The biggest social effects of AI may arrive after the technology is already normal: when a generation grows up that never experienced a world without synthetic intelligence.
Most people I meet are still either neutral or negative about AI.Many use it at the most basic level: draft an email, clean up a presentation, summarize a document. A lot of people deliberately stop there. They do not want to learn it more deeply because of environmental concerns, economic concerns, copyright, labor displacement, or a general feeling that something about it is wrong.I understand those concerns. But I also think we may be looking at the wrong timescale.AI may reshape how the world works first. The more radical social change could arrive later, when a generation grows up that never experienced a world without it.
The generation that debates a technology is not always the one that normalizes it
The internet is a useful analogy, but not because its negative effects were negligible. They were not. It transformed privacy, journalism, retail, politics, attention, communication, and work.The more interesting similarity is generational.Adults who encountered the internet later in life had to decide whether they trusted it, what they would use it for, and what parts of their lives belonged online.Children born into it did not make that decision. The network was simply part of reality.The same happened with mobile phones.For an adult today, the question around AI is often: should I use this?For a child growing up with GPT or Gemini, the future question may be: why would I do this without it?That is a much bigger shift.
Our objections come from a pre-AI reference world
Most current arguments about AI compare it with the world we already know.Copyright debates compare model training with how creative work used to be produced. Employment debates compare an AI economy with the jobs people currently have. Environmental debates compare inference costs with performing the same task without inference.A child born into the resulting system will not necessarily share those reference points.If their first tutor was partially synthetic, if their first creative tool could talk back, if software agents were already normal when they learned to use a computer, then AI will not feel like an intrusion into an older system.It will feel like infrastructure.They may still care deeply about ownership, energy use, surveillance, labor, or the concentration of power. But they will argue about those things from inside an AI-native society, not from the position of deciding whether AI should exist at all.
Refusing AI could eventually become a privilege
There is another uncomfortable possibility.Today, refusing to use AI can be framed as a principled position. But imagine being seventeen in 2040 without wealthy parents, expensive education, a professional network, or meaningful capital.
If an AI system can teach you mathematics, review your code, translate your work, explain bureaucracy, help you start a company, create media, and coordinate other software, then refusing it might carry an enormous opportunity cost.
That does not mean AI automatically fixes inequality. The opposite could happen at the level of ownership. Compute, energy, robotics, distribution, and capital could become even more concentrated.
Both things can be true at once: AI can concentrate wealth while dramatically increasing the capabilities of an individual who has very little wealth.
For someone born into that world, AI may not look like the force taking opportunity away. It may look like one of the few tools powerful enough to give them leverage against inherited advantage.
Children will internalize a different constraint system
The deepest change may be less about whether children use AI and more about what they learn to assume is possible.
Previous generations grew up with different constraints.
Information was scarce, then the internet made it abundant.
Communication was expensive, then phones and networks made it cheap.
Publishing was difficult, then social platforms made distribution nearly instant.
Expertise has remained relatively scarce.
AI changes that assumption.
A child who grows up with an always-available tutor, programmer, translator, illustrator, researcher, and assistant may develop an intuition that intelligence itself is callable.
That changes what feels reasonable to attempt.
Why should a small organization need the administrative structure it has today? Why should demonstrating a skill require four years of permission before practicing it? Why should a single person not be able to build something that previously required fifty?
Those questions sound radical when viewed from current institutions. They may sound obvious to someone who has been delegating cognitive work to machines since childhood.
Normalization does not mean the critics were wrong
There is an important distinction here.
A technology becoming normal does not prove that its objections were misplaced.
Cars became normal while reshaping cities and emissions. Social media became normal before society understood many of its incentive problems. Cheap global manufacturing became normal while pushing labor and environmental costs out of sight.
AI could follow the same pattern.
The generation growing up with it may eventually become more critical of how models are owned, how energy is produced, whose work is captured, and who benefits economically.
They simply will not experience non-use as the obvious default.
The real transition may happen in two stages
I suspect the first phase is what we are watching now.
AI gets inserted into existing companies, schools, professions, laws, and workflows. Adults use it to make the old system faster. There is friction everywhere because the institutions were designed for a world in which intelligence and skilled labor were expensive.
The second phase comes later.
People who never experienced the pre-AI world start designing organizations, education, media, relationships, and culture around the assumption that synthetic intelligence has always existed.
That is potentially much more disruptive than the current ChatGPT adoption curve.
The important distinction may eventually be less AI-literate versus AI-illiterate and more AI-adapted versus AI-native.
Teaching someone how to use the internet in 1998 was significant. But it was not the same phenomenon as a child growing up assuming that every person, business, song, video, map, and piece of knowledge should simply be accessible through a network.
Right now we are watching people learn AI.
The more consequential generation may be the one that never remembers learning it at all.
