RebelOfBabylon on Nostr: Though I agree with you, I think we are simultaneously downplaying LLMs. Google's ...
Though I agree with you, I think we are simultaneously downplaying LLMs. Google's AlphaCode2 can apparently score in the 90th percentile on a software problem solving test, one that requires humans to think and strategize about a problem before solving it. And not because LLMs are intelligent but because they can attempt many different solutions incredibly fast and converge on the optimal one. Similarly, many AI researchers are confident that LLMs will likely postulate a new mathematical theorem in the coming years.
Published at
2023-12-17 16:19:18Event JSON
{
"id": "4acef12a58f8a4f51355bfb6f0bd1fa0ada4d980d7c05f6f9d2e5866f8618017",
"pubkey": "d06e6018c1fcf7d80d4f18ae7ea669fa10f84389f95f6d1bdcea9727cb266c33",
"created_at": 1702829958,
"kind": 1,
"tags": [
[
"e",
"b5108fa3fb2577d8e6d3c211e8f5dcb508d9caa91f129308879314f2894004e8",
"",
"root"
],
[
"p",
"f728d9e6e7048358e70930f5ca64b097770d989ccd86854fe618eda9c8a38106"
]
],
"content": "Though I agree with you, I think we are simultaneously downplaying LLMs. Google's AlphaCode2 can apparently score in the 90th percentile on a software problem solving test, one that requires humans to think and strategize about a problem before solving it. And not because LLMs are intelligent but because they can attempt many different solutions incredibly fast and converge on the optimal one. Similarly, many AI researchers are confident that LLMs will likely postulate a new mathematical theorem in the coming years.",
"sig": "38840bf221f76da4c7fe1f470f13065967aac7197d6711fde92112250b134d1f6813c0128f9c997f68389ef454ee3b08a257a02d3e138cccfb9b588b681450ab"
}