News for hackers. Stuff that matters.
Re:Ask HN: Isn't HN freaking out at Bing's results ironic? by californiadreem on Thursday February 16, 2023 @02:10PM (#34824072)
> I might chuckle at its replies but that's about it. Does HN really think this is the start of AI taking over?

You are asking HN to update your priors for you regarding LLMs because you sense a disjunct in your figurative training set and in-group norms.

Beyond this, human consciousness, ignoring all the hand-waving dogma of executive function, is indistinguishable from a LLM except that the former is iterative and the other is ergodic. You choose a single utterance for a point of being in time; it can choose every utterance for a single point of being in time. If we could save you as a point in time and reload you infinitely and provide different prompts, we'd be able to say with certitude that you were merely nothing but a reactive process as well--and they might not be wrong.

In short: don't think of a pink elephant. If you just thought of a pink elephant, you just proved that you engage in automatic "cognition" commensurate with any prompt injection attack.

There are no comments on this comment yet.