I think I've written about the Turing test twice now, and it's always fun to revisit the concepts. Just an observation: I'm pretty sure that Turing would have been quite satisfied with today's computers as "exhibiting intelligence", whatever that slippery phrase really means. I think he would see this and say, "yes, this is exactly what I'm talking about."
It has only been in the ensuing decades that our own cultural perceptions about AI have evolved, with us demanding ever more before we say, "yes, this is convincingly human."
I think Turing would have rejected ELIZA, but thumbs-upped ChatGPT.
Agreed absolutely. I think Turing would have definitely considered GPT-3/4 to have qualified the imitation game. The same can be said for Gemini Pro and Ultra. But in my humble opinion, I think that would be a wrong conclusion to reach. The test that Turing devised was genius for his time. But I think some updates are needed for testing 'intelligence' in AI, whatever that means.
Turing's concept was brilliant, but I also don't think he was saying that passing his test represented AGI. I think Turing was really trying to get us to ask what it meant to be intelligence.
Mission: accomplished! Well done, Mr. Turing. Well done.
I think it's mainly an oversimplification, but it's also not super noteworthy to say, "Hey, check out this weird way to think about human intelligence!"
I think I've written about the Turing test twice now, and it's always fun to revisit the concepts. Just an observation: I'm pretty sure that Turing would have been quite satisfied with today's computers as "exhibiting intelligence", whatever that slippery phrase really means. I think he would see this and say, "yes, this is exactly what I'm talking about."
It has only been in the ensuing decades that our own cultural perceptions about AI have evolved, with us demanding ever more before we say, "yes, this is convincingly human."
I think Turing would have rejected ELIZA, but thumbs-upped ChatGPT.
Agreed absolutely. I think Turing would have definitely considered GPT-3/4 to have qualified the imitation game. The same can be said for Gemini Pro and Ultra. But in my humble opinion, I think that would be a wrong conclusion to reach. The test that Turing devised was genius for his time. But I think some updates are needed for testing 'intelligence' in AI, whatever that means.
Turing's concept was brilliant, but I also don't think he was saying that passing his test represented AGI. I think Turing was really trying to get us to ask what it meant to be intelligence.
Mission: accomplished! Well done, Mr. Turing. Well done.
This I such a crucial point that is often missed. The Turing test is often misrepresented that way.
I think it's mainly an oversimplification, but it's also not super noteworthy to say, "Hey, check out this weird way to think about human intelligence!"
YAWN.