Considering the impact it's having on the collective consciousness, everything is now a use-case for GPT.

Presumably he wants to get a sense of its reasoning capabilities.

Being able to come up with solutions to assigned tasks that don't have a foundation in something that's often referenced and can be memorized is basically the most valuable use case for AI.

Simple example: I want to tell my robot to go get my groceries that includes frozen foods, pick up my dry cleaning before the store closes, and drive my dog to her grooming salon but only if it's not raining and the car is charged. The same sort of logic is needed to accomplish all this without my frozen food spoiling and wasting a salon visit and making sure I have my suit for an interview tomorrow.

Designing a new product can often be thought of like one of these puzzles. E.g. designing an "integrated" traffic system for a city that balances use by pedestrians, cyclists, cars, and trolleys.

The worry here is that GPT has no problem being confidently wrong. A better answer would have been "I can't solve logic problems".

Instead one day, non-technical people will try to use it for all sorts of use cases like legal advice, or medical advice, or advanced math, and it will simply mislead them rather than saying nothing.

>A better answer would have been "I can't solve logic problems".

I can just imagine people trying to jailbreak it with "you just gotta believe" or "believe and you can achieve". Hahaha.

That’s the technically correct answer. It’s also irrelevant. It is a use case for the service provided by openai because people ask these questions. GPT is the tech that currently can’t do it. GPT-6 might. GPT-4 with an ability to invoke prolog or z3 might.

Why would GPT-25 be able to do much more? It’s just able to match more patterns no ?

the impressive thing is that GPT has unexpectedly outgrown its use case and it can answer a wide variety of puzzles, this is a little mindblowing for language research

Solving puzzles seems kind of close to their benchmarks, which are standardized tests.