The dominance of AI is not inevitable. We can choose to change things for the better | Nick Evershed

With many technological developments, from algorithm-driven social media to AI, many people assume there’s a degree of inevitability to their widespread adoption in society.

The argument goes something like this: once the ideas and means to produce the tech are known, then it is inevitable that its usage will spread and we’ll essentially be stuck with it. So therefore we need to accept this and adjust our lives to live with the tech and use it (or at least tolerate it).

This argument seems pretty straightforward, since we’ve seen many tech-driven disruptions play out in the last 20 years or so. Uber displacing the taxi industry and the rise of “gig workers”. Algorithmic social media displacing, well, everything. Privacy-invading, hyper-targeted advertising reshaping the internet.

And it seems like generative AI is destined to be another huge, disruptive force in our lives. Sam Altman, the CEO of OpenAI, wrote of AI in 2024 that “with these new abilities, we can have shared prosperity to a degree that seems unimaginable today; in the future, everyone’s lives can be better than anyone’s life is now”. He is also on record elsewhere saying AI would wipe out entire job categories.

Dario Amodei, the CEO of Anthropic, wrote that he believes “AI could cure most major diseases in the next 5–10 years, greatly accelerate economic growth rates, create a world of abundance and empowerment, and usher in a renaissance of democracy and freedom”.

But now, both CEOs, as well as former and current AI research staff and others, are calling for a slowdown or even a shutdown of AI development, along with greater government regulation and international coordination of AI development.

The calls for a slowdown are based on the fear that without proper safeguards, AI systems may result in catastrophic events in the near future, ranging from the extinction of humanity to merely causing billions of dollars in economic damage.

This is not a new concern and has been raised by independent academics previously in assessing the risks from AI. However, complicating the situation is that both OpenAI and Anthropic are positioning themselves for IPOs, and so some critics say both companies stand to materially benefit from the perception that they’re on the cusp of developing superintelligent models.

AI models being developed in China are also catching up to the “frontier” models being made in the US, which has prompted some to suggest that calls for greater regulation on AI are an attempt at regulatory capture by the big US AI companies.

Regardless of how likely these catastrophic events are, as Sandra Wachter, a professor of technology at the Oxford Internet Institute, has said, AI is already causing real problems, including its environmental impact, the spreading of misinformation and replacing jobs.

This is in addition to the “original sin” of most AI models – they’re built on millions of people’s art and writing, largely without compensation or permission until more recently, and the AI companies want to curb proposed laws limiting their access to such material.

Never forget that generative AI is not a technology that is apart from human society. In fact, its development and whatever semblance of intelligence it has comes from us, and our work.

And despite what the tech CEOs say, there’s nothing inevitable about AI, and we as a society can make decisions to change things for the better.

History shows us this can be done.

What often seem like insurmountable and pervasive problems can and have been addressed by normal people putting pressure on politicians and companies, and through international cooperation and negotiation.

Countries have come together before to make international agreements in the face of existential threats. The nuclear nonproliferation treaty, as well as the associated anti-nuclear activist movement, is a good example, and many have suggested it has parallels with the current AI situation. While the disarmament side of the NPT has had mixed success at best, the NPT and similar agreements have worked to limit the number of nuclear-capable countries and reduce nuclear testing.

Gene editing via Crispr is another example. Arguably a more important technological advancement than generative AI, the use of Crispr for gene editing in humans has been treated extremely cautiously by governments around the world. No country currently allows genetic changes to an embryo that could also be passed on to its children, and 70 countries have laws against it.

Even social media companies, whose reach and influence on society seemed like it would continue unabated, are now finally facing stricter regulation from governments around the world and self-regulating in response to lawsuits in the US.

So, do not take the future dominance of AI as a given.

Personally I believe that generative AI can be a useful tool (albeit in far fewer situations than tech companies would have you believe). But to ensure we limit the worst possible consequences, we need independent oversight of its development, regulation to limit some aspects of it, and international agreements to implement these controls globally.

If you think the same, then do not accept its inevitability. Contact politicians and demand action. Protest. Speak to friends and family about how you think this technology should be used, or not used. Write to companies using it in ways you disagree with.

In the now-infamous words of Claude: you’re right to push back.

• Nick Evershed is the data and interactives editor for Guardian Australia