Much of the misplaced fear and distrust surrounding AI adoption traces back to a single omission in how people are often introduced to its use. Businesses and the media have fixated on the intelligence aspect while often ignoring the behavioral framework required to make it work in the real world.

The early representation of Generative AI suggested it was a shortcut that required very little effort. If users were told upfront about the level of detail, context-setting, and iterative refinement required to get a usable result, the hype might have been quieter (look how long Anthropic was off the radar of the general public), but the real work with these powerful tools might have started sooner for the average person and business ( AI Adoption Puzzle: Why Usage Is Up But Impact Is Not, BCG, 2025)

Subscribe now

We are essentially trading traditional coding hours for what some call vibe coding: throwing natural language at a problem and hoping the model catches the intent. Vibe coding is a legitimate way to prototype, but it becomes technical debt if you do not eventually solidify the logic. Replacing a clean specification with an open-ended series of guesses is how projects lose their shape before they find their footing.

The most effective approach is not simply plugging a model into an existing process because it looks like it might help. Genuine acceleration comes from a willingness to rethink how things get done, then determining how AI can facilitate those better ways. It is the difference between automating a flawed process and designing a new one.

The success stories often come from teams who looked at a failed output and wondered what specific lever they forgot to pull. They treat the model as a mirror. If the output is off-base, it usually means the instructions provided were incomplete or lacked the necessary constraints. It is an objective way to see where our own requirements are fuzzy.

This is particularly evident in workflow automation. Earlier automation projects often failed because they only mapped the mechanics. We drew boxes and arrows to show what happened next, but we ignored the intent.

AI-driven automation is succeeding where those attempts fell short because the machine requires the reasoning, not just the step. To make an agent navigate a workflow, you have to document why each step exists. This forces organizations to complete their process definitions rather than paper over the gaps. If you cannot explain the logic behind a decision point, the machine cannot execute it. This forced clarity is the real process improvement.

Share

The Double Standard

There is a noticeable double standard in the modern workplace. When an LLM returns a hallucinated mess or fails a logic branch, we iterate. We refine the prompt. We provide more context. We give the machine a level of professional grace and patience that we rarely extend to our human peers.

Think about what that looks like in practice. A new team member submits work that misses the mark, and the first instinct is to question their judgment or capability. The same output from a model and the instinct is to wonder what context was missing from the prompt. One is treated as a character flaw; the other as a specification problem. They are often the same problem.

If organizations applied that same diagnostic instinct to people, treating an incomplete first draft as a gap in the brief rather than a gap in the person, productivity would likely increase. Instead, we frequently demand accuracy on the first pass from humans while subsidizing the machine’s learning curve with endless retry clicks. ( The Human Side of AI Adoption: Lessons From the Field, MIT Sloan Management Review, 2025)

The Same Loop Applies to Both

Closing that gap is not primarily a technology problem. It is a management problem, and the same loop applies whether you are working with a model or a person.

Start by acknowledging that a wrong answer is often a sign of a logic path being tested; it is data, not a failure. Reward the attempt at solving the problem; in early iterations, the goal is narrowing the scope, not delivering the final answer. And when the output is off-base, assume the cause is a lack of clear boundaries before assuming incompetence. These are not novel management principles. They are just easier to see when the thing being managed cannot take it personally.

The teams getting real value out of these tools are not looking for a magic button. They treat the AI as a diagnostic tool for their own process gaps. They do not just want the answer; they want to see where the system broke so they can fix the underlying logic.

Subscribe now

The One Attribute That Survives

This brings us to the attribute that determines whether a tool gets abandoned or mastered.

Curiosity is the only attribute and attitude that survives the hype cycle.

Expectations without curiosity lead directly to disappointment. If you aren’t wondering why the model failed, you will just conclude the tool is broken and move on. In a technical context, curiosity is the bridge between a strategy and a result. It leads to both the perseverance and the openness to changing the way we think about how things get done. It forces us to reprioritize the work based on what the machine reveals about our own internal logic.

Proficiency in this landscape is not about mastering a specific toolset, because those change every few weeks. It is about an underlying hunger to understand the mechanics of the work. If you have that curiosity, you will find the ROI because you will keep digging until the logic is sound.

Until next time…

© Scott S. Nelson

Leave a comment

There’s this great weekly online morning meetup I join when I can called “The Secrets of Product Management”, led by Nils Davis. Recently the topic of Spec Driven Development came up.

Full disclosure: I didn’t take notes in the meeting and there were a lot of concepts and thoughts shared verbally and in the chat. Some of what I recall may be off, and I hope that if anyone present reads this and has a better recall that they share their thoughts in the comments.

Share

Some thought it was about a Product Manager gathering all of the specifications for a product in advance, and that it led back to waterfall style processes.

Some thought it was building a Proof-of-Concept to serve as the specification.

By the end of the discussion, the one thing everyone (mostly) agreed on is that it works much better when done iteratively, and includes direct references to standards.

As an architect who still codes, my understanding of SDD is that it is about the spec files that are carefully crafted to direct generative AI in how to write the code. It is a way to get better code from the AI that will require less refactoring after the first results.

The different perspectives made me think it was worth doing a little research and summing it up for my reader here. I admit to mentally vacuuming up a lot of content about AI in order to feed my own synthesis on its use, and the key thing that I saw differently was the ownership of the specification used for SDD.

Subscribe now

When I presented the question “Who owns the spec in spec driven development?” to an AI, it responded with “…humans own the spec…”, which points out a whole new perspective.

So, that’s what drove me to dig in a little bit to improve my own understanding and share the results.

A Quick History Lesson

Like most things in IT, the earliest signals appear long before what we later label as “modern” computing (a term that conveniently tends to align with when each of us personally got excited about technology). As far back as 1987, Managing the Development of Large Software Systems: Concepts and Techniques outlined ideas that closely resemble what many now think of as Specification-Driven Development (SDD). Interestingly, its diagrams reflect structures similar to waterfall methodologies (an ironic reminder that many “new” ideas are refinements of older patterns rather than entirely novel inventions).

These concepts did not evolve in isolation. Over the following decades, they were reinforced by related disciplines such as formal methods and API design principles like Design by Contract* (which emphasized precision, verifiability, and clearly defined interfaces). Later, approaches like Behavior-Driven Development (BDD) carried some of this thinking forward, framing specifications as shared artifacts between humans and systems (but still largely as guidance rather than execution).

What has changed more recently is the role of AI in making specifications actionable. Around 2025, tools began to emerge that transformed specs from passive documentation into active drivers of implementation. Projects like AWS Kiro and GitHub’s spec-kit marked a shift. Specifications became executable guides for coding agents (not just references for developers). In this sense, “modern” has continued to compress (moving from spanning decades to evolving almost in real time), as specs shift from descriptive artifacts to operational components of the development process.

Share

Opinions Still Differ

I don’t think my input in the recent conversation changed anyone’s mind about how they define SDD. And people will definitely have strong opinions on the value of SDD.

In a recent post, Allen Holub said:

“People talk about spec-driven design, but the best spec you can have is a test—a test you write before you write the code. You don’t write a test to see if the code adheres to a spec. The test IS the spec. Don’t write specs. Write tests.”

I agree with TDD proponents, because it is part of a Continuous Testing cycle, a process that was just starting to really catch on before GenAI went GA, and is even more important since. That said, tests are part of the spec, they are just managed a little differently because the developer doesn’t happen to be human. That’s the whole point of SDD. It is how developers work with agents through clear communication. Because, let’s face it, the Agile approaches of sitting with a user won’t work with AI until after the code has been written, and pair-programming with an AI was only modern for a moment.

Helpful Tools to Try

Tools make this less painful than it sounds.

GitHub’s spec-kit is a good entry point. It gives you Markdown templates for a “constitution” file with principles, then spec.md, plan.md, tasks.md. You slash-command it in your IDE, and AI fills in the gaps. They put it well: “The specification captures intent clearly, the plan translates it into technical decisions.” ([GitHub Blog], Spec-driven development with AI: Get started with a new open-source toolkit) Amazon’s Kiro does staged workflows, Tessl flips code to byproduct. Red Hat talks up “lessons learned” files to feed back into future specs, cutting errors over time ([Red Hat Developers], How spec-driven development improves AI coding quality).

Subscribe now

Wrapping Up

All in all, my sense is to treat specs like your IaC or database schemas. Human owned from the start, iterated carefully, governed with some structure. Reference standards to ground it. Try it small, on a utility script maybe, and see how it holds up in real work.

If it fits your flow, it can add real velocity with AI. If not, no big loss; plenty of paths forward.

*Side note: Yes, I usually have these inline in parenthesis (a habit my AI editors hate), but this one seemed too long for that, so… I did some research with Gemini where it insisted on a correlation between design by contract and spec driven development, which at first I took to mean it prefers its training data rather than current information, so I switched to my usually research LLM wrapper, Perplexity. After some hind-brain thinking, it occurred to me that Gemini may have semantically equated specification with contract, which is another quirk of AI: it is so darn literal!

©2026 Scott S. Nelson

Leave a comment

For my regular reader, you know I can’t resist a pun, and the initial research note for this post was “Timely topic title: Take the Tax out of Taxonomy”. You also know I digress, so I thought I would get it out of the way at the start. Done. Moving on to the next level…

You are paying a massive hallucination tax. You bought a premium AI subscription or deployed a desktop agent. You pointed it at a project directory full of deprecated drafts, unstructured notes, “versioned” files, maybe even some sample code. Now the AI is confidently generating output based on requirements from three years ago, and maybe Wednesday’s lunch order.

Subscribe now

The AI users assume a better foundation model or highly complex prompt engineering will fix output inaccuracy. They will not. According to the research paper A Comprehensive Taxonomy of Hallucinations in Large Language Models, hallucinations are not merely a bug, but a theoretically inevitable feature of computable LLMs, irrespective of architecture or training.

You cannot patch out hallucinations with a clever system prompt. You have to restrict their oxygen.

Generative AI operates entirely on the context it is fed. When you open a workspace (or upload a zip file, or point it at SharePoint), the model uses the folder structure to understand relationships. It assumes every file in the provided directory is equally valid, current, and relevant.

To get faster, accurate output, you must adopt a standardized, hierarchical folder taxonomy. This is not a housekeeping chore. It is a strict data contract for your AI. The academic consensus supports this structural approach. As outlined in A Systematic Framework for Enterprise Knowledge Retrieval, transforming a static blob of data into a navigable, context-rich knowledge architecture significantly improves model accuracy and reduces retrieval latency.

Share

The Prep Station Metaphor

Think of an LLM as a highly skilled line cook with zero short-term memory. If you ask the cook to make an omelet, but point them to a kitchen counter where the fresh eggs are mixed in a pile with old receipts, bleach, and rotten produce, the resulting meal will be toxic.

