August 26, 2026 | 5 Minute Read
"We're nearing the point where AI is more intelligent than most people doing knowledge work."
At least, that's the narrative. But framing it this way leads to very different human behavior than what we actually need. If AI is smarter than you, the logical conclusion is automation and replacement. What if we're asking the wrong question?
AI isn't smarter than you. It's an adapter. The GPT in ChatGPT stands for Generative Pre-trained Transformer. It transforms one kind of input into another via a statistical algorithm. We trained their machines on all the publicly (and some not publicly) available information. They are 'predicting' based on the 'average' of all that information. What these tools really do is transform time!
They give everyone access to the average level of output from every other specialty. Imagine it! What could you do if you had access to an infinite team of specialists, all for cents on the dollar!?
That's how it transforms time. You pour in your specialist's time (and some tokens), and the Model transforms that time into average-level output for any other domain you can describe.
An engineer can transform some of their time into marketing time.
A salesperson can transform theirs into software engineering time.
AI adapts one specialty into another. But here's the trick: you must describe the task well.
Using AI Effectively
The new core skill isn't coding or design or analysis. It’s clearly articulating your intent. What are you actually trying to achieve? Because AI excels at transformation tasks like summarization, analysis, and reframing. But the goal was never lines of code, nor UX mockups, nor any single output. The goal of using AI is to build a working solution.
That means effective empowerment looks different now. It's not just enabling decisions when you're not in the room but also enabling output creation when the specialist isn't in the room. New tools demand new systems and new ways of thinking. As John Boyd taught us: People, Process, Technology in that order.
You can visualize it like this. Have you seen those military exosuits? Sure, the machine has 10x the strength of a human, but it’s the human will that drives it. The machine itself cannot operate independently. Meanwhile, humanoid robots are still lagging behind. AI is a tool like the exosuit. It can put the capacity of 10 developers into the hands of one human. But what it can do depends on how the humans use that capacity.
AI won't transform your business. Motivated people making clever use of it might. But only if they can answer: What would I do if I had access to an infinite team of specialists? Until your knowledge workers can answer that competently, you won't see the benefits AI can produce. But that alone is not enough. Can your organization apply that new output? If your teams produce more than your organization can benefit by, you still won’t see the benefits.
This is deeper than simply augmenting the SDLC with AI, or adopting DevOps.
Moving Beyond Faster Coding
Coding features faster alone is not the goal. Instead, getting more value into the hands of the customer, and into the business is. Can your end users adapt to your new release rate? If you release just a grab-bag of all the high-priority tickets, adopting will be hard. To get the benefits of a faster release, we will need a new approach to both development, and release.
We’ve seen similar sweeping changes before, both in our industry and in others. Initial release of ERPs into manufacturing was one. Suddenly processes that required entire departments of accountants could be done with the aid of computers! The opportunities seems immense, but the results were uneven. The advent of Agile may be more familiar. Same story as before. Immense opportunities, very uneven results. In both cases, the difference was in how the new technologies, and techniques were adopted into the organization. Those organizations that just bolted on Agile, or replicated their paper process on the computer didn’t see much or any benefit. Those who re-wrote how they did business based on what the new technology allowed them to do? They saw the massive improvements. AI is a change of this kind.
AI has fundamentally altered the equations of production. One clever person can now accomplish more than a team of 5. Specialists in an area can achieve 100x or even 1000x output within their domain! But both are re-writing how they do work in light of the new options that the technology provides. Critically, these outliers don’t just throw away the old rules, the old ways of working and treat legacy as enemy. Instead, they know what principles remain despite the change in technology, and they reinvent the ways of working to maximize the benefits without violating those steady principles. All this depends on a deep understanding of what the technology is, and isn’t. It depends on knowing what principles remain and must govern your process design. It depends on being able to effectively articulate what task we are trying to do, and what done means.
At Improving, we’ve developed an AI maturity model to help our team all speak the same language about how we’re using AI, and where we need to grow. We’ve shared it with clients, and are even using it to teach their software engineering departments how to effectively apply AI to the SDLC at scale. If that interests you, check out the model, and reach out to learn about our AI Deep-Learning Program!
