@VictorGallagher What are your current thoughts on this? Would you say AI meets your criteria? I'd argue language models still struggle with a lot of basic arithmetic that high schoolers should be able to do with ease, e. g., multiplying together 20 digit numbers without using a Python extension or internal calculator to do computation. And if it can't do this, it is not yet general enough for this market, even if it can do far more advanced things.

@spiderduckpig The latest AI models can do a lot more than high school math now. I went with high school academic performance because back in the eighties a high school friend of mine was running a store in the mall at the age of 16. So I reasoned that if an AI can match that then it should be able to do real world work. But it seems that things did not work out that way. We are not going to get to AGI by adding more and more knowledge. We are in effect making the pie bigger but it is still missing a slice. Perhaps with world models we will be able to close the gap. It could also be that there is something less tangible involved, such as self reflection or personal evolution that a digital environment may not allow.

@VictorGallagher so current ai, which can do highschool math and logical reasoning, doesn't count for your market? just making sure i got this right

High school math and "some logical reasoning" seems like a very low bar for AGI

I can see LLM getting good at high school math without being nearly as good as humans elsewhere, and can't current ones do some logical reasoning already?