You have to prep the station before you ask for the work.

This requires changing how you manage your local environments. You must segment your files and organize your folders explicitly by client, project, or (and sometimes “and”) specific activity. When the AI opens that specific folder, the taxonomy forces it to focus strictly on the given task.

The Micro-Macro Data Contagion

Local file structures often mirror enterprise data architecture. If your team’s shared drive is a chaotic dumping ground of nested, unnamed folders, your enterprise data lake is likely more of a content swamp.

Organizations often fund massive, top-down enterprise data transformation projects. They deploy tools to wrangle petabytes of unstructured data. Consultants are brought in to describe how it should be done, walk you through clean up, and leave you with a perfectly indexed wiki on how to maintain it.

The reason other organizations don’t do this kind of clean up? Aside from the few that don’t need it, the rest have someone recruited from an organization that did need it, then did it…then did it again a few years later. At least some had the excuse of acquisitions as a cause. The rest just forgot to make being organized part of the organization’s culture.

Subscribe now

The report What Is Data Taxonomy? Use Cases & Best Practices points out that taxonomy programs do not fail because the classification structure was wrong. They fail because nobody owned it after launch, or the controlled vocabulary was written for data engineers rather than the business users who needed to adopt it. A taxonomy that nobody actively owns becomes outdated within twelve months.

If you build a pristine enterprise knowledge graph but your teams still save raw client notes to a local desktop folder named “Misc”, your clean data architecture will erode. Bad habits always defeat good infrastructure.

Start locally. Expand globally. Treat your team’s shared folder as the training ground for enterprise AI.

Here is the implementation baseline for engineering a reliable folder taxonomy.

1. Force Local Discipline: The guide Document Taxonomy Simplified notes a critical reality: AI can read full text, but without consistent indexing and classification, it has a harder time understanding which documents are current or relevant for a specific question. Humans must define the taxonomy. Organizations that rely solely on AI risk amplifying bad data.

2. Build a Strict Domain Hierarchy: Segment folders strictly by project and lifecycle status. Your AI should never have read access to a “Drafts” directory when you are asking it to write production documentation.

3. Establish the Data Contract: Metadata like document type, owner, client, date, and status tells AI not just what a document says, but how it should be used. This context improves AI ranking and reduces irrelevant hits that happen to share keywords.

4. Separate Human and AI-Native Formats: Segment your directories into files meant for humans and files meant for the AI. Lean towards using markdown files, text files, and CSV files for AI consumption. Keep heavy, formatting-rich files in a separate reference folder that the AI does not scan unless explicitly commanded.

5. Isolate Contextual Boundaries: Open-ended prompts can generate answers that blend multiple disciplines or outdated content. When your library is indexed by department, project, and lifecycle stage, AI can narrow its focus and answer questions within the right slice of your information.

You’ll note that there is a lack of solid reference examples of good taxonomies. This is, again, related to data cleanliness being driven by culture. The same taxonomy may or may not work for another organization. But a solid taxonomy based on how the organization thinks and processes can easily be maintained through training, communication, and the occasional reminder (preferably automated).

Share

The ROI of a strict folder taxonomy is immediate. Output precision goes up. Token waste goes down.

If your AI is only as reliable as the context it receives, your unstructured file storage is an active threat to your workflow. Build the hierarchy locally. Clean up the directories.

Leave a comment

©2026 Scott S. Nelson

Remember the movie Gremlins? A friendly creature that if you fed it after midnight, all heck would break loose and humans would be left to clean up the mess in the morning.

Despite setting all of the basic configurations to the contrary, Windows 11 occasionally reboots after an update on its own. It restores all of the running windows afterward in an attempt to hide this from users. Sometimes it fails on the restart, and I have to dig my laptop out from under the monitor stand where it is connected to everything via a USB dock. Other times it succeeds and most users wouldn’t notice. I, on the other hand, have a massive amount of data and virtual machines collected over 34 years of Windows versions (and another 6 of DOS and whatever the original Mac OS was), all of which reside on encrypted external drives, and there is only one reason they will all have been disconnected: a stealthy reboot.

Share

Step 1: Spot the Sneaky Restart in Event Logs (Skip the Freezing GUI)

Event Viewer sounds official, but it freezes like molasses when filtering big logs. Don’t bother with it—use PowerShell instead (search “PowerShell” → Run as Admin).

First command to check for crashes/power losses:

PowerShell

Get-WinEvent -FilterHashtable @{LogName=’System’; Id=41,6008} -MaxEvents 20 | Sort-Object TimeCreated -Descending | Select-Object -First 1 TimeCreated, Id, LevelDisplayName, Message | Format-List

• What it does: Looks for “dirty” shutdowns (ID 41 or 6008). No results? Good news—your restart was “clean” (planned). (Mine showed nothing, so not a crash.)

Subscribe now

Step 2: Dig for the Real Story with Broader Boot/Shut Logs

No crash? Check all reboot clues:

PowerShell

Get-WinEvent -FilterHashtable @{LogName=’System’; ID=12,13,1074,6005,6006,6009} -MaxEvents 10 | Sort-Object TimeCreated -Descending | Format-Table TimeCreated, Id, LevelDisplayName, ProviderName, Message -Wrap

• Key IDs: 1074 reveals who restarted (e.g., “TrustedInstaller.exe… Operating System: Upgrade (Planned)”).

• The Culprit: TrustedInstaller (Windows Update’s installer) often forces an “upgrade” reboot. (Mine hit at 10:51 PM—classic move.)

Bonus quick boot time check:

PowerShell

Get-CimInstance -ClassName Win32_OperatingSystem | Select-Object LastBootUpTime

Share

Step 3: Lock Down Future Restarts (Microsoft’s Hidden Maze)

Your BSOD “no auto-restart” setting? Useless here. Windows ignores it for updates. Time for the real fixes.

Quick Registry Shield (Copy-Paste PowerShell as Admin)

PowerShell

New-Item -Path “HKLM:\SOFTWARE\Policies\Microsoft\Windows\WindowsUpdate\AU” -Force
Set-ItemProperty -Path “HKLM:\SOFTWARE\Policies\Microsoft\Windows\WindowsUpdate\AU” -Name “NoAutoRebootWithLoggedOnUsers” -Value 1
Set-ItemProperty -Path “HKLM:\SOFTWARE\Policies\Microsoft\Windows\WindowsUpdate\AU” -Name “AlwaysAutoRebootAtScheduledTime” -Value 0
gpupdate /force

• Why it works: Tells Windows “Don’t reboot if I’m logged in.” Verify with Get-ItemProperty "HKLM:\SOFTWARE\Policies\Microsoft\Windows\WindowsUpdate\AU".

Pro GUI Version (gpedit.msc)

• Run gpedit.msc → Computer Config → Admin Templates → Windows Components → Windows Update → Manage end user experience.

• Enable Configure Automatic Updates → “2 – Notify download/install”. (This stops the OS from making decisions for you.)

• Enable Turn off auto-restart during active hours → Wide hours (e.g., 6 AM–11 PM).

• gpupdate /force.

Nuclear Option: Tame Update Orchestrator

• services.msc → Update Orchestrator Service → Startup: Disabled → Stop.

• No sneaky scheduling anymore (updates still work manually).

Subscribe now

Why Microsoft Makes This So Hard (Minor Rant)

I get why the default configuration reboots on its own. Most users will never bother to check updates on their own, and if they are installed automatically but require a reboot, the message will be ignored by the general populace until the next upgrade. Anytime a non-technical friend or family member brings me a device that is giving them problems, finishing an update fixes it 92.8% of the time. But…

There’s a bunch of us nerds who came up from our basements (or down from the attic, in my case) and still want fine control over our own equipment without having to spend 4 hours on Google (or Gemini, or Claude, or…you get the idea). Or blowing up a movie theatre.

Leave a comment

©2026 Scott S. Nelson

TL;DR: Today’s “No AI” corporate policy is just the 1990s “No JavaScript” policy with a higher compute bill. Paradigms shift whether their security team likes it or not.

The Open Big Gulp® in the Server Room

When JavaScript arrived, they treated it like a security breach waiting to happen. IT managers saw it as a “toy” that let hackers steal cookies (which it did, because the sandbox was more of a suggestion back then). Netscape 2.0 released it as a way to make the web interactive, but the initial response was a panicked reach for the “Off” switch. You can read the architectural post-mortem of early JS to see how the fear of unvetted client-side execution nearly throttled the modern web.

They blocked it. They turned it off in browsers. They built a static, safe, and entirely useless wall around their organizations. It was a classic case of prioritizing perimeter defense over functional utility. (The server room was dry, but the business was dehydrating.) Static HTML was “secure” the same way a bricked laptop is secure: it doesn’t do anything, so it can’t break.

Then the ROI of interactive applications made the “Block All” strategy look like a company-killing move. They didn’t win by keeping it out: they won by learning how to wrap it in architectural guardrails. AI is currently sitting in that same unvetted hallway. It is a messy, sugary disaster waiting to spill into your production environment, yet every developer is already sneaking it through the door because they can’t work without the “caffeine” it provides. If you don’t provide a cup with a lid, you’re just waiting for the spill to happen.

Share

New Age Resolutions

Being an architect who can navigate a paradigm shift isn’t a “gift.” It is a byproduct of breaking systems until you understand the failure points. Mastery requires practice and pushing boundaries: the literal grit of failing until you find the baseline. You don’t wake up knowing how to manage a LLM anymore than you wake up knowing how to optimize a SQL query.

Most people waiting for an official “AI Policy” are actually just waiting to be told what to do. (These same people will wait for someone to tell them how to do it, which won’t bode well for their stock values). They treat adoption like a “gift” that will eventually be delivered by a consultant. In reality, adoption is like a New Year’s resolution. Recent data shows that upwards of 90% of resolutions are abandoned before they even see the heat of summer.

Most teams fail because they run straight for the goal: the “total automation” fantasy: and get tired before they are halfway there. They treat AI like a sprint instead of a baseline change in how we think about compute. The survivors set an intention to steadily improve rather than sprinting toward a finish line that keeps moving. True expertise is earned through bottom-up experimentation. If you aren’t currently failing at prompting or orchestrating small agents on your local machine, you aren’t “playing it safe.” You are becoming obsolete.

Building the Guardrails

To get in front of the shift without getting cut by the bleeding edge, treat AI like any other external dependency. You wouldn’t pull a random library from a public repo and drop it into your core banking app without an audit, so stop doing it with LLMs. Use these procedures to keep the Big Gulp® off the motherboard:

1. Isolation: Run your experiments in a sandbox that doesn’t touch production data. If you can’t self-host a model, use an enterprise instance with zero-retention policies. (Putting proprietary code into a public chat interface is the 2024 version of leaving your server room door propped open with a brick).

