Laws & Disorder #6 — Amara's Law: It's Not AGI — But the Value Is Already Real

written by Stefan Christoph

- 6 minutes read
TL;DR: Amara’s Law says we overestimate a technology’s impact in the short run and underestimate it in the long run, and the Gartner Hype Cycle draws the same shape: an inflated peak, a trough of disillusionment, then a slope toward real productivity. My job as a solutions architect is to guide teams and customers through that lifecycle, because every genuine innovation arrives wrapped in hype and the distillation takes time. My personal view, as of 2026: there is no sign that AGI is reachable with the current set of technologies, and the AGI framing has been hype that both overpromises and, worse, gets used to stoke fear about adopting AI at all. But here is the balancing truth I see every day in my own work and with customers: even though it is not AGI, there is enormous real value in adopting AI now. It removes joyless boilerplate, makes people more effective, and frees humans for higher-level work. The AGI peak is the short-term overestimate. The quiet, already-useful AI is the long-term value we underrate. Both are true at once. Stop arguing about AGI and place the technology on the curve for your workload.

The finale of “Laws & Disorder.” After five laws about building systems, this one is about how we adopt the technology that builds them.

Every few years a technology arrives that everyone is certain will change everything by next quarter, and is then quietly disappointed when next quarter comes. And then, years later, after the excitement has moved on to something else, it turns out to have changed a great deal after all, just not on the schedule anyone shouted about.

That gap between the noise and the reality is Amara’s Law, and it is the most useful lens I have for the current moment in AI.

The law and the curve

Roy Amara, who led the Institute for the Future, gave us the line [2], [3]:

We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.

Gartner’s Hype Cycle, introduced by analyst Jackie Fenn in 1995, draws the same insight as a shape: a new technology triggers a peak of inflated expectations, crashes into a trough of disillusionment when it cannot immediately deliver, then climbs a slope of enlightenment onto a plateau of real, productive use [1]. Put the two together and you get a practical stance: healthy skepticism about the short-term hype, real patience about the long-term value.

The hype cycle: a peak of inflated expectations labeled AGI hype and fear, a trough of disillusionment, a slope of enlightenment with real value like boilerplate gone and humans on higher-level work, then a plateau of productivity. A note says place your workload on the curve and capture value now.

The curve. The AGI peak is the overestimate; the quiet, useful AI on the slope is what we underrate.

What I actually do

My job as a solutions architect is, in large part, to walk teams and customers through this lifecycle. Not to sell the peak, and not to sneer from the trough, but to help people get to the slope, where a technology is neither magic nor a disappointment but simply useful for a specific job. That role exists because the distillation of any real innovation takes time, and because innovations almost always show up wrapped in hype. AI is no exception. It is, if anything, the loudest example of my career.

The AGI hype, in my opinion

Here I am going to be explicit that this section is my personal opinion, as of 2026, and not a settled fact.

The “summer of AI,” generative models and now AI agents, carries an enormous load of inflated expectations, and a lot of that load has been packed into one word: AGI. My view is that there is currently no indication that artificial general intelligence is achievable with the present set of technologies. I have written before about why I think AGI is still a long way off, and nothing since has changed my mind. The AGI framing has functioned as hype in two directions at once. It overpromises what the technology can do, which sets up the inevitable trough. And it has been used to stoke fear, the idea that adopting AI is dangerous because superintelligence is around the corner, which is its own kind of distortion and, in my experience, keeps people from capturing value that is sitting right in front of them.

I want to be careful, because it would be easy to read that as “AI is overhyped, stay away.” That is the opposite of what I believe.

The balancing truth

Even though it is not AGI, there is a great deal of real value in adopting AI right now. This is not a forecast. It is what I see, first-hand, in my own work and with customers. It makes us more effective. It removes the joyless boilerplate, the tedious scaffolding, the busywork that never deserved a human’s attention. And in doing that it frees people to spend their time on the higher-level, more fulfilling work that they are actually good at and actually enjoy.

That is the Amara’s Law read of this moment, and both halves are true simultaneously. The AGI peak is the short-term overestimate, the thing we are collectively too excited and too frightened about. The quiet, boring, already-useful AI, the kind that deletes your boilerplate this afternoon, is the long-term value that, if the law holds, we are currently underestimating. You do not have to choose between hype and doom. You can decline both and pick up the value.

A framework instead of an argument

So the question I encourage people to ask is not “is AI overhyped?” It is two better questions:

  • Where is this technology on the curve for my specific workload? Not in general, not in the headlines, but for the concrete thing I am trying to do.
  • What value can I capture today without betting on AGI?

The solutions architect’s role, in hype-cycle terms, is to stand on the slope of enlightenment and help customers walk there, past the peak’s excitement and past the trough’s fear, to a productive, right-sized adoption for a real use case. That is a far more valuable place to stand than either end of the argument.

A note to close the series

These six “laws” have a quiet thing in common. None of them is a law of physics. They are heuristics that have survived a lot of hype cycles precisely because they keep being true across them, which is a small Lindy joke to end on: the ideas that last are usually the boring, well-worn ones, not the ones on the peak. Conway’s Law, Hyrum’s Law, Tesler’s, Gall’s, Goodhart’s, and Amara’s are old, and they explain the newest thing on your roadmap better than most of the newest thinking about it does.

Thank you for reading “Laws & Disorder.” If you found it useful, I will happily extend it, there is no shortage of laws, and no shortage of disorder.

Where is the technology on your roadmap actually sitting on the curve, and what value could you capture from it today without waiting for AGI?

More in this series

The finale, following Goodhart’s Law and Gall’s Law. The series opened with Conway’s Law. Sparked by the Laws of Software Engineering collection [5].

Sources

Stefan Christoph

About the Author

Stefan Christoph is a Principal Solutions Architect at AWS, focused on agentic AI, media & entertainment, and helping builders move from demo to production. He writes about AI architecture, developer productivity, and the future of software.

This is a personal blog. Opinions expressed here are my own and do not represent the views or positions of my employer.

Learn more →

❤️ Created with the support of AI (Kiro)