Well, there's an argument that the Search usage pattern is incompatible with
next token predictors. E.g.
https://www.theguardian.com/commentisfree/2023/jun/03/lawyer-chatgpt-research-avianca-statement-ai-risk-openai-deepmind.
But maybe it depends on what one's searching for?
On 6/19/23 09:43,
glen wrote:
IDK. The implication that we already have laws that cover (80%?) of
the use cases for new tech we, as a society, want to discourage, is a
good default. It resists the "there ought to be a law" sensibility
held by old people and curmudgeons everywhere. And it keeps our legal
There probably already is a law, but no one knows what it is? The law
suffers from the same curse as the scientific literature, most of it gets
ignored because no one has the time to read it all.
So maybe that's what LLM's are for. We can set one to read the collected
works of Carl Friederich
glen wrote:
IDK. The implication that we already have laws that cover (80%?) of
the use cases for new tech we, as a society, want to discourage, is a
good default. It resists the "there ought to be a law" sensibility
held by old people and curmudgeons everywhere. And it keeps our legal
IDK. The implication that we already have laws that cover (80%?) of the use cases for new
tech we, as a society, want to discourage, is a good default. It resists the "there
ought to be a law" sensibility held by old people and curmudgeons everywhere. And it
keeps our legal system a little