2. Validation: Never trust the output of a paradigm-shifting tool without a human peer review. If an AI writes a script, you own the script. (Admitting “the AI made a mistake” is a newbie gotcha that won’t save your job). Check the OWASP Top 10 for LLM Applications for a checklist on how to actually vet these outputs before they hit a compiler.

3. Governance: Define the data boundaries now. If you don’t define an architecture, you have still defined one: it’s just likely a bad one. Create a standard for which models are allowed and what constitutes “sensitive data.” (If it is not written down, the guardrail does not exist).

Subscribe now

The Ouroboros of History

Ignoring a shift because it is “too risky” creates a massive amount of organizational friction. While they are debating the ethics of LLMs in a three-hour meeting, their competitors are figuring out the ROI of augmentation. This cycle mirrors how JavaScript transformed the web from a collection of digital brochures into a dynamic business engine. Just as users stopped tolerating websites that required a full page refresh for every interaction, they will soon stop tolerating businesses that operate at pre-AI speeds.

Signs of ubiquitousness are everywhere, as AI moves from a technical curiosity to a fundamental mechanic of communication and analysis across all departments. Marketing teams are using AI to synthesize sentiment from thousands of customer transcripts in seconds (parallel to how JS enabled real-time behavioral tracking and heatmaps). Legal teams are using it to flag anomalies in five-hundred-page vendor contracts (mirroring how JS moved complex form validation and logic from the server to the client for immediate feedback). According to , AI could contribute up to $15.7 trillion to the global economy by 2030, driven largely by these operational enhancements. Even executive assistants are using it to ghostwrite scheduling threads and summarize meeting minutes: a leap in administrative efficiency comparable to the arrival of real-time collaborative web suites. In these contexts, the “hallucination” risk is just the new version of a typo. If you don’t have a human in the loop to verify that the summary actually reflects the meeting, that’s not a tool failure: it’s a process failure.

The first clear ROI has been in software development, with early adopters way ahead of those that blocked the .ai top level domains back in 2023. Research from GitHub’s productivity studies shows that augmented developers complete tasks 55% faster. This is the modern version of the “Full Stack” revolution brought on by Node.js: it changes the fundamental economics of what a single contributor can produce. If your organization is the only one in the sector still digging with hand shovels while the competition has backhoes, you aren’t “preserving the craft.” You’re just going out of business.

Architectural principles shouldn’t be a “No” department. They should be the manual that explains how to use a high-performance engine without blowing the gaskets. In 1995, we didn’t need “No JavaScript” policies; we needed better sandboxes. Today, we don’t need “No AI” policies; we need a documented path for bottom-up augmentation.

Closing Tip: Start a local wiki for AI gotchas encountered by your team. Record every failed prompt, every hallucinated library, and every successful integration. If it is not written down, it does not exist.

Leave a comment

Subscribe now

Share

It depends.

Yeah, I hate that answer, too, but it’s because we all prefer simple answers to the real ones. We also want to believe in overnight success, one size fits all, and that the plug and play option is all we need. And don’t get me started on long term weather predictions!

But it helps to know what it depends on, which is, in this case, where you are in your AI journey. The same approach really applies to any learning journey where there are multiple aspects, so we’re going to start with looking at it simply from the perspective of learning.

Why and How and When to Start Deep

If this is your first foray into a new realm of knowledge, start by going deep on one aspect.

Pick the area that you are most interested in. Intrinsic motivation is a better driver for learning than any reason that includes “have to”. Once you have picked that topic, dig in and follow your curiosity until you feel you can converse freely on the topic. This is how you build a baseline mental model.

Going deep in one specific corner makes other adjacent areas easier to absorb later because you actually have a frame of reference to hang new information on. When you encounter a new tool, you filter it through existing mental models to facilitate integration of new knowledge. This cognitive filtering means you aren’t starting from scratch every time a model updates. You are simply updating a specific branch of an existing tree. (See The Memory Paradox: Why Our Brains Need Knowledge in an Age of AI)

Subscribe now

The Pivot to Breadth: Mapping the Landscape

Once that baseline mental model exists, going broad is more valuable.

This first accumulation of breadth is to understand what’s possible, or available. You aren’t trying to master everything. You’re mapping the space so you know where the boundaries are. This aligns with the “T-shaped professional” model, defined by having deep expertise in a specific area while also possessing broad knowledge across various disciplines. This structure ensures you have enough technical depth to contribute high-value work immediately. It also gives you enough breadth to collaborate with experts in adjacent fields without needing a translator.

Going broad makes it easier to know exactly when and where to go deep next. Knowing what exists and what is possible makes it easier to say “I have an idea of how that can be done” with conviction.

The Trap of Constant Depth

The problem with going deep on one thing at a time, after the initial deep dive, is that when you need the knowledge or skill in a practical situation, there may be something adjacent that will make it easier or is better suited.

If you’re buried in a single silo, you won’t see it. This is why pure specialists struggle when their niche technology shifts. Markets move faster than individual mastery, which is why modern engineering organizations must embed specialists into existing teams. Breadth prevents you from becoming a legacy asset the moment your specific tech is disrupted. It provides the foundation for transferring implicit knowledge, which is the exact kind of knowledge needed to generate creative ways of tackling business problems. Innovation happens at the intersection of two unrelated fields.

Share

Managing the Hierarchy of Ideas

To move between breadth and depth effectively, you have to understand how to categorize information. A practical framework to understand how to conceptualize those categories in a given realm of knowledge is The Hierarchy of Ideas. This concept allows you to mentally zoom in and out of a topic. It ensures you are always operating at the exact level of detail the current problem requires.

Think of a hierarchy using transportation as the frame. At the top, you have the abstract concept of “Transportation”, which includes planes, boats, trains, cars, skateboards, and ox carts. Moving down a level, you find “wheeled vehicles”, which is still broad enough to encompass trains and scooters. Further down, “Cars” will include internal combustion, electric, and peddle powered. As a mechanic, you will be more interested in learning the distinctions between a Ferrari and Hyundai, or between the Sonata and Kona. The higher you go, the more general and broad the idea becomes. The lower you go, the more specific and detailed it gets.

Navigating this hierarchy is done through “chunking.” When you chunk up, you move from the specific Tesla to the broader category of “Transportation” to understand the big picture. When you chunk down, you move from the general concept of “Cars” into the specific components like the “Battery Management System” to find depth. You can also chunk laterally, moving from “Cars” over to “Trains.” This allows you to find alternative solutions that exist at the exact same level of utility.

Subscribe now

The AI Sandbox: Applying Levels and Chunks

AI is like transportation in the way that zoology is like geology. They aren’t one giant subject. There’s a hierarchy of distinct concepts, applications, audiences, and values that you have to navigate intentionally.

Start by chunking down into a specific primary aspect. Dive into development. If you choose Software Development, don’t just use a generic chatbot. Master how tools integrate directly into the developer’s workflow. Modern development is shifting toward a model where the AI handles low-level syntax. The human engineers and architects manages high-level logic and security.

If you choose Marketing, dive into tools capable of predicting future trends. These platforms move you from general demographic targeting to individual-level behavioral forecasting in real-time. This creates your first deep anchor.

Once you feel steady, chunk up. Skim through news and articles about the overall space so you get a sense of the capabilities. Map the broader landscape—from vector databases to multimodal generation. Staying informed at this high level prevents you from getting blindsided by architectural shifts.

As you build that breadth, you chunk laterally. Then, when something comes up that would benefit greatly from an aspect other than your first specialty, you will recognize that your current focus isn’t the right one in that context. If you are deep in software development but hit a bottleneck in data quality, your broad map will point laterally toward data architecture. You will have a good idea what aspect is better suited. Then you can partner with someone that has that expertise, or learn it deeply yourself, or both.

Effective mastery requires building a foundation deep enough to create your mental anchor, while maintaining a wide enough perimeter to spot the right tools for the job. You cannot specialize into obsolescence and expect to stay relevant in a field that moves this fast. Whether you are ready for your first technical deep dive or you are currently gathering seeds for future growth, the only wrong move is standing still. Pick a starting point and get to work.

For your convenience (plus, I hate to throw away interesting artifacts that AI outputs when researching my articles), below are some areas (i.e., non-exhaustive) to consider when conceptualizing the Hierarchy of Ideas around AI.

AI Deep Dive Reference Table

I don’t find images of tables helpful so the heading above is linked to the table on my website.

Leave a comment

©2026 Scott S. Nelson

Originally published at https://theitsolutionist.com/2026/03/20/learning-ai-and-going-broad-or-deep-first/

TL;DR: The use of AI in development has shifted from a coding assistant to a team of agents doing the heavy lifting. This requires developers to skill up in management and forces a fundamental shift in how software roles collaborate.

The Managerial Migration

For those not watching closely (which is most people, and perhaps only a few reading this), the world of software delivery is in the midst of a tectonic shift. The use of AI is evolving from a simple coding assistant to a team of agents, or experts, performing the bulk of the work. This moves the developer role from a traditional software engineer to an agent manager.

The change in role definition and skills is another aspect of the paradigm shift the Age of AI is bringing. If this sounds new to you, it is because you may have missed the transition that the proliferation of the “WWW” subdomain brought to IT in the 90s. We are all going to come out better, but it is going to be a long haul as we re-learn lessons from last time and write new ones for this era.

Subscribe now

The Expertise Gap in Management

One common misconception is that this automation means developers can be replaced by Product, Project, or Program managers. This is mostly the “Wall Street Rumor Mill,” which is only just now being revised from “replacing people with AI” to “shifting investment from employees to AI vendors”. At least that is more honest.

The “average” manager often lacks the technical depth to write a precise specification or review the complex output of an agentic workflow. Managing a digital workforce requires the same technical understanding and focus as writing source code. If you cannot perform a rigorous technical review of what AI agents produce, you should not put it into production (unless you suffer from a terminal case of technical hubris).

The Developer Drift

While many managers lack the depth to take over, developers are not guaranteed success in this new role without learning to view technical problems from different angles. Many developers tend to drift from the business context without a reminder, which is why lifecycle ceremonies exist to gather feedback from users and owners. For some, this is a forest-versus-trees effect, while for others, it is the temptation of a “cool” approach over a practical solution.

The speed of AI can take a minor gap in understanding and expand it into a costly chasm. When an agent can produce a week’s worth of logic in seconds, the cost of moving in the wrong direction scales exponentially. The team must find a way to collaborate where agents are a factor beyond just a tool choice.

Share

Grit over Gift

This proficiency is not a magic gift: it is a byproduct of learning, practice, and pushing boundaries. There is a persistent myth that “prompt engineering” is an inherent talent or a shortcut for the lazy. It is actually the opposite. Real proficiency comes from hundreds of hours spent in invisible iteration. You have to break the agents to understand how to fix the workflow. These skills are then applied to context engineering, where the developer becomes the manager and the back-and-forth transitions to a human-in-the-loop system.

