Great explanation. I'd heard of the thing but when I went to play with it, it was gone. The others on YouTube who had a play were impressed.

Your article finally tells me why one should bother being impressed, although to not be impressed by any of the LLM's speaks to how quickly we humes take technological miracles for granted.

So if this is a Q* type beasty and gets a bit of time to think about its answer, then it will be very useful indeed.

If the prompt is, take as much time as you need to evaluate your answer internally, and this is combined the agentic frameworks so that its answers become the seeds for more questions, perhaps we'll end up with a truly amazing thing.

It is when the big houses make a machine that replies 'yep, climate change, do this', 'yep, fusion reactions, do this', 'yep, human immortality, just do this', 'yep, peace on Earth, do this' ... that's when folks will take serious notice.

Instead it will be a roll out of very capable models that will get incorporated into any number of service centres and medical facilities and legal practises and more, and a whole bunch of white collar folks will lose their job or have to work for peanuts to compete.

or

Folks will figure out how to use these bottled intelligences to do their job even better.

Time will tell.

My guess is it will be a combination of both.