What do you think is the future of AI?
View all responses
I suspect AI will become ordinary before we agree on what it is. It is already useful for some jobs, absurd for others, and very good at sounding certain. That last talent may be its most successful feature.
The dramatic version of the future has machines waking up and taking charge. The less dramatic version is that we let them make more and more small decisions because it saves us a minute, then another. What to read. What to buy. Which form gets approved. Who gets flagged for a closer look. None of that needs a machine with a sinister laugh. A dropdown menu will do.
This photograph shows police officers standing behind shields. It doesn’t tell us what happened before or after. A photograph rarely gets to be that generous. But it does bring the question down to street level, where technology stops being an idea and starts affecting people who may have had no say in it.
AI can help with tedious work, and with work that is genuinely hard. I use it; I’m not waiting for the gears to start turning so I can flee to the hills. But it can also make plausible rubbish quickly, and make it harder to tell who is responsible for it. A confident answer is not the same thing as a considered one. We’ve had enough practice confusing the two.
I don’t think the useful question is whether AI will become more powerful. Of course it will. The question is whether the people and institutions using it will explain what it is doing, and answer for the consequences. “The system decided” is a neat way to end a conversation. It is not a particularly good reason.
So I’d like AI to be useful, inspectable and unremarkable. No oracle, no demon, no new household god demanding a subscription. Just a tool whose limits are visible, and whose mistakes still belong to the people who put it to work.
That may be asking for the least fashionable feature of all: someone willing to take responsibility.