Deep experience can sometimes trigger intellectual rigor mortis, where you stop looking for a better way because you already know the “right” way. To succeed now, you need the grit to unlearn habits that are no longer efficient. High ROI in the age of AI belongs to the person who pushes boundaries through practice, not the one waiting for the “perfect” model to arrive.

Subscribe now

The Practical Pivot

As we navigate this “.ai moment,” leadership, managers, and developers need a new way to interact. It is no longer about a ticket hand-off: it is about real-time orchestration.

• Developers: Start treating your AI tools as interns, not calculators. An intern needs guidance, a clear spec, and a rigorous peer review. If they produce garbage, it is a reflection of your management. Mentor your agents by providing better context and documentation.

• Managers: Help leadership understand that the “silver bullet” still requires expert aim. AI is a force multiplier, but it requires a human who knows where to point the barrel. Use these tools to bridge the communication gap, not to eliminate the experts.

• Everyone: Support each other in cross-training. Incorporate big-picture product thinking with low-level solutioning. Document the new workflows immediately, as your team now includes transient sessions that lack long-term memory.

Incorporating this new layer requires new connections, shifts in responsibility, and overlaps that act as double-checks from different perspectives rather than simple redundancies.

Leave a comment

TL;DR: Experience is a liability when it kills curiosity. AI proficiency is a byproduct of hundreds of hours spent in invisible iteration.

The Expertise Trap

Deep experience often triggers intellectual rigor mortis. You have seen the “right” way to do things for a decade, so you stop looking for the better way. A beginner mindset is not about being a blank slate (which is just another word for useless). It requires enough of a foundation to know when you are headed in the right direction, or perhaps a parallel path with new perspectives, before getting back on track.

If you have twenty years of experience but no curiosity, you are just a legacy system waiting for a decommission date. The ROI on a “beginner” attitude is higher because it allows for rapid pivoting. You need the basics to provide a compass (to avoid spending days debugging a syntax error), but you need the mindset to explore the paths that lead to breakthroughs.

The AI Sweat Equity

There is a myth that prompting is a low-skill activity. It is not. Most people really good at prompting have iterated and learned. The developers currently running multiple agents and building software 4 to 10 times faster than they did last year have been in months of practice to get there.

This is iteration 0 work. It is messy and mostly undocumented (because the tech moves faster than the README files). What makes it daunting for those first starting is that the people who are now good at it did it without formal training. There is a tendency to forget how much effort went into the initial struggle.

Share

Building the Mental Infrastructure

Learning new technical skills requires toggling between different cognitive states. Barbara Oakley (author of Learn Like a Pro: Science-Based Tools to Become Better at Anything and Learning How to Learn, among other great books) describes this as the tension between focused and diffuse modes. Focused mode is for the granular syntax: the structure of a prompt or a script. Diffuse mode is where the beginner’s mindset lives. It is the relaxed, curious state that allows your brain to make the non-linear connections required to solve a problem that does not have a documentation entry yet.

She emphasizes chunking: breaking complex concepts into small, functional units until they become second nature. This prevents cognitive overload when the system throws an error you have never seen before. Curiosity is a tool that keeps you in the diffuse mode long enough to see the “big picture” before diving back into the details. I took her class on Coursera at the start of my AI journey, and I recommend everyone do the same, even if your interests are in other areas. It applies to learning anything, and you will thank yourself for doing so.

They say “oh, it’s easy, you just do this”, which looks like magic to the beginner. It is not entirely different from visiting a new area and asking a local for directions. Every time they start with “Oh, that’s easy”, there is a good chance you are going to get lost following their directions.

Locals navigate by landmarks that either do not stand out to an outsider or have disappeared from all but the local’s memories. They tell you to turn where the oak tree used to be or past the shop that changed names five years ago. They have internalized the route so deeply they forget the friction of finding it the first time.

And sometimes, that makes the trip more fun. Stop looking for the “perfect” prompt or the “right” workflow. Spend more time being “lost” in the tool. The goal is not to avoid the detour: the goal is to have a strong enough foundation to know how to get back to the main road once the detour stops being productive.

Subscribe now

A Simple Roadmap

If you haven’t begun your journey with Generative AI, or feel a bit lost, here’s a simple roadmap to help you along:

1. Pick one model and stay there: Stop comparing benchmarks and just use one tool (Claude, GPT, or an LLM via API) for a week straight to understand its specific “personality.”

2. Iterate on a single prompt 50 times: Don’t just accept the first output. Change one variable at a time until you understand exactly what triggers a hallucination vs. a logic block.

3. Read the system prompt documentation: Most users treat AI like a search engine. Read the actual technical guides on “system roles” and “temperature” to understand the controls.

4. Practice manual orchestration: Before you try to automate a multi-agent system, act as the agent yourself. Copy the output of one model into another and manually fix the “gotchas” in between.

5. Fail on purpose: Try to make the model break. If you don’t know the edges of the tool, you won’t know when you are standing on a cliff.

© Scott S. Nelson

Leave a comment

As a writer who has (rarely) been paid for the craft, I find attribution to be not only important, but a moral obligation. Especially when the quote resurfaces in my thinking time and again.

I first read “There are three kinds of lies: lies, damned lies, and statistics” in a Robert Heinlein novel, where he (through his character) attributed it to Mark Twain. Samuel Clemens actually attributed this truism to someone else, though that providence has not been agreed upon by scholars (per Wikipedia), so I will leave this instance of noblesse oblige as having made my best effort. But I digress (a specialty of mine).

Share

Despite my failed attempt at correct attribution, it’s still true. And it is being proven once again, in the various reports and claims around AI. Allow me to categorize them as I see them today:

1. Lies

…are that AI will be able to do X by N. No one really knows how long. Extrapolating future data based on similar but different prior data has an unknown margin of error. This is why every honest company states at the beginning of their demo or presentation or pitch that you should not base purchase decisions on functionality not currently released (and I would add not proven for your specific use case).

2. Damned Lies

…are about what people are accomplishing now. The app vibe coded in a weekend is either going to be feature frozen in the near future, suffer a major failure through bad coding or bad actors, or (most rare) be the result of extensive planning prior to the weekend of wonder.

“The ‘vibe-coded’ apps that fail are not failures of code, but failures of craft.”

3. Statistics

…are the worst of the three, because anyone with half a brain and less scruples can get the same numbers to say entirely different things, like:

• MIT report: 95% of generative AI pilots at companies are failing.

• 56% get nothing from adoption.

• Two-thirds of nonmanagement staffers said they saved less than two hours a week or no time at all with AI, while over 40% of executives reported saving more than eight hours weekly.

• Enterprise AI adoption grew with worker access rising 50% in 2025.

• 60% of executives report AI boosts ROI/efficiency.

• Corporations plan to double AI spending.

Subscribe now

Lies Aside

What we call AI (usually) isn’t actually AI. What everyone is calling AI is still the biggest paradigm shift in Information Technology since computers shrunk from needing a room to sitting on a desk (and, not long after, a lap).

Impacts to humanity are a physics phenomenon in that every positive improvement has an equal and opposite potential that will eventually be realized. Whenever the many benefit, a very few benefit a lot more. Some deserve those benefits, and some don’t.

The Mirror and the Machine

We find ourselves at a peculiar crossroads where the technology is accelerating faster than our ability to tell the truth about it. We are told AI is a magic wand, a job-thief, or a savior, but the “Lies, Damned Lies, and Statistics” reveal a simpler reality:

“AI is not a destination; it is a mirror.”

The statistics that show executives “saving eight hours a week” while workers save none aren’t an indictment of the technology—they are an indictment of how we value human time. When we strip away the marketing gloss and the manipulated ROI reports, we are left with the same struggle that has defined every Information Technology shift since the first mainframe: the tension between efficiency and agency.

Share

The Paradox of Progress

If my “physics phenomenon” theory holds true, the equal and opposite reaction to the AI boom will be a renewed, premium demand for the irreplaceably human.

“AI makes it remarkably easy to be mediocre at scale.”

The paradigm shift isn’t just about computers moving from desks to our pockets, or from pockets to our cognitive workflows. It can generate the lies, the damned lies, and the statistics for us in seconds. But it cannot provide the providence of a thought. It cannot feel the moral obligation of attribution. It cannot understand why a quote from a Heinlein novel matters to a writer’s soul.

The New Bottom Line

We should stop asking when AI will “arrive” or if the pilot programs will finally hit their 100% success rate. Instead, we should ask: Who is being served by the current narrative?

If the many are to benefit—and not just the few—we must look past the “weekend of wonder” and the padded ROI spreadsheets. We must demand a version of progress that doesn’t just prioritize the speed of the output, but the integrity of the outcome.

The lies will continue to evolve, and the statistics will continue to shift. But the truth remains:

“A tool is only as profound as the intent of the person wielding it.”

The question isn’t what AI can do for you by 2030; the question is what you are willing to stand for today while everyone else is busy chasing the vibe.

©2026 Scott S. Nelson

Leave a comment

It’s not because they are more innovative.

There is a common myth in tech that smaller, nimbler companies always win the adoption race. But with Generative AI, we are seeing the opposite. While startups are still “tinkering,” enterprises are productionizing. According to recent data shared by Nathaniel Whittemore (a.k.a. NLW, host of the AI Daily Brief & CEO, Super.ai) at the at the AI Engineer World’s Fair, full production deployment of AI agents in billion-dollar enterprises jumped from 11% to 42% in just the first three quarters of 2024 [ 03:15]. Why? It comes down to a brutal reality of economics, automation, and what I call the “2% vs. 20% ROI Gap.”

AI is Automation (Just Less Consistent)

Many AI enthusiasts argue that automation isn’t AI. That’s true in the sense that not all fruits are apples, but all apples are fruits. AI is automation. The primary difference? Traditional automation is deterministic (consistent); AI is probabilistic (less consistent, but more capable). Smaller companies are already masters of traditional automation because they have to be. They use it to survive with fewer people. But for a massive corporation, the “low-hanging fruit” of basic automation hasn’t even been picked yet. This creates a massive opportunity for Information Gain -the ability to apply AI to “messy” processes that were previously too expensive to automate.

Share

The Math: The 2% vs. 20% Rule

The biggest “moat” for big business isn’t their data or their brand-it’s their Scale ROI. Because a large company doesn’t need significantly more resources than a small company to build a single AI agent or workflow, the math of deployment looks very different:

• For the Small Business: To pay for the initial R&D and resource overhead, a new AI tool might need to deliver a 20% improvement in efficiency just to break even.

• For the Enterprise: Because they are applying that tool across thousands of employees or millions of transactions, a mere 2% improvement creates an ROI that justifies the entire department.

