Technology and Society: Who Is Shaping Whom in 2026

Every few years the question of technology and society comes back with new urgency, and in 2026 it is back with more force, as AI rules take effect and workplaces reorganize around the tools. Generative AI went mainstream when ChatGPT launched on November 30, 2022, and in the years since, the conversation has moved from novelty to something closer to unease. Governments are rewriting rules, companies are reorganizing around tools they only half understand, and ordinary people are quietly adjusting how they write, search, study, and decide. I have come to think the most useful question is no longer "what will this technology do to us?" It is a quieter one: who is shaping whom?

I want to work through that question here, not because I have a tidy answer, but because I think the way we frame it determines what we notice. If you assume technology arrives like weather, you will dress for it. If you assume people steer it, you will ask who is holding the wheel.

What Is the Relationship Between Technology and Society?

The relationship runs in both directions. Technology changes what people can do, and society decides, often slowly and unevenly, what those new abilities are for. Scholars call the first idea technological determinism and the second the social shaping of technology. Neither holds up alone.

The determinist story is appealing because it feels true from the inside. You did not vote on whether your workplace would use email. It simply appeared, and then it was impossible to function without it. But the social shaping view points out something we forget: email, like every tool, was built by particular people with particular incentives, adopted under particular conditions, and regulated (or not) by particular institutions. Different choices at several points would have produced a different email.

The honest picture is a loop. A tool opens a possibility, people use it in ways its designers did not expect, institutions react, and the reaction changes the next version of the tool. We are always somewhere in the middle of that loop, which is exactly why it is hard to see.

Has Technology Always Reshaped Society This Way?

Yes, though the speed has changed, and speed matters more than we usually admit.

Johannes Gutenberg's movable-type press, developed in Mainz around 1440, is the example everyone reaches for, and for good reason. It did not simply make books cheaper. It changed who could challenge whom. Religious and political authorities that had relied on controlling scarce manuscripts suddenly faced a world in which arguments could travel faster than the people who wanted to suppress them. The institutions of the time responded with licensing, censorship, and index lists, and some of those responses lasted for generations.

The Luddite episode of 1811 to 1816 in England is a more uncomfortable example. Parliament responded to machine-breaking with the Frame Breaking Act of 1812, which made destroying textile machinery a capital offense. The popular memory of the Luddites is that they were afraid of machines. Historians such as E. P. Thompson, in The Making of the English Working Class (1963), argued instead that they were bargaining, by force, over who would benefit from the machines, with no legitimate seat at the table. Their story is worth remembering whenever someone tells you a transition is simply inevitable and everyone will be fine on the other side.

Compare a few eras side by side and a pattern shows up:

The pattern is not that technology wins or society wins. Each time, a thing became cheap, someone gained leverage before everyone else, and institutions scrambled to catch up with a rulebook written for the previous era. The gap between the new capability and the new rules is where most of the harm and most of the opportunity live.

Why Does the Pace of Change Feel Different Now?

Two reasons stand out to me, and neither is just "things are faster."

The first is that previous technologies mostly extended our hands or our voices. A loom does physical work. A printing press carries words. Current AI systems operate on language, images, and judgment, which are the materials we use to think, persuade, and decide. When a tool touches that layer, it is not just changing what we can produce; it is changing the environment in which we form beliefs. I have written before that this is part of why AI is not a tool shift in the way a spreadsheet was. A spreadsheet did not talk back.

The second is distribution. A new factory machine took years to spread across an industry. A single software update can now reach a very large user base almost overnight. The social adjustment period, the time in which norms form and laws catch up, gets squeezed, and sometimes disappears altogether.

There is a name for the resulting trap. In his 1980 book The Social Control of Technology, David Collingridge described a dilemma: early in a technology's life, when it would be easy to change its direction, we do not yet know what harms it will cause; later, when the harms are clear, it is entrenched and expensive to change. The idea dates to 1980, and it remains a useful way to think about AI.

What Are Governments Actually Doing About It?

