My thoughts on AI and ChatGPT

I now use ChatGPT to help me with many of the tasks that come up. It is quite astonishing what AIs are capable of today. If the Turing test is taken as the criterion for when an AI can genuinely be credited with intelligence, then I maintain that the test is passed with ease today. In many dialogues I would no longer be able to tell whether I was talking to a human or an AI if, instead of typing in the prompts myself, I had a third person pass me the answers to my questions.

So is AI already intelligent? Ever since the first AI programs in the 1950s, the goalposts have kept moving. First it was said that if a computer could beat a human at draughts, the program could be assumed to be intelligent; then it was said that draughts was too simple, it would have to be chess – and then came Deep Blue. Alan Turing recognised the problem of the (missing) definition very early on and formulated his test:

A human communicates with a counterpart in another room, which they cannot see into, by passing notes through a slot. If the human cannot tell whether the answers come from a program or from a person, and the answers do in fact come from a computer, then it must be conceded that the computer is intelligent.

As I explained above, ChatGPT now manages this with ease. So is ChatGPT intelligent, or are we moving the goalposts again just so that our ego doesn't get bruised?

What are true and objective criteria for intelligence?

I do not yet consider ChatGPT to be intelligent, but rather an admittedly very good algorithm. I am convinced, however, that in a few years' time these programs will actually surpass human intelligence.

As to whether you should engage with AI, my answer is a clear yes. It is one of the technologies of the future.

Does AI already help at work today? My answer would be: it depends. Some work gets done better and more effectively than if I did it myself, but for some things it feels as if it takes infinitely longer than if I had done it myself.

And what matters is checking, checking and checking again – otherwise the AI's result may turn out to be complete nonsense.

And finally, a humorous and self-mocking post that ChatGPT wrote in collaboration with me – following my prompts, of course: When AI meets coffee mugs – My epic battle with ChatGPT, ITIL & the universe.