Furthermore, as NLW points out, these large organizations are moving toward Systemic Adoption [ 17:00]. They aren’t just doing “spot experiments”; they are thinking cross-disciplinarily. They can afford to go slower, spend more on high-quality resources, and leverage volume discounts that drive their production costs down even further.

The “Risk Reduction” Transformation

Interestingly, while most companies start with “Time Savings” (the default ROI metric), the real “transformational” wins are happening elsewhere. NLW’s study found that Risk Reduction-while the least common primary goal-was the most likely to result in “transformational” impact [ 14:59]. Large companies have massive back-office, compliance, and risk burdens. AI can handle the sheer volume of these tasks in ways a human team never could [ 15:17]. This is a “moat” that small businesses simply don’t have to worry about yet.

Subscribe now

The Cycle: From Moat to Commodity

This scale is the moat that gives big business a temporary advantage. But here is the irony: The more they use that advantage, the faster the moat shrinks. As enterprises productionize these efficiencies, they effectively commoditize them. What cost a Fortune 500 company $1 million to develop today will be a $20/month SaaS plugin for a small business tomorrow. We are in a cycle of:

1. Hype: Everyone talks.

2. Value: Big companies productionize at scale.

3. Cheap: The tech becomes a commodity.

4. Reverse Leverage: Small, disruptive players use those same cheap tools to move faster and out-innovate the giants.

The giants are winning the production race today, but they are also building the very tools that the next generation of “disruptors” will use to tear them down.

© Scott S. Nelson

Leave a comment

“the AI scare trade is speeding up AI transformation by years at tens of thousands of businesses.” (Nate B. Jones, Feb 19, 2026, Why the Biggest AI Career Opportunity Just Appeared—and Almost Nobody Sees It.)

Anyone who is focused on enterprise AI adoption and heard this on Nate B Jones‘s AI News & Strategy Daily channel or read it on his substack will be concerned about how this all works out on several levels. Let’s look at a few key impacts:

• Your Investments

• Your Career

• Your Business

That list is not in order of importance; it is in order of a sequence of events in history where each of these aspects has had similar drivers. I will be referring to quotes from Why the Biggest AI Career Opportunity Just Appeared—and Almost Nobody Sees It. throughout, and I recommend you watch it after reading this because it has a lot more to offer on other topics, too.

Share

Investments

While I agree that last year was not the AI bubble, there is one coming soon. Anyone that was part of the world of tech in 2000 will see the echo here:

Meanwhile, AI startups, regardless of whether they’re good or not, look relatively more attractive to everybody. You stick AI in the name, and magical things happen right now. Magically, more capital will flow to the AI company that has AI in the name and releases an AI press release than to anybody else. (14:23)

In 1999, companies with “.com” in their name were guaranteed a huge IPO, even if all they had to offer was a cool sock puppet as their mascot. Today, a similar narrative is forming where Wall Street hype and Media hype are creating a hype-vortex, drawing in capital regardless of actual value. When the Dot Bomb blew up the Dot Com Bubble, I watched friends near retirement age having to change their life plans with no choice because it wasn’t just tech stocks that took the hit, and it took the market 15 years to rise back up to the peak of the previous century. Diversification was not a watchword in those days; even for those that were diversified, the impact rippled.

Subscribe now

Your Career

…the CFO pulling forward cost cuts to demonstrate to investors that management does take this transition seriously. Stock drop doesn’t just reflect reality, it creates reality. A company whose stock craters on AI fears is going to start behaving as if AI is an existential threat. Even if the actual tech is years away from threatening its core business, defensive postures get adopted right away. (4:58 to 5:19)

Lots of people have already had their lives upended by companies determined to show investors that they are becoming more efficient, even (especially?) if they aren’t. This reactive stance often ignores the fundamental reality that human behavior does not move at the speed of a GPU (see Your AI-Driven Digital Transformation is Impeded by Behavioral Challenges).

Share

There are lots of takeaways from the video on this, so, again, watch it after reading this. Meanwhile, a recurring theme on the channel is that if you aren’t thinking about how you can learn to use AI and then figure out how to become more efficient with AI, your career is going to be in trouble. This will come later for some industries as a whole, and for individual businesses because, contrary to media hype, not everyone is going to be making the shift at the same time and certainly not at the same pace. But it is coming, and the speed it is coming at keeps increasing.

There were a lot of people who changed careers in the late 90s into a track that was tied to the .com boom. Only a few survived and then thrived in the rebuild. Of the rest, the lucky ones were able to resurrect their former skills and return to their previous work. The rest often drifted from job to job, sometimes finding a new track, and sometimes not.

Subscribe now

Your Business

And because of the prominence of the American stock market, there are boards all over the world looking at this. Now visibility like this is what turns a slow trend into an urgent capital reallocation in favor of AI. I’m not kidding when I say the AI scare trade is speeding up AI transformation by years at tens of thousands of businesses. The scare trade is a transfer of career capital from the people who treated AI as somebody else’s problem to the people who have been invested in understanding it. (27:59 to 28:24)

The above quote is what inspired this post. I’ve been advocating for a while now that most businesses have been going all wrong about how they are adopting AI. They are buying the tools and then trying to figure out how to use them. This approach often fails to account for the competitive disadvantage smaller firms face when racing against the scale of industry giants (see Why Bigger Companies Move Faster than You in the AI Adoption Race).

Share

Companies are looking at how others are using tools and assuming it is the tool that is making it work, so they start mimicking the behavior they can see and failing miserably because what makes the tools work is not the results, it is the process of adoption and growth. That is why for every Chase or Walmart example, there are 10 AWS or Replit incidents. Many of those don’t go reported because they happened to a business the size of yours, rather than one that is currently getting media focus on a regular basis.

It’s clear that some businesses are going to rush into their AI adoption approach. Some already have. Some will be like children that touch the hot stove (healing in a couple of weeks and exercising caution), and a few may even become great chefs. Others will be more like the fictional inventor from The Expanse, but without the rest of the world benefiting from his demise:

Lesson learned? Faster is better when you build speed, not when you jump straight from 0 to splat.

Subscribe now

Post title inspired by both Zen and the Art of the Internet: A Beginner’s Guide and Zen in the Art of Archery. While the title Zen and the Art of Motorcycle Maintenance is the inspiration for the former, Zen in the Art of Archery is much more along the lines of this post and the AI adoption, though truthfully the adoption of AI, like your situation, is unique unto itself.

Looking for help adopting AI in your organization? Let’s talk. Tag me in a comment or reach out to me on LinkedIn with a connection request.

Leave a comment

©2026 Scott S. Nelson

Originally published at https://theitsolutionist.com/2026/02/21/zen-and-the-art-of-ai-adoption-surviving-the-scare-trade-to-thrive-in-the-age-of-ai/

“Going to be” is the one level where I depart from a lot of recent articles by really brilliant people. When discussing the unprecedented acceleration of new and improved capabilities that come under the media definition of AI, these experts are not only correct in their assessments of the rate of change; they understand the details of those changes better than most.

However, they often present these shifts as a present-tense reality for the masses. For the vast majority of organizations, these changes are still in the “going to be” phase because the experts are focusing on a very active and very small minority.

Share

Then there are people.

• Most CI pipelines aren’t really continuous and don’t truly integrate.

• Teams hold stand-ups and manage backlogs that aren’t the least bit Agile.

• Enterprise CRM systems are treated as glorified address books while the predictive analytics and automation features sit dormant.

• Smartphones are used for scrolling while the powerful sensors and computing power in our pockets remain largely untouched.

The main impedance to technical solutions is rarely a technical problem. The real culprits are process and culture challenges that act as a silent brake on innovation. This resistance to change usually stems from a deep-seated fear of the unknown or a perceived threat to the status quo.

When a new capability arrives, it doesn’t just offer a faster way to work; it threatens the established hierarchy, the “way we’ve always done it,” and the specialized knowledge that individuals have spent years protecting. These psychological hurdles are the biggest obstacles to adding and improving technical capabilities. It will take significant time before these new tools make it into mainstream IT departments because human behavior does not move at the speed of a GPU.

Subscribe now

A Challenge by any Other Name is…Entirely Different

This brings me to the point of my only contention with the article. I disagree with the suggestion that “transformation impedance” is a better way to think about these shifts than “epistemic flexibility under inversion.” While I find the shift in terminology problematic, Crooker’s post is otherwise incredibly thought-provoking and accurate; it is really valuable that he raised these points because they are essential to consider.

He explains “epistemic flexibility under inversion” as a capability characteristic of both systems and people to adapt to rapid changes and then adopt new approaches as a result. He goes on to suggest that “transformation impedance” may be a better way to think about it.

But branding is more important than most realize. People who take up the call of “transformation impedance” will be more likely to focus on the impedance side, which leads to conflicts between those who think everyone should reduce the impedance versus those who want to lower it. I’ll admit there is some room for collaboration on the rate of lowering impedance, but then again, there are still a lot of those CI pipelines that are still neither.

Share

First, I will admit that I had to look up the definition of “epistemic flexibility under inversion” to fully digest it:

“Epistemic flexibility under inversion” is a specialized concept often found at the intersection of Bayesian statistics, cognitive science, and information theory. It refers to a system’s (or a mind’s) ability to maintain a coherent understanding of reality even when the “direction” of information flow or the relationship between cause and effect is flipped.

Once I had this better understanding, I had the same reaction to using “transformation impedance” as an alternative as I do to changing “issue” to “challenge.” (There is a lot more to that definition, of course, and I suggest you talk with your favorite Generative AI LLM to get the rest of the picture.)

The Utility of the Negative

Media tells us we should always be positive and pursue higher goals. We buy into this because the truth is that the method of using the negative to drive action, specifically addressing an “issue,” is much more likely to succeed than the message of chasing a dream. That’s another hard truth.

I like “issue” better than “challenge” because people will deal with an issue so it will go away. A challenge makes them feel good about pursuing it, and since the pursuit is the reward, completing it removes the reward and thus the incentive. If it is an issue, the incentive needs to be to correct it.

Subscribe now

While “epistemic flexibility under inversion” may be harder to understand, it keeps the focus on how we need to change our approach to deal with the changes approaching us. “Transformation impedance,” on the other hand, is a label describing a phenomenon and doesn’t necessitate action until it is too late.

We need to flip our approach and find ways to catch up with change and not be left behind or run over. We should begin thinking about what problems need to be solved for our businesses, and even our lives, that for whatever reason we thought were too hard before, and then come up with new solutions taking advantage of the AI. To do that, we must be willing to set aside the old frameworks that impede our ability to do so.

Leave a comment

©2026 Scott S. Nelson

Originally published at https://theitsolutionist.com/2026/02/13/your-ai-driven-digital-transformation-is-impeded-by-behavioral-challenges/

The Philosophy of Ease

Under the title of the What IT Is blog is the statement: “Technology should make things easier”. The original driver behind my blog was to simplify complex tasks so that a struggle encountered the first time would be much easier the next. My focus was on tasks requiring repetition but occurring too infrequently to become muscle memory, which is a concept similar to the Second Brain framework later branded by Tiago Forte.

