[int0x2e]

There's little doubt in my mind we're not yet close to making self-aware models or anything of that nature, but I think it's hard to deny we're getting closer at an incredible pace. The most telling metric to me is not what they can or can't do, but rather, how similar their errors and patterns are to children. My 4 year old seems to share more with GPT's hallucinations than I'd initially guess, and as time goes by I see more and more of these small similarities. These models aren't there yet, but they will get to the point where they would at least fool the best of us, at which point you must ask yourself if our wetware has any special merit over their hardware...