Talking about large language models
arxiv.orgSaved by sari
Talking about large language models
LLMs and all such products tend to converge: on styles, on forms, on patterns; the “sameness” isn’t incidental, it’s actually core to what these technologies do, how they work; they will always have this quality; this quality is their modus operandi, really! There is no world in which LLMs write interesting things.
Humans care about what other human
... See moreBecause these vectors are built from the way humans use words, they end up reflecting many of the biases that are present in human language. For example, in some word vector models, doctor minus man plus woman yields nurse . Mitigating biases like this is an area of active research.
Timothy B Lee • Large language models, explained with a minimum of math and jargon
We’re far from replacing humans for many tasks now being delegated to large language models. That doesn’t mean LLMs can’t be helpful; it just means they’re being used wrong. The key is applying them to tasks they’re well-suited to, such as analyzing, synthesizing, and manipulating data, and not for making important decisions on behalf of people.
An ... See more
An ... See more
Reconsidering the Role of AI: Valuing Process Over Output
What’s ChatGPT really bad at? Being counterintuitive and surprising us. Large language models (LLMs) are trained on the mass output of humanity to optimize for what is expected. As a result, ChatGPT routinely let’s us down when it comes to getting something we didn’t expect. If you start paying attention to the music, books, movies, speeches, essay... See more