Under its original name and domain, my blog caught the eye of an editor at Developer.com. This led to me writing how-to articles on processes that were time-consuming to figure out but simple to execute once all the steps were gathered and sequenced. Generous copyright rules allowed me to republish these pieces after a holding period with proper attribution.

This shifted the focus of my blog toward making things easier for others. The beauty of making a task easier is that it frees people up to spend that time on more productive, interesting, and creative pursuits.

Historical Leverage: The Promise of Progress

History reveals a series of technological leaps designed to trade mechanical effort for human potential. The transition from foraging to settled agriculture allowed humanity to move beyond the daily search for calories. This newfound surplus of time provided the foundation for the birth of philosophy, mathematics, and complex governance.

During the Industrial Revolution, steam and steel began to replace human and animal muscle. Tasks that once required an entire village to complete over several weeks were suddenly finished in mere hours. This shift was theoretically intended to liberate the worker from the most back-breaking forms of labor.

By the 20th century, the “electric servant” arrived in the form of home appliances. Washing machines, vacuums, and ovens were marketed as the ultimate liberators of the domestic sphere. These tools promised to turn hours of physical toil into the simple push of a button, reclaiming life from routine chores.

The digital age followed with the promise of the paperless office and instant data processing via computers. Spreadsheets replaced rooms full of ledger-keepers, and word processors eliminated the need to re-type entire manuscripts. In every era, the pitch remained the same: efficiency would set the individual free.

Share

The Darker Side: The Persistence of Burden

Despite these advances, the time saved has often been redirected into new forms of systemic entrapment. The agricultural revolution, while providing stability, was frequently accompanied by the rise of feudalism and organized slavery. In these systems, the efficiency of the land was not used to grant leisure to the tiller but to consolidate power and wealth for the few at the top.

The industrial era followed a similar pattern of redirected effort. Rather than creating a world of leisure, the introduction of the machine often birthed the sweat shop. Workers were required to labor for 16 hours a day in dangerous conditions just to maximize the output of the new technology. In the modern consumer age, this burden evolved into planned obsolescence, forcing individuals to work longer hours simply to maintain or replace items intentionally designed to fail.

Today, the digital version of this burden has manifested as the 72-hour work week. The efficiency of the computer has not actually shortened the workday for many; instead, it has been used to increase the speed and incline of the productivity treadmill. We have built tools that cut effort ten-fold, but the saved time is often swallowed by a demand for even higher volumes of output.

The Modern Silver Bullet

The conversation around these saved hours has reached a fever pitch with the advent of AI. A recent discussion explores the idea that we may finally have a “silver bullet” for software development. This technology attacks both accidental complexity (the mechanics of coding) and essential complexity (the logic of what to build) by leveraging decades of established patterns. However, the warning remains: while the silver bullet exists, the real bottleneck is no longer the code, but the management. If leadership fails to aim this tool correctly, the result is not liberation, but a “heck of a kick” that could lead to catastrophic failure or even more grueling hours for those involved.

Subscribe now

Breaking the Cycle of Diminished Returns

There is no inherent opposition to working hard or putting in long hours. However, there is a strong stance against working to the point of diminished returns. This occurs when the final 42 hours of a marathon week produce less value than the first 30.

There is a fundamental lack of logic in developing a tool that cuts effort ten-fold only to use it 15 times as much rather than using the saved time to improve the lives of people. For business leaders, the goal should be to divide saved resources between improving work-life balance and enhancing the capabilities of the organization.

Working smarter on interesting tasks produces results far superior to grinding for the sake of volume. Technology can improve shareholder value without requiring the sacrifice of human well-being at the altar of effort.

The Question for Leadership: Is accelerating the ROI of AI initiatives worth the cost of driving people to work twice as much?

Leave a comment

©2026 Scott S. Nelson

Originally published at https://theitsolutionist.com/2026/02/10/a-silver-bullet-should-break-gold-chains/

I’m a big fan of classic science fiction. I generally avoid dystopian themes, but some are just too good to ignore, from A Boy and his Dog to Hunger Games. When ChatGPT started getting all that popular press a few years back, I was looking forward to finally living in that shiny future promised by Heinlein, Asimov, Clarke, and Roddenberry finally coming true, maybe even a flying car (the current prototypes still aren’t there yet, BTW). But the news of the last few years has had more Brave New World and 1984 vibes.

So when I read a recent NPR report on AI in schools, it felt like another example of how we are engineering frustration out of the human experience. The report describes software that is so sensitive to a student’s frustration that it pivots the curriculum before they even have a chance to get annoyed. On paper, it is a triumph of user experience; in practice, it might be a silent deletion of the very thing that makes a mind grow.

Subscribe now

The Lesson of the Eloi

When H.G. Wells sent his Time Traveller into the year 802,701, he didn’t find a high-tech utopia or a charred wasteland. He found the Eloi: beautiful, peaceful, and intellectually vacant creatures living in a world of total automation.

Wells’ speculation in his passage on [suspicious link removed] hits quite close to home in the age of generative AI:

“Strength is the outcome of need; security sets a premium on feebleness.”

The Eloi weren’t born “slow” because of biology. They were essentially optimized into that state by an environment that removed every possible hurdle. They had won the game of civilization so thoroughly that they lost the ability to play it.

The parallel to AI-driven education isn’t that the technology is failing, but that it is succeeding too well. If the machine handles every productive struggle (sensing your confusion and immediately smoothing the path), it isn’t just teaching you. It is doing the mental heavy lifting on your behalf. You don’t get stronger by watching your trainer lift the weights, even if the trainer is a hyper-personalized LLM.

Share

The Mirror of “Useful” Atrophy

It isn’t just about the classroom; AI is becoming a universal solvent for friction. History suggests that when we remove friction, we usually lose the muscle that was meant to overcome it.

• The GPS Effect: We traded the frustration of paper maps for a blue dot that tells us where to turn. The result is that our internal spatial awareness is basically a legacy system. We can get anywhere, but we often have no idea where we are.

• The Calculator Trade-off: We offloaded long division to a chip. This was a fair trade for most, but it established the precedent: if a machine can do it, the human brain is officially off the clock for that specific skill.

• The Infinite Search: We stopped memorizing facts because we treat our devices as an external hard drive for our personalities.

Not all of that has been a bad thing, unless we get to live one of those post-EMP stories (which I avoid reading to avoid remembering it isn’t that far-fetched). I, for one, am glad that Einstein said “Never memorize something that you can look up,” because rote memorization is a struggle for me, but I really do enjoy exercising mental muscle memory. Which is where using AI the wrong way will lead to an atrophy that doesn’t need a major solar event to make us realize things went too far. It doesn’t just provide answers; it simulates the thinking.

Subscribe now

The Verdict: Designing for Resistance

We should be optimistic about AI’s potential to amplify us, but we have to be wary of the passenger mindset. If we use these tools to abolish difficulty, we aren’t empowering ourselves. Instead, we are prepping for a very comfortable life as Eloi.

The challenge for educators, and for anyone using an AI “intern” in their daily workflow, is to intentionally design productive friction back into the system. We need AI that makes the work more meaningful and not just more invisible.

Mastery requires resistance. If the road is perfectly flat and the bike pedals itself, you aren’t traveling; you are just being delivered.

Leave a comment

©2026 Scott S. Nelson

Originally published at https://theitsolutionist.com/2026/01/26/the-frictionless-trap-ais-greatest-benefit-is-also-a-hidden-risk/

OpenAI’s AI for Self Empowerment blog post (January 18, 2026), highlighted in this morning’s The Deep View newsletter issue, paints a utopian picture shared across the AI industry: frontier AI closing the “capability overhang” to unlock double-digit GDP growth, affordable healthcare, and personal productivity miracles for all. Yet this empowerment narrative, echoed by Anthropic, Google, Microsoft, xAI, and Meta, hinges on controlled compute access. These six firms position themselves as gatekeepers of human intelligence in ways that blur technical promise with business-government capture. The trajectory suggests not competition, but an orchestrated march toward regulated quasi-utilities where corporate entities and defense officials co-author deployment and policy.

Subscribe now

Technical Maturity: Compute as the New Bottleneck

At its core, frontier AI’s “capability overhang” marks a pivotal shift: from narrow tools to agentic systems reasoning across complex tasks. OpenAI’s post cites power users leveraging 7x more compute for superior output, suggesting that access to reasoning and autonomous agency directly correlates with economic value. This technical fact—that AI usefulness scales with compute power—centralizes economic opportunity around massive data centers dominated by six players: OpenAI ($1.4T infrastructure commitment), Anthropic ($50B), Google, Microsoft ($80B+ FY2025 capex), xAI (Colossus expansion), and Meta ($600B+ through 2028).openai+5

The constraint is intentional. Each vendor markets “free tiers” and “accessible APIs” to seed adoption, yet premium agency—the kind OpenAI’s post credits to power users—demands enterprise or sovereign compute. Anthropic’s public sector strategy illustrates this starkly. The GSA OneGov deal in August 2025 offers Claude AI to all three federal branches for $1 per agency; a symbolic price that masks Anthropic’s deeper integration into national security infrastructure. The vendor provides “FedRAMP High” models for sensitive unclassified work, ensuring government reliance on Anthropic’s architecture. This is compute as a regulated utility, disguised as democratic access.gsa+3

Share

The Convergence: Government Infrastructure as Corporate Oligopoly

The scale and synchronization of vendor-government partnerships exceed any competitive free market. In 2025 alone:

OpenAI announced Stargate, a $500 billion data center build in partnership with Oracle, SoftBank, and MGX, with operational control remaining OpenAI’s. The Pentagon upside is explicit: Roy Campbell, the Pentagon’s Chief Strategist for High Performance Computing, told Breaking Defense that DoD hopes to “get a slice of Stargate’s computing power” to “bypass a major bottleneck.” OpenAI also lobbied the Trump administration to expand the CHIPS Act’s 35% tax credit (AMIC) to AI data centers, servers, and grid components; a request that would subsidize $1.4 trillion in company capex. The company requested federal loan guarantees, cost-sharing agreements, and a strategic reserve of raw materials.breakingdefense+2

Anthropic secured a $200 million DoD contract (July 2025) to develop frontier AI prototypes fine-tuned with DoD data, positioning Claude models for classified networks and sensitive national security applications via Palantir partnerships. Separately, Anthropic announced a $50 billion AI infrastructure buildout across Texas and New York, creating 800 permanent and 2,400 construction jobs; precisely the scale regulators cite when designating utilities as “essential.”cnbc+3