This is where the current momentum is most visible. Rule-making has accelerated on several fronts at once, and the approaches differ more than the headlines suggest.

The European Union adopted its AI Act as Regulation (EU) 2024/1689, which entered into force on August 1, 2024, with obligations phasing in over several years: prohibited practices and AI literacy duties applied from February 2, 2025; obligations for general-purpose AI models from August 2, 2025; and most remaining provisions, including the Annex III high-risk requirements, were scheduled for August 2, 2026, with high-risk systems embedded in regulated products following on August 2, 2027. The European Commission has proposed delaying some high-risk deadlines through its digital omnibus package, so check the Official Journal for the current status before relying on any date. It takes a risk-tiered approach: the more a system can affect people's rights or safety, the more it is expected to document, test, and disclose. The United States has leaned on a patchwork of executive actions, agency guidance, and state legislation, and the federal posture has shifted noticeably between administrations. I covered some of that churn in a week-in-review on how Washington rewrote the AI rules, and what struck me then still strikes me now: the rules are moving, but the underlying questions about accountability barely are.

I would offer one caution about reading all this activity as progress. A new law, a new voluntary commitment, and a new safety framework can each be real, or each can be a performance. It is worth asking, for any given announcement, what actually changes on Monday morning for the people affected. Rules that cannot be checked by anyone outside the organization they govern tend to produce paperwork rather than protection.

Who Decides What Technology Is For?

If I had to pick the single most important fact about technology and society, it is that the answer to "what is this for?" is rarely settled by the people who live with the consequences.

In my judgment, most consequential decisions about AI are made in three places: inside a small number of companies that can afford to train frontier models, inside the procurement offices of large institutions that adopt them, and inside legislatures that are writing rules with limited technical staff. Ordinary users mostly meet the technology after those decisions are made, as a product with a text box.

That is not a conspiracy. It is just how capital-intensive technology tends to work. But it does mean that "society shapes technology" is true only in a thin sense unless people actually have ways to push back: a vote, a complaint process, a union contract, a professional norm, a purchasing decision that says no. Where those channels exist, the loop I described earlier works. Where they do not, the loop is mostly one-directional and the language of inevitability fills the gap.

How Is Technology Changing Work and Institutions?

The most honest answer is that we are still early enough that confident predictions deserve suspicion. An observation often attributed to the futurist Roy Amara, and known as Amara's law, says we tend to overestimate the effect of a technology in the short run and underestimate it in the long run. It is usually quoted from memory rather than from a primary publication, so treat it as a rule of thumb, not a finding. It is still a useful brake on both the breathless and the dismissive.

What I can say with more confidence is what organizations tend to do under pressure. Many of them adopt visible symbols of transformation before they change anything underneath: an AI task force, a pilot, a slide with a roadmap. I have called this the difference between symbol and substance in AI adoption, and I think it explains a lot of the gap between how much people talk about AI at work and how little some workplaces have actually changed.

Individual workers feel a different version of the same tension. Using these tools well takes judgment: knowing when a fluent answer is a correct one, when to trust, and when to check. The skill that matters is not prompting; it is discernment. That skill is learned the way most professional skills are learned, by doing the work badly for a while under the eye of someone who does it well. Which raises a question I do not think we have answered: if the entry-level tasks that used to teach that judgment get automated, where does the next generation learn it?

What Happens to Truth, Trust, and Shared Reality?

This is the part of the technology and society conversation that I think is most underrated, because the effects are slow and spread thin.

When producing convincing text costs almost nothing, the old signals we used to separate careful work from careless work get weaker. A polished paragraph used to suggest that someone had spent time on it. Now polish is free. That does not mean everything is false; it means the surface of a document tells us less than it used to, and we have to read differently.

There is also the quieter risk that many people end up consulting the same few systems and receiving similar framings, so that agreement starts to look like evidence when it is closer to an echo. I keep coming back to this idea of synthetic consensus because it is so easy to miss from the inside. When everyone agrees, it feels like truth. Sometimes it is just a shared source.

