3 ms·
>It feels more like an argument that chatbots will never exhibit general AI. I feel it's even more specific, that the current approach to training and building
by marcinzm 7y ago
>It feels more like an argument that chatbots will never exhibit general AI.
I feel it's even more specific, that the current approach to training and building chat bots will never exhibit general AI. Theoretically, you could make chat bots in other ways such as directly simulating a human brain down to some arbitrary level. Maybe there's things in-between as well.
- roywiggins 7y agoOh, yes, that's what I'm referring to as a chatbot- something built specifically to converse by modeling natural language, rather than trying to model cognition at any deeper level. Not that we can do that successfully either, but at least trying to tackle that has an honest chance of working. This sort of feels like an updated version of Searle's argument... a chatbot is an awful lot like his 'Chinese room.'
- marcinzm 7y agoGot it. I think the paper actually makes a reference to 'Chinese rooms.'