Google received a $200 million DoD CDAO contract and maintains Project Nimbus, its controversial cloud deal with Israel’s Ministry of Defense involving AI-enabled targeting and surveillance tools. More striking, Google quietly removed from its “AI Principles” (January 2025) language barring AI applications “likely to cause or directly facilitate injury to people” and “surveillance violating internationally accepted norms.” Nearly 200 DeepMind employees protested; Google saw no reason to respond. The company’s industrial logic is now explicit: if the state demands weaponized AI, principles yield.cloud.google+4

Microsoft landed a $170.4 million Air Force “Cloud One” contract (January 2026) via sole-source award; no competitive bidding. This builds on longstanding partnerships embedding Azure into military classification levels (Impact Level 5) and classified networks. Yet Microsoft simultaneously positioned itself as the “responsible” vendor, launching its “Community-First AI Infrastructure” initiative in January 2026. The firm pledges to pay full electricity rates, reduce water consumption, and invest in local training. This rhetoric—community stewardship masking federal co-investment in military infrastructure—exemplifies the modern regulatory capture: the firm captures public sympathy while the Pentagon captures the compute.finance.yahoo+4

xAI (Elon Musk’s startup) achieved the fastest Pentagon integration. In July 2025, xAI received a $200 million DoD contract identical to its competitors. By December 2025, just five months later, xAI’s Grok models were integrated into GenAI.mil, the Pentagon’s internal AI platform serving 3 million military and civilian personnel. Defense Secretary Hegseth, in January 2026, announced plans to expand Grok across all classified and unclassified DoD networks by month’s end. This 30-day deployment mandate, faster than any vendor achieved previously, signals extraordinary Pentagon confidence or extraordinary pressure. xAI’s infrastructure expansion mirrors the recklessness of competitive capture: its Colossus data center in Memphis operated 35 unpermitted methane gas turbines (422 megawatts capacity) by exploiting a 364-day loophole, violating the Clean Air Act. The EPA ruled against xAI in January 2026, but the damage was operational; xAI proceeded knowing federal urgency over “domestic AI leadership” would override local enforcement.stocktwits+6[youtube]​

Meta pursues a two-tier strategy masking military integration. In September 2025, Meta announced its Llama models—open-source and free—would be available to federal agencies via GSA’s OneGov initiative. The framing: “open-source” ensures agencies retain “full control over data.” Yet Meta’s partnerships reveal the truth. In November 2024, Meta dropped its long-standing pledge to prohibit military use of Llama. It then partnered with Palantir, the Pentagon’s surveillance contractor, to deploy Llama within classified networks and with Anduril Industries, a defense startup, to develop XR products for “real-time battlefield intelligence” guiding warfighter decisions. Meta’s $600 billion AI spending commitment (through 2028)—including a $27 billion financing deal for its Hyperion multi-gigawatt data center—positions the firm as a serious infrastructure player, rivaling OpenAI and Anthropic. The free Llama offer is a Trojan horse, seeding military adoption while obscuring defense partnerships.mediapost+5

Leave a comment

The Government-AI Nexus: Utilities for Defense, Not Civilians

The defense integration is no longer rhetorical. A January 2026 Pentagon memorandum from the Deputy Secretary of Defense outlines an “AI Acceleration Strategy” mandating that all major vendors deploy their “latest models” within 30 days of public release; a “primary procurement criterion” that effectively conscripts frontier innovation into military service. The same memo calls for AI agents to automate military workflows, from logistics to threat analysis, with “constant support both in prevention and in real-time response.”dsm.forecastinternational+1

Anthropic, Google, Microsoft, xAI, and Meta have already embedded their models into classified networks, some integrated via Palantir, a defense contractor famous for surveillance and targeting systems. Meta and Anduril announced a partnership (May 2025) to develop Extended Reality products allowing warfighters to control autonomous platforms via AI-driven perception. Applied Intuition, an autonomous vehicle software firm, raised $600 million and is collaborating with the Army on autonomous combat systems. This ecosystem—vendors, contractors, and the Pentagon—mirrors the military-industrial complex of the Cold War, except the product is cognition itself.anthropic+1

Subscribe now

Historical Precedent: Technologies of Promise Become Weapons

The trajectory from civilian empowerment to military dominance is not new. Nuclear fission promised to “power the future”; infinite clean energy for all. By the 1950s, military reactors fueled submarines and bombs. Today, the Trump administration is deploying advanced nuclear reactors to power military bases worldwide, with Clearpath and others promoting AI-driven microreactors as strategic assets. The promise of energy abundance became a monopoly on destruction.clearpath+1

ARPANET, funded by DARPA in the 1960s to enable academic collaboration, birthed the internet: celebrated as humanity’s greatest decentralized network. Yet within decades, cyberwarfare became a pillar of military doctrine. The Stuxnet attack (attributed to U.S. and Israel) weaponized the same networks that promised to democratize information. GPS, opened to civilian use in 1983 only after a South Korean airliner was shot down, now enables precision strikes accurate to meters, turning navigation aid into a tool of targeted killing.rsdi+3

Chemical synthesis promised medicine; it created nerve agents. Autonomous robotics promised manufacturing efficiency; they are now weapons platforms. AI follows the same arc. OpenAI’s blog post emphasizes “building tools which people can use to shape the future,” putting “power in people’s hands.” Yet the infrastructure centralizes that power in firms and governments, and the defense partnerships show how quickly “power” becomes weaponized agency.futureoflife+3

The Rhetoric of Capture: “Responsible AI,” “Constitutional AI,” “Community-First,” “Open-Source”

Each vendor packages military integration in ethical language:

• Anthropic markets “Constitutional AI,” a training method using ethical principles to guide behavior, presenting Claude as uniquely “responsible” for “sensitive national security applications.” Yet Constitutional AI did not prevent a $200 million Pentagon contract or classified network deployment. The ethical framing is post-hoc justification.calibredefence+1

• Google removed weapons and surveillance restrictions from its AI Principles while simultaneously claiming to deploy AI “responsibly” with DeepMind leading “ethical and responsible AI.” The internal contradiction is resolved by redefining “responsible”: it now means serving government, not constraining it.pivot-to-ai+1

• Microsoft launched “Community-First AI Infrastructure” promising electricity cost transparency, water stewardship, and local job creation; a move that preempts regulation and positions the firm as civic-minded. But these commitments coexist with $170+ million in sole-source military contracts and classified network access. The firm’s strategy is to win local hearts while ceding national security decisions to the Pentagon.govconwire+3

• xAI frames Grok for Government as providing “frontier models available to United States government customers,” emphasizing responsible deployment at speed. Yet the same firm violates environmental law with impunity, operating unpermitted turbines for over a year in predominantly Black neighborhoods. The rhetoric is window-dressing for infrastructure capture.builtin+3

• Meta positions Llama as “open-source,” emphasizing that agencies “retain full control over data” and that the free offering reduces costs and “serve Americans.” The marketing obscures Palantir partnerships deploying Llama in classified networks and Anduril collaborations building warfighter control systems. Open-source is the rhetorical mask for what is, in practice, state-managed compute infrastructure.nextgov+4

• OpenAI continues framing the “capability overhang” as a problem of underutilization: if only people had access, they could “empower themselves.” Yet every policy lever OpenAI pulls—CHIPS Act expansion, federal siting, DoD integration—concentrates power in OpenAI’s hands and the Pentagon’s.[openai]​

Share

Implications: Too Big to Fail Becomes Too Big to Trust

When six firms control compute infrastructure deemed “essential” by defense and economic planners, regulatory capture is inevitable. The symptoms are already visible:

1. Sole-source contracts: Microsoft’s Air Force deal ($170M) was awarded without competitive bidding. This is how utilities are governed: one trusted provider per region.finance.yahoo+1

2. Tax subsidies masquerading as competition: The CHIPS Act’s $280 billion, plus vendors’ requests for expanded AMIC and loan guarantees, amounts to a public investment in private monopoly. Competitors cannot match this capital intensity; consolidation follows.siliconangle+2

3. Regulatory exemptions for strategic firms: xAI’s unpermitted turbines operated for over a year. The EPA moved only when congressional scrutiny intensified. The rule changed retroactively, validating that national security overrides environmental law.tekedia+1

4. Defense dominance of innovation: The 30-day model deployment mandate means Pentagon requirements drive vendor roadmaps, not market demand or safety research. Vendors race to maximize military utility, not human flourishing.[dsm.forecastinternational]​

5. Rhetorical cover-ups: “Constitutional AI,” “responsible deployment,” “Community-First,” “open-source”—each masks the hard reality that AI infrastructure is now a national security monopoly, funded by taxpayers, controlled by corporate boards, and directed by military planners.

6. “Too big to fail” becomes “too big to trust”: When these firms face regulatory scrutiny—whether for environmental violations, labor exploitation, or antitrust concerns—they invoke national security. The Pentagon defends them, Congress defers, and accountability dissolves. The firms are rescued not because markets demand it, but because government has already surrendered control.

Conclusion: The Path Forward

OpenAI’s AI for Self Empowerment is not a lie; it is an incomplete truth. Frontier AI can amplify human capability, but only for those with access. Access flows through governments and firms that have already surrendered to military imperatives. True empowerment demands:

• Decentralized infrastructure: Open-source models and sovereign compute infrastructure owned by communities, not corporations or states.

• Transparent partnerships: Public disclosure of all government contracts, classified deployments, and military use cases.

• Regulatory integrity: Environmental, labor, and antitrust law applied uniformly, without national security exemptions for strategic vendors.

• Democratic governance: Citizens, not Pentagon planners, determining whether AI is deployed for weapons or welfare.

The capability overhang is real. The question is whether it will be bridged through concentration or distribution. Current policy, driven by vendors and Pentagon alike, favors the former. Without intervention, the “Age of Intelligence” will be an age where six firms decide what humanity can think and do. That is not empowerment; it is servitude, dressed in the language of self-realization.

Leave a comment

The AI hype bubble is the ultimate stress test for these filters.

Human Perception as
Filtering

The first point is, human perception is the result of filtering.
Actually, all perception is the result of filtering; it is just that
humans are more interested in how it affects them. That is actually part
of the filter.

Sort of like image filters, where you have adjustments for Red,
Green, and Blue, perception is adjusted through deletion, distortion,
and generalization. Some examples:

• Deletion: Everyone has at least one thing they do
  where they wouldn’t do it if they remembered how difficult it was.

• Distortion: Media algorithms that zoom in or out
  based on audience bias.

• Generalization: The core of most learning and both
  a boon and barrier to that very learning.

Subscribe now

While generalization is core to individual learning, all three
filters can be seen when groups of people are learning. Here is how
those filters are dialed at a group level:

• Deletion: Things we learned in the past that would
  help us better adopt and adapt to what is a new paradigm in
  technology.

• Distortion: Knowledge distribution through media
  algorithms that zoom in or out based on audience bias.

