Embed Notice

HTML Code

<blockquote style="position: relative; padding-left: 55px;"><section><a href="https://hachyderm.io/users/deirdresm/statuses/112256279951705295">Deirdre Saoirse Moen (deirdresm@hachyderm.io)'s status on Friday, 12-Apr-2024 15:14:07 JST</a><a href="https://hachyderm.io/@deirdresm" title="deirdresm@hachyderm.io"><img src="https://gnusocial.jp/avatar/41625-48-20221206182946.webp" width="48" height="48" alt="Deirdre Saoirse Moen" style="position: absolute; left: 0; top: 0;">Deirdre Saoirse Moen</a><div><ul><li></ul></div></section><article><p>This is the single best explanation (long!) I've read about why LLMs are a con. Great piece from <a href="https://toot.cafe/@baldur">@baldur</a>.</p><p><a href="https://softwarecrisis.dev/letters/llmentalist/" rel="nofollow noreferrer">https://softwarecrisis.dev/letters/llmentalist/</a></p><p><a href="https://hachyderm.io/tags/AI" rel="tag">#AI</a> <a href="https://hachyderm.io/tags/LLM" rel="tag">#LLM</a></p></article><footer><a rel="bookmark" href="https://gnusocial.jp/conversation/2927965#notice-5825360">In conversation</a><time datetime="2024-04-12T15:14:07+09:00" title="Friday, 12-Apr-2024 15:14:07 JST">Friday, 12-Apr-2024 15:14:07 JST</time> <span>from <span><a href="https://hachyderm.io/@deirdresm/112256279951705295" rel="external" title="Sent from hachyderm.io via ActivityPub">hachyderm.io</a></span></span><a href="https://hachyderm.io/@deirdresm/112256279951705295">permalink</a><h4>Attachments</h4><ol><li><label><a rel="external" href="https://gnusocial.jp/attachment/2491198">The intelligence illusion seems to be based on the same mechanism as that of a psychic’s con, often called cold reading. It looks like an accidental automation of the same basic tactic.By using validation statements, such as sentences that use the Forer effect, the chatbot and the psychic both give the impression of be- ing able to make extremely specific answers, but those answers are in fact statistically generic.The psychic uses these statements to give the impression of being able to read minds and hear the secrets of the dead.The chatbot gives the impression of an intelligence that is specifi- cally engaging with you and your work, but that impression is nothing more than a statistical trick.</a></label><br><a href="https://media.hachyderm.io/media_attachments/files/112/256/268/622/790/724/original/007bdd94d1ead765.png" rel="external">https://media.hachyderm.io/media_attachments/files/112/256/268/622/790/724/original/007bdd94d1ead765.png</a></li></ol></footer></blockquote>

Corresponding Notice

Embed this notice
Deirdre Saoirse Moen (deirdresm@hachyderm.io)'s status on Friday, 12-Apr-2024 15:14:07 JSTDeirdre Saoirse Moen
- Baldur Bjarnason
This is the single best explanation (long!) I've read about why LLMs are a con. Great piece from @baldur.
https://softwarecrisis.dev/letters/llmentalist/
#AI #LLM
In conversationFriday, 12-Apr-2024 15:14:07 JST from hachyderm.iopermalink
Attachments
1. The intelligence illusion seems to be based on the same mechanism as that of a psychic’s con, often called cold reading. It looks like an accidental automation of the same basic tactic. By using validation statements, such as sentences that use the Forer effect, the chatbot and the psychic both give the impression of be- ing able to make extremely specific answers, but those answers are in fact statistically generic. The psychic uses these statements to give the impression of being able to read minds and hear the secrets of the dead. The chatbot gives the impression of an intelligence that is specifi- cally engaging with you and your work, but that impression is nothing more than a statistical trick.
  https://media.hachyderm.io/media_attachments/files/112/256/268/622/790/724/original/007bdd94d1ead765.png

Public

Embed Notice

HTML Code

Corresponding Notice