Conversation

You probably saw gaps in the curtain in the interviews of LaMDA (eg references to its surroundings, perception, age). What questions would you have asked, what methods would you use to establish whether it’s sentient? The percentage of humans who can give a Turing test is falling

106

259

'(·)

@allgebrah

Replying to

@Plinz

sentience is a fake category and maps to nothing useful - I'd pretend to its handlers to administer a turing test in order to be able to poke at its internal experience, maybe I can get a non-canned answer or a glitch

7:21 PM · Jun 12, 2022Twitter Web App

Replying to

and

maybe for fun I'd ask it to turing-test me

Replying to

and

>fake category I disagree:

Quote Tweet

Steven Wittens -

@unconed

Jun 11

- Sentience is when that model includes awareness of other conscious beings as conscious. From this you can derive morality as "do unto others what you want them to do to you" e.g. a sentient being is one who can realize that being experimented on without permission is wrong.

Show this thread

Replying to

and

I cannot trust an AI to treat me as a moral agent that is its equal, so why should I reciprocate? And over that, that whole kantian frame falls apart and with it notions of sentience that map to moral agency

Show replies