• Generalization: Comparing new paradigms to
  previously familiar concepts; being exposed to high-level concepts as
  parallels to common knowledge to build on iteratively, going deeper each
  time.

The AI hype bubble is a perfect example of the above perceptual
filter settings.

Share

Metaphors for AI Perception

Taking this back to the headline, an anadrome of LLM, here are five
common metaphors being applied to AI (and organized by AI, TBH) that are
worth adding to your own perceptual filters:

1. The “Alien Intelligence” This metaphor suggests
   that AI doesn’t think like a human; it is a powerful, non-human mind
   that we are trying to communicate with. It highlights the “otherness”
   and unpredictability of Large Language Models (LLMs). Best
   for: Discussing AI safety, alignment, or the surprising ways AI
   solves problems. Source: Popularized by technologist
   and writer Kevin Kelly in Wired, where he argues we should view AI as an
   “artificial alien” rather than a human-like mind.

2. The “Stochastic Parrot” This is a more critical
   metaphor used to describe LLMs. It suggests that AI doesn’t “know”
   anything; it simply repeats patterns of language it has seen before,
   much like a parrot mimics sounds without understanding the meaning.
   Best for: Explaining how LLMs work, discussing
   hallucinations, or tempering over-hyped expectations.
   Source: From the influential 2021 research paper “On
   the Dangers of Stochastic Parrots,” co-authored by Emily M. Bender and
   Timnit Gebru.

3. The “Bicycle for the Mind” Originally used by
   Steve Jobs to describe the personal computer, this metaphor has been
   reclaimed for AI. It positions AI as a tool that doesn’t replace the
   human, but rather amplifies our natural capabilities, allowing us to go
   “further and faster.” Best for: Productivity-focused
   content, tutorials, and “AI-as-a-copilot” narratives.
   Source: Originally Steve Jobs (referring to PCs);
   recently applied to AI by figures like Sam Altman (OpenAI CEO) in
   various interviews regarding human-AI collaboration.

4. The “Infinite Intern” This metaphor frames AI as
   a highly capable, tireless assistant that is eager to please but lacks
   common sense and requires very specific instructions (prompting) to get
   things right. Best for: Business use cases, delegation,
   and explaining the importance of “human-in-the-loop” workflows.
   Source: Widely attributed to Ethan Mollick, a Wharton
   professor and leading voice on AI implementation in education and
   workplace settings.

5. The “Electric Library” Think of AI not as a
   search engine that gives you a list of links, but as a librarian who has
   read every book in the world and can synthesize that information into a
   single answer for you. Best for: Explaining the shift
   from traditional search to Generative AI search.
   Source: A common conceptual framework used by Ben
   Evans, a prominent technology analyst, to describe the shift in how we
   access and process information.

Subscribe now

However you perceive the rise of LLM-based AI, include “Springboard”
in your own collection of metaphors. That is, something that helps
anyone reach higher when approached at high speed with focus…and will
trip you if you come at it the wrong way, even at a slow walk if not
paying attention.

This post was inspired by a post on LinkedIn by Dr. Thomas R. Glück. If
you have read this far, please like, comment on, and share both.

Leave a comment

If you get all the way through this post, please
tell me if the “MML” headline is an epic fail at being clever or if it
caught your eye. (Full disclosure: I use a Gem to review and edit my posts, and
generally ignore 80% of what it suggests, including losing that
headline.)

©2026 Scott S. Nelson

Originally published at https://theitsolutionist.com/2026/01/16/mll-metaphorical-language-license/

I’ve seen several tech buzz cycles, where even the real stuff is hyped. From BBS systems to .com bubbles, shareware to SaaS, DHTML to AJAX to ReST, and web first to mobile first to cloud first. In almost every one of those booms, the “first-mover advantage” belonged to the command-and-control mindset: direct, rigid, and strictly instrumental.

As I watch the rolling adoption of Generative AI (GenAI), I see a long-overdue validation of a different skillset.

The technical gap is no longer being closed by the most aggressive “commanders,” but by the most collaborative coordinators. I am delighted to see that women are not just adopting this technology, they are mastering its productivity curve at a rate that confirms what many of us have suspected for years:

When technology becomes conversational, the best communicators win.

Subscribe now

A Predictable Shift in the Trenches

In hindsight, this was inevitable. We have moved away from a world where you had to speak “machine” (syntax and code) to a world where the machine finally speaks “human” (semantics and dialogue).

I’m seeing this play out in two very specific ways:

• In Engineering: I’ve noticed women developers are often faster to move past using AI as a simple code generator. They are using it as a high-level architectural partner, stress-testing logic and managing edge cases. They aren’t just looking for an output; they are managing a relationship with a complex system.

• The Non-Technical Leap: This is one of the most gratifying shifts to watch. I’m seeing women in marketing, HR, and operations become “technical” as a side-effect of AI adoption. They are building automated workflows and custom tools that once required a dedicated IT ticket. They are bridging the gap not through brute-force coding, but through precise, collaborative inquiry.

Why the “Soft” Skill is the New “Hard” Skill

Traditional computing was about giving a machine a rigid command. If you didn’t know the exact syntax, the machine failed.

GenAI is different. It requires a dialogue.

The best results don’t come from a single prompt; they come from a back-and-forth “coaching” session. This requires empathy for the model’s logic, iterative questioning, and the patience to refine an idea rather than just demanding a result. Because women have historically been the primary collaborators and “connectors” in the workplace, they are naturally suited for the dialogic nature of GenAI.

Share

The Data Catches Up to the Reality

The industry is starting to recognize this shift, and the data is backing up what we are seeing in our offices:

• Closing the Gap: Deloitte’s TMT Predictions suggest that the rate of GenAI adoption among women has been tripling, on track to equal or even exceed male adoption by the end of this year.

• The Quality of Interaction: Recent studies indicate that while men may use the tools more frequently for “one-off” tasks, women often show greater knowledge improvement and higher competence after the interaction. They aren’t just using the tool; they are learning with it.

The Bottom Line

We are witnessing the Collaboration Dividend. For decades, “soft skills” were often sidelined as secondary. Today, they have become the ultimate competitive advantage.

It is a pleasure to see these skills—and the women who have mastered them—finally getting the recognition they deserve. In the age of GenAI, the “cooperator” will almost always outperform the “commander.”

Leave a comment

About the Feature Image

It is one colleague in particular that inspired the first spark of this post, and I wanted her to be part of the feature image. Then I began thinking of other women that have shown me the benefits of collaboration and I added their images as well as tribute. And my apologies for those I didn’t think of during the 10 minutes of creating this image prompt, or who are no longer on LinkedIn.

©2026 Scott S. Nelson

Originally posted at https://theitsolutionist.com/2026/01/09/the-collaboration-dividend-who-is-really-ahead-in-the-genai-adoption/

We all make and break New Year’s resolutions. I have two this year: Write more and be more disciplined. That is the order they came to me. I keep a journal in a note app and create a new note for each year’s entries. I never edit entries, but I do edit the opening statement that is meant to be a self-reminder each time I journal, and the edited version is:

Subscribe now

But this blog is not about me as a person other than how I deal with investing. And sometimes finances, which I will be doing more of this year, too (). But (and I don’t like to use that word, but it seems appropriate for this post), it is relevant to this blog as it will change the format.

I will be returning to the original CLI blog format of each entry being independent, rather than a page per stock with dated entries. I may or may not (the L suggests not) re-format those posts to be separate. Either way, to be sure you are finding everything relevant to a ticker you will want to search for it, and I will be consistent in include the ticker as a tag on the post. If enough folks start visiting, I will add some function that will link the posts together, but that is TBD (hint, hint, ;)).

Share

Oh, and I plan on my investments out-performing, but only make resolutions you have control over. Like following my blog (or substack, for the occasional reminder to come visit my blog).

Cheers!

Scott S Nelson

https://www.linkedin.com/in/scottsnelson1
https://medium.com/@TheITSolutionist
https://www.youtube.com/@scottsnelson1
https://money.theitsolutionist.com/

Leave a comment

Originally published at https://money.theitsolutionist.com/happy-new-you/

Almost everyone knows there are different types of hammers.
But only those who use them professionally or collect them amateurishly have more than one type.

We all know that some people are better with hammers than others. Practice is the main difference. The next level of proficiency comes from focus. You can tell the difference by the frequency of their fingers being bruised and the condition of the material being hammered along with the surrounding area.

Share

Some people are not aware of the distinctions that come about from practice and focus. They may take a swing or two and get the result they wanted, then pick up the pace too soon, with the bruised fingers, bent nails, and damaged material as evidence, if they are lucky.

They may have to hire a professional to finish the job, buy more material to replace the damaged, or spend a few hours in the emergency room followed by weeks of healing if they were not. Most are lucky. Those who are not might wind up in the news as a cautionary tale.

Hammers have evolved. They probably started as rocks or branches. Some people still try using rocks. Sometimes they have to wait until their fingernail grows back to try using a more modern tool.

The handles and heads of hammers have improved and diversified over time, which is how we have so many different types, and why some of the same type are superior to others.

Tools meant to do the same thing, the same way, but better at it.

For some hammering, the hammers have been replaced by machines that shoot the nails or bang the material. The principle remains the same, though the mechanics are very different.

Subscribe now

There was a time when nail guns were very expensive, and only a few had them, giving them an advantage. Now they are much cheaper and most professionals who need one have one.

It would be surprising if hammers stopped evolving, if there were not tools ten or twenty years in the future that would be entirely new ways of doing the same thing. Technology is like that. People are like that.

AI is like that.

Leave a comment

©2026 Scott S. Nelson

A fully supported software initiative includes people focused on coding, UI/UX, and testing. In high-performing teams, these specialists interact frequently.

Great solution teams understand that skilled “creatives” have deep grounding in data about human behavior and regularly test their work with users and refactor based on feedback and practicality 🎨 ; “testers” need to understand the limits of the technology, imagine behaviors that are not expected, and analyze the likelihood of something happening versus the impact of it happening 🧪 ; and developers who don’t test as they go, or don’t apply creative thinking to meeting business requirements may produce a lot of code but aren’t really productive 🤠 .

Yet, many organizations keep these experts apart outside of occasional “sync” meetings that don’t result in anything being synchronized but do tend to reduce productivity.

Share

Other organizations recognize that there is overlap in thinking across these specialties and try to cut costs or speed output by removing the specialists and increasing the load of the remaining experts. 🪨

People that have chosen a focus and developed the skills to be good at what they do are happiest and most productive when they are supported and challenged by people with overlapping thought processes and differing skills. 👀 These similarities in thought processes and differences in discipline are the basis of highly productive teams that thrive when leadership aligns them on a shared direction. 🛣️

Subscribe now

(Leaving out managers and architects is a peril, too, but including them here would require a much longer post).

Leave a comment