AI will be as intelligent as we want it to be

AI has been around for a long time now, and the debate about whether it will nudge us aside and take over or let us remain in charge rumbles on.

When you see an event advertised as the Artificially Unintelligent Awards, you simply have to sign up. Not for a few days yet, but you can only imagine that these awards are the AI equivalent of the Internet of Silly Things. We have had a lot of fun with this concept over the years, and perhaps we are now about to embark on the AI version.

While many applications probably benefit not a jot from being powered by AI, there is a momentum building that will drive AI into everything we do and, from there, who knows where.

Up to this point, AI has been about fast lookup and is being implemented to do repetitive tasks more cheaply and more effectively than before. This is where people believe that robots will replace their jobs, and they are right to be worried.

Now, though, the focus is tightening on what happens next and where AI can ultimately go.

There is talk of Artificial General Intelligence, where we have machines that think like humans. While this is a long way away, it is both exciting and scary in equal measure. Like Neuralink and other ways that machine and human become one.

Meanwhile, a Google Engineering Fellow called Geoffrey Hinton is putting together a roadmap to harness AI thinking and machines into heading in the same direction. His goal is to instil AI with intuition. His project is called GLOM because he is glomming together the best ideas and thinking. At least the man has a sense of humour, which may be helpful down the line.

This project is, as many things seem right now, scary and exciting.

There is no doubt that AI is useful. It is cutting costs. It is (debatably) improving customer service, and it was instrumental in producing a vaccine for COVID faster than was (literally) humanly possible. It will be used in climate change modelling and work. It will be…you get the picture.

Then, of course, there are the not so happy applications for AI. The future of security will be fought by machines, using humans as attack vectors. The future of war will be driven by AI, AI will write books, and pictures will be painted by robots. And all of that is scary news.

If the nice Mr Hinton and friends succeed, the real worry will be that AI will decide (whether intuitively or logically) that humans have really made a mess and will set about cleaning it up.

At which point AI might just make a decision about the usefulness of humans. For now, though, we can poke a little fun at the thing. Disruptive.Asia