3 ms·
But the fact remains - we are predictable. LLMs do nothing other than predict the next tokens. I recently met some candidates who are looking to join our team,
by pk-protect-ai 3y ago
But the fact remains - we are predictable. LLMs do nothing other than predict the next tokens. I recently met some candidates who are looking to join our team, and I found it really amusing how some of them were "hallucinating", not knowing the answer to the question, instead of admitting that they do not know the answer.
- dahart 3y agoOh totally there are lots of parallels, but we still tend to anthropomorphize AI and exaggerate its ability to think. In the case of humans, there are deeper reasons we don’t admit to not knowing, especially in a job interview, so we can’t conclude it’s the same thing as the hallucinations of a neural network, right?
- pk-protect-ai 3y agoOh here's another one for you. I recently got into an argument with Bing. This is where everyone should start laughing. How stupid must one be to argue with an LLM? Especially when one tries to prove a point to an LLM, knowing that the LLM does not understand and simply predicts the next token. Yet, the urge to explain and correct the falsehood, for the sake of future conversation, persists. It's so irrational and superficial, knowing that the LLM cannot comprehend you, but still, you hold out hope...