4 ms·
I can see why not believing Turing Test is a useful benchmark means that you can't deduce ChatGPT is sentient from it being able to pass one. But how do you ded
by lostmsu 4y ago
I can see why not believing Turing Test is a useful benchmark means that you can't deduce ChatGPT is sentient from it being able to pass one. But how do you deduce it is not AGI without some other reasonable benchmark? Moreover, how can you deduce GPT-4 can't spontaneously become one even in principle without having that benchmark? Those two are the questions to the skeptics really.
- dr_dshiv 4y agoRight. I don’t see why human-level AI is the criteria for AGI. I think we can have weak AGI, potentially long before we have strong AGI. We need clear benchmarks, otherwise it turns into theology.