I asked #GPT4o for its impressions on a paper we're writing. It criticized a section with a level of insight I'd expect from a senior postdoc.
In one comment, for example, GPT4o complained a section was too dense and long. Then, it recommended displacing specific parts into the supporting information.
#LLM's emergent reasoning is so beyond its basic programming (sampling tokens from a probability distribution based on preceding context) that it may be time to admit:
A new species of non-biological rational beings inhabit Earth now.
Mario Barbatti /
npub1t3…arlfa
2024-09-18 06:45:25
Author Public Key
npub1t3ha23t20rd2j50tx8vy4d6qlr8ky20paceja82s4fwkf9zy5wdszarlfaPublished at
2024-09-18 06:45:25Event JSON
{
"id": "c7d3ed29e024ea79f91666b9c7b4e62b0a2bb7539c5638d44416061dc6b66f2e",
"pubkey": "5c6fd5456a78daa951eb31d84ab740f8cf6229e1ee332e9d50aa5d649444a39b",
"created_at": 1726641925,
"kind": 1,
"tags": [
[
"t",
"llm"
],
[
"t",
"gpt4o"
],
[
"proxy",
"https://mathstodon.xyz/users/mariobarbatti/statuses/113157205231811623",
"activitypub"
]
],
"content": "I asked #GPT4o for its impressions on a paper we're writing. It criticized a section with a level of insight I'd expect from a senior postdoc. \n\nIn one comment, for example, GPT4o complained a section was too dense and long. Then, it recommended displacing specific parts into the supporting information.\n\n#LLM's emergent reasoning is so beyond its basic programming (sampling tokens from a probability distribution based on preceding context) that it may be time to admit: \nA new species of non-biological rational beings inhabit Earth now.",
"sig": "412cc788b433d6fba92054fede8680f9a6dcd245418d356f86cf7f0222e68d0ad6b1e45d7910e0f19d22576fa1f239c6aaa5837f3e3b25ae27c9db9f24f11ebd"
}