None of this requires panic. It does ask for habits that earlier generations got by default from friction: checking a claim against a second independent source, noticing when you are being handed a conclusion rather than reasoning, and keeping at least some of your thinking in places where no model is listening.

Is the Fear Itself Part of the Problem?

I think it can be. Public debate about AI swings between two stories, one of salvation and one of catastrophe, and both stories share a feature: they make ordinary people into spectators. If the future is either a utopia or an extinction event, there is nothing for a school board, a hospital administrator, or a small-business owner to do except wait.

The more useful frame is the dull and difficult one. Specific systems are being deployed in specific settings by specific people, and each deployment is a decision that could have gone differently. That is where society actually shapes technology: in thousands of small, unglamorous choices about what to adopt, what to disclose, what to refuse, and who gets to object.

Urgent framing can also be a tool in its own right. Claims that we must act immediately, or fall behind, or lose to a rival, are sometimes accurate and sometimes a way of avoiding scrutiny. The test I use is simple: when someone tells me there is no time to deliberate, I ask who benefits from the deliberation not happening.

What Should a Thoughtful Person Do With All This?

The best response depends on where you sit, but a few orientations travel well:

• Treat any claim of inevitability as a prompt to ask who is making it and what they would like you to stop doing.
• Look at the layer under the tool: who owns it, who can change it, and what it records about you.
• Verify anything that matters before you act on it, especially when it arrives sounding confident.
• Protect slow thinking that happens away from a screen, such as drafting your own first view of a problem before consulting a model.

At the level of institutions, watch for three things more than any single product launch:

• Whether affected people have a real way to contest automated decisions.
• Whether claims about safety can be checked by someone independent.
• Whether the people absorbing the costs of a transition are in the room when it is planned.

If you work on governance, published frameworks such as the NIST AI Risk Management Framework (AI RMF 1.0, January 2023) and the ISO/IEC 42001:2023 AI management system standard give these tests concrete structure, including roles, documentation, and review cycles. The Luddites did not have that. We could.

Where Does This Leave Us?

Technology and society are not two separate things, one acting on the other. They are one long conversation, and right now one side of it is talking very fast. The speed is real, the stakes are real, and so is the uncertainty, which is why I distrust anyone who speaks about the next decade with total confidence.

What I keep returning to is that every previous technological shift looked inevitable from the inside and contingent from the outside. The printing press, the loom, the broadcast tower, the platform: each one could have been built, owned, and governed differently, and each time the real fight was about exactly that. I suspect we are in that fight again, and that most of it will be settled in rooms far less dramatic than the headlines suggest. The question I would leave you with is a personal one: in the rooms you are actually in, what are you being asked to accept, and what are you allowed to ask?

Last updated: 2026-10-10

Frequently Asked Questions

What is the relationship between technology and society?

It is a two-way loop. Technology expands what people can do, and society, through laws, norms, markets, and individual choices, shapes how those abilities are built, adopted, and limited. Neither pure technological determinism nor pure social control describes it well on its own.

How is AI different from earlier technologies in its effect on society?

Earlier technologies mostly extended physical effort or the reach of communication. Generative AI works on language, images, and judgment, the materials people use to form beliefs and make decisions, and it spreads through software to very large audiences very quickly, which compresses the time institutions have to adapt.

What is the Collingridge dilemma?

Described by David Collingridge in his 1980 book The Social Control of Technology, it says that early in a technology's life it is easy to change but its harms are not yet known, while later the harms are clear but the technology is entrenched and costly to change.

Are governments regulating AI?

Yes, with different approaches. The European Union adopted Regulation (EU) 2024/1689, the AI Act, which entered into force on August 1, 2024 and uses a risk-tiered model with obligations phasing in over several years. The United States has relied more on executive actions, agency guidance, and state-level legislation.

What can ordinary people do about technology's influence on society?

Question claims of inevitability, verify important information from independent sources, understand who owns and controls the tools they rely on, and use whatever channels exist, such as votes, workplace policies, professional norms, and purchasing choices, to influence how systems are adopted.