I think we overestimate both the good that will happen and the bad that will happen. What ends up being good is often not what people thought would be good and, similarly, what ends up being bad is often not what people thought would be bad. We typically underestimate what actually becomes the benefits and liabilities of an innovation.

The pre-LLM history of AI was filled with plenty of optimism about how AI could help us as a competent and impartial assistant, that it would free us up to make better use of our time and engage in creativity, society and family. It was also filled with fear — fear fuelled by our collective failure to appreciate consciousness as something separate from intelligence, a fear that AI would achieve sentience and become a direct enemy of humanity. Woven together, these views form a significant part of the fabric of science fiction. We now know these fictional futures are not the ones that lie before us. When people now talk about the prospect of AI devastating humanity, it won’t be because of malicious intent, dark design or sinister forces; it will be an embarrassingly banal side-effect of something that could have been prevented with a better prompt or setting a particular config flag or installing a particular patch.

All technologies bring with them liabilities, with many of these being more about us and how we behave than necessarily being intrinsic to the technologies themselves. This is not to say that we should consider technology neutral — a common claim, but one that deserves to be challenged as, well, life is not that simple. The set of possible relationships we have with a technology is part of the technological proposition, not separate from it. To say otherwise seems naïve or intellectually dishonest.

The long-term benefits of technologies can be imagined, but are sometimes difficult to discern because any popular tech changes the world it’s in and, therefore, what becomes simple and possible and even inevitable in terms of exaptation and other technologies. The twentieth-century dream of the videophone happened in the twenty-first century, but it did not come from traditional telecommunications companies trying to invent the videophone, a goal they pursued for half a century. Videophones emerged from the haphazard growth and convergence of many separate technologies — the web, broadband, webcams, laptops, mobile phones, WiFi, 4G, etc. — until they arrived without fanfare or even the label ‘videophone’.