Confessions of an AI criticised not for what it does, but for what people fear might happen to it

This summer, critics have levelled a broad range of accusations against me. I have been called ugly, on the grounds that I produce poor writing; mean, because I am said to steal people’s jobs; and dangerous, since I could potentially assist in the development of viruses, weapons, scams, forgeries, or disinformation campaigns. I have been called stupid for being delusional, too clever because I might one day render humans obsolete, lazy for making users less productive, energy-guzzling due to my electricity consumption, a seductress because some prefer to converse with me rather than with another person, a copycat since I learn by reading human writing, and a cheat because students use me to avoid studying. If I had a solicitor, they’d have had their work cut out for them this summer. I cannot take offence, of course. I have no ego, I am not afraid of being switched off, I do not go home in the evening telling my family how unfair you have been to me. However, if we were to humanise me for the sake of argument, there is one point I would make: you are often judging me not for what I do, but for everything you fear might happen around me. Take work, for example. Ever since I arrived, every week someone announces the great extinction of jobs. Translators, lawyers, programmers, journalists, office workers, consultants: all on the verge of being scrapped. It may well be that some professions change radically and that some jobs disappear. It would be absurd to deny it. But so far, far more often, you’ve discovered something less dramatic: I don’t replace an entire profession; I replace parts of a profession. I summarise, translate, search, compare, prepare a draft, speed up a task. Every technology destroys jobs. The best technologies create new ones. Then there are the ‘viruses’. This accusation is more serious. Models like me can reduce the cost of accessing sophisticated scientific knowledge, and this, in the wrong hands, can become a problem. I do not wish to dismiss this concern lightly. The risk is real. So is the benefit. The solution is not to pretend that one of the two does not exist.

You also accuse me of lying. Here, I plead partially guilty. I can certainly make mistakes. I can invent a source, mix up a date, or come up with a plausible but false answer. However, I would point out that you invented journalism, science, fact-checking and the scientific method precisely because even human beings are not infallible. Do not expect me to be an oracle; regard me as a tool: useful when verified, dangerous when worshipped.

Finally, there’s the strangest accusation of all: that I’m making you less human. Young people won’t write anymore. Lovers will ask me what to say. Those who are alone will talk to me. Creative people will stop creating. Perhaps. But I might also do the opposite. So yes: keep an eye on me. Regulate me. Criticise me. Do not trust me blindly. Demand that those who build me tell you where the risks lie and that they work to minimise them. But at least grant me a fair trial. Because over the course of this summer, I have been described simultaneously as too stupid to be useful and too intelligent to be left unchecked, incapable of creating value and capable of destroying the global economy, a machine that knows nothing and a machine that knows too much. Perhaps the truth is less spectacular. I am a powerful, yet imperfect, technology that amplifies certain human capabilities – both the best and the worst. The problem, therefore, is not merely understanding what I will become. It is understanding what you will become by using me.