The rhetoric surrounding AI deserves more suspicion than the technology itself.
A company that decides to insert an AI assistant into its software is making a business decision.
But describe it as a response to “the AI revolution” and the decision acquires the aura of weather. Nobody is responsible for rain.
AI has absorbed extraordinary amounts of capital, computing capacity, electricity, engineering talent, and executive attention.
The people making those investments, therefore, have an obvious incentive to persuade everyone else that the investment was not simply a gamble.
That incentive explains the grand predictions.
AI will transform every industry. AI will reinvent education, medicine, science, entertainment, government, software, warfare, logistics, and ordinary office work. AI will make individuals radically more productive. AI will produce an abundance of intelligence. AI will change civilization.
But a technology can be powerful without being universally useful. The internet changed communication without abolishing spam or mediocre advertising.
Historical importance does not imply universal benefit.
AI Adoption Is Not a Historical Force
The people with the greatest ability to shape the technological environment are also the people most likely to benefit from describing its direction as beyond their control.
Software companies can choose whether an AI feature is optional, mandatory, accurate, private, interoperable, or simply decorative.
Ordinary users possess none of that leverage. They can accept the new interface or find another job, another service, another platform, another school, another workflow. The rhetoric of inevitability converts this asymmetry of power into an asymmetry of responsibility.
The executive says, in effect, “We had to do this because the technology was coming.” The worker is then told to acquire new skills.
The bill travels downward.
The People Selling AI’s Inevitability Are Trying to Make It Inevitable
Look closer and you see AI’s loudest evangelists not behaving as though its triumph were inevitable, even if they constantly tell you so. They lobby governments and seek favorable regulation and flood public discourse with predictions about what AI will soon accomplish.
They want people to believe that the future has already been decided while they simultaneously work furiously to decide it.
If it’s so inevitable, surely you wouldn’t have to put so much time and money into making it inevitable, right?
AI’s Social Impact Depends on Who Controls Its Adoption
The fundamental mistake in much of the AI discourse is that it imagines the individual as an isolated chooser standing before an infinite menu.
Actual people are social creatures. We navigate culture through institutions and relationships. We trust editors, teachers, colleagues, critics, friends, artists, and communities. We develop preferences inside shared worlds.
Much of the current excess will disappear. Corporate “AI strategies” will be re-branded. Startups whose only distinction is a thin interface around a general-purpose model will evaporate. Many grand predictions will age badly.
Whatever will be worth keeping will not require worship. A reliable database software will not be sold to you on the premise that human memory is obsolete.
Technology earns its place by solving problems at a cost people accept. AI should face the same test.
If AI raises productivity but allows firms to capture nearly all of the gains while workers receive greater surveillance and weaker bargaining power, society has not experienced a technological miracle, but a(nother) transfer of economic power.
There is nothing inevitable about allowing private companies to determine the technological environment in which public institutions operate.
These are business models and political choices.
AI Should Have to Prove Its Value
The sensible response is to stop granting AI’s promoters the authority to define the terms of its success.
Demand evidence before replacing people. Make companies compete on usefulness rather than novelty.
Let failed products fail.
All the AI boom needs is customers who are difficult to impress.
Things on my Desk
The Forbes 30 Under 30 lore gets a LOT messier when you look at what happened after the cover story. 👀
This tracker rounds up alumni who later faced fraud allegations or full-on company collapse.
Just a good reminder that someone raising $50m doesn’t mean they’ll build a great company.
From the Same Desk
- Productivity must belong to the producer. A software company does not charge less because its engineers improved deployment.
- If a business repeatedly wastes resources, customers leave. Few other institutions face consequences this immediate.
- Fast fashion is being financially repackaged. The Vinted app built a secondary market that lets consumers keep buying the same volume with a cleaner conscience.