Ketan Joshi on Nostr: This is such a good illustration of the danger of over-reliance on machine-generated ...
This is such a good illustration of the danger of over-reliance on machine-generated "research" tools - sometimes it reproduces human-created accurate information well, and other times it garbles it but in a seriously convincing way.
So the false confidence effect is extremely powerful here
RE:
https://bsky.app/profile/did:plc:mdjhvva6vlrswsj26cftjttd/post/3ljivpiwags2a ChatGPT recently released a new mode, Deep Research, and @caseynewton.bsky.social tested it out with a prompt to write a 5k word report on the fediverse.
The best way to judge the quality of Deep Research is to do deep research on the output, so here you go
fediversereport.com/fediverse-re...
Fediverse Report’s deep resear...
Published at
2025-03-04 14:28:35Event JSON
{
"id": "d45b68ae9497024f912bed74d064f2e63120e37229061bd28a57bb9419bfd418",
"pubkey": "cbcf0f555e3cd259d33c093054a665167cf1b6688dbe38df58118920981716c7",
"created_at": 1741098515,
"kind": 1,
"tags": [
[
"q",
"97472817de8163d1c44258f7eaedd0f9a3ed3720d1170ecbd66d650a57442c0f",
"wss://relay.mostr.pub"
],
[
"proxy",
"https://bsky.brid.gy/convert/ap/at://did:plc:66lbtw2porscqpmair6mir37/app.bsky.feed.post/3ljkofcxcgk2p",
"activitypub"
]
],
"content": "This is such a good illustration of the danger of over-reliance on machine-generated \"research\" tools - sometimes it reproduces human-created accurate information well, and other times it garbles it but in a seriously convincing way. \n\nSo the false confidence effect is extremely powerful here\n\nRE: https://bsky.app/profile/did:plc:mdjhvva6vlrswsj26cftjttd/post/3ljivpiwags2a\n\nnostr:nevent1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqjarjs977s93ar3zztrm74mwslx376deq6ytsaj7kd4js546y9s8sp5nsm9",
"sig": "b91806c306c8e54d1e3ae804a393448ab563341b81146c92f32cb6a665eb11e34ffa97bd8876d305423273a497fa01f4d59802ab03ef90efe932e8269c2c5cc9"
}