
Must be an official acknowledgment (e.g., statement, exploratory paper), not a random tweet by one employee
People are also trading
@UnspecifiedPerson Which specific statement in this blogpost or video makes you think that the company officially acknowledges that models could be conscious? (I haven't watched the full video)
In particular, it's worth distinguishing between "_It is unclear whether_ they could be conscious" and "They could be conscious". These two claims are subtly different.
Statements made in the blogpost such as 'There’s no scientific consensus on whether current or future AI systems could be conscious, or could have experiences that deserve consideration' suggests they were making the former claim but not the latter.
@LeeSharkey Those statements are already acknowledging the possibility, though. They're acknowledging the possibility while saying that there's no consensus on whether the possibility in fact holds, but that's still acknowledging the possibility.
Does this count? https://arxiv.org/abs/2402.12422
Release thread:
@GraceKind Thanks for bringing this to my attention! The paper notes:
The opinions expressed in this article are those of the author (at the time of writing). They do not necessarily reflect the views of his employers or the institutions to which he is affiliated.
I think it would have to be reposted from the Google DeepMind website or Twitter or otherwise endorsed to qualify.