at://jaystevens.me/app.bsky.feed.post/3m63n726grk2o
Back to Collection
Record JSON
{
"$type": "app.bsky.feed.post",
"createdAt": "2025-11-20T20:46:19.702Z",
"langs": [
"en"
],
"reply": {
"parent": {
"cid": "bafyreierqccg2wrnrtovuhq4j5bjfftsupy5lejadeqn32kppfkweybc2y",
"uri": "at://did:plc:jupvzthhxsac2mwtqwulknsk/app.bsky.feed.post/3m63lupyq2s2t"
},
"root": {
"cid": "bafyreierqccg2wrnrtovuhq4j5bjfftsupy5lejadeqn32kppfkweybc2y",
"uri": "at://did:plc:jupvzthhxsac2mwtqwulknsk/app.bsky.feed.post/3m63lupyq2s2t"
}
},
"text": "These all basically measure \"how often does it make shit up\" in various ways. The numbers reflect how often it was right, vs. how often it made something up that was wrong.\n\nFor example, Humanity's Last Exam is a bunch of doctorate-level questions with known answers. It was right 37.5% of the time"
}