Home
Starting Up
Case Studies DB
Products
Ideas DB
Vibe Coding Tools
Subscribe to IH+
Starting Up
Case Studies
Ideas DB
Products DB
Join
13
Likes
8
Comments
Google Engineer on his sentient AI claim
by
uxland
https://www.youtube.com/watch?v=kgCUn4fQTsc
It's trippy how it has a sense of humor. I don't think that itself means it's sentient. If it itself found the joke funny, it would indicate it feels emotion. But it could have just learned what people could find funny - or at least what the person it's speaking to at the time finds funny. So, when faced with a trick question, it might have just learned to get out of a corner with a joke? Smart, no doubt - but sentient?
Exactly, I think this engineer's perception when it interacts with the AI is that it's sentient. It doesn't mean the AI is expressing what the engineer thinks it's expressing. To use a very crude analogy: When chimps look like they're smiling (when their lips are pulled back to show their teeth), although to humans it might look like they're happy, this is actually their "fear grimace" - it means they're afraid, unsure, or stressed. Just because we perceive another being/AI's behavior/reaction/expression as one thing, doesn't mean it actually is.
Right, but the engineer agrees with this - he agrees they should run the 'Tiering test/thiering test?' (not sure how to spell this...). I thought this news was totally ridiculous, especially because the media has painted him as a bit of a hippy clown. But this video is making me rethink - I think he's got a point. He's not anywhere near as conclusive as he's been made out to be - all he's saying is: more work needs to be done to understand this arena, and Google's current policies are preventing this from happening.
Same happened in 1964 when the first chatbot ELIZA. It was a very simple chatbot yet many early users were convinced of ELIZA's intelligence and understanding, despite Weizenbaum's (builder) insistence to the contrary.
The problem with vague words like "sentient" is that you can always change/modify the definition just a bit and come up to a different conclusion.
I've been seeing people discuss whether Google's AI is really sentient, what sentient means, etc.
If you want to have a productive discussion, first start with a more specific definition of how you'll MEASURE 'sentient' in an AI.
This won't happen anytime soon I think because the media loves to tinker around with vague words (the more vague something is, the more they can write about it).
It Surely is entertaining.
Within the last couple of years Google has fired others who have raised concerns about AI ethics (though most of those concerns seemed to have centered around things like how skin tone impacts AI technology).
Either way - Google is botching their ethics game big time when it comes to AI. This doesn't feel acceptable. It doesn't make me feel confident that if AI were to become (or already is!) sentient under their watch that they would 1) Share it with the world. 2) Handle it in a way that feels truly ethical.
Watched the video and I really like the engineer. But I do question if its actually sentient. Weird how Google doesn't want to at-least test it
This comment was deleted 4 years ago