#hallucination — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #hallucination, aggregated by home.social.
-
11/
AI INTERPRETATION (read HALLUCINATION) of Adam Goodes: -
New arXiv policy: 1-year ban for hallucinated references
https://twitter.com/tdietterich/status/2055000956144935055
#HackerNews #arxiv #policy #hallucination #references #academic #integrity #research #ethics
-
New arXiv policy: 1-year ban for hallucinated references
https://twitter.com/tdietterich/status/2055000956144935055
#HackerNews #arxiv #policy #hallucination #references #academic #integrity #research #ethics
-
New arXiv policy: 1-year ban for hallucinated references
https://twitter.com/tdietterich/status/2055000956144935055
#HackerNews #arxiv #policy #hallucination #references #academic #integrity #research #ethics
-
New arXiv policy: 1-year ban for hallucinated references
https://twitter.com/tdietterich/status/2055000956144935055
#HackerNews #arxiv #policy #hallucination #references #academic #integrity #research #ethics
-
New arXiv policy: 1-year ban for hallucinated references
https://twitter.com/tdietterich/status/2055000956144935055
#HackerNews #arxiv #policy #hallucination #references #academic #integrity #research #ethics
-
The Irish Council for Civil Liberties (#ICCL) has released an open source tool designed to reduce LLM hallucinations.
If my 15-year-old PC had a GPU I'd be tempted to give it a try.
#verity #hallucination #ai #opensource #qwen #granite #ibm #gpu #ireland
-
The Irish Council for Civil Liberties (#ICCL) has released an open source tool designed to reduce LLM hallucinations.
If my 15-year-old PC had a GPU I'd be tempted to give it a try.
#verity #hallucination #ai #opensource #qwen #granite #ibm #gpu #ireland
-
The Irish Council for Civil Liberties (#ICCL) has released an open source tool designed to reduce LLM hallucinations.
If my 15-year-old PC had a GPU I'd be tempted to give it a try.
#verity #hallucination #ai #opensource #qwen #granite #ibm #gpu #ireland
-
The Irish Council for Civil Liberties (#ICCL) has released an open source tool designed to reduce LLM hallucinations.
If my 15-year-old PC had a GPU I'd be tempted to give it a try.
#verity #hallucination #ai #opensource #qwen #granite #ibm #gpu #ireland
-
The Irish Council for Civil Liberties (#ICCL) has released an open source tool designed to reduce LLM hallucinations.
If my 15-year-old PC had a GPU I'd be tempted to give it a try.
#verity #hallucination #ai #opensource #qwen #granite #ibm #gpu #ireland
-
What a wonderful discovery! #AI #hallucination #writingcommunity
https://halupedia.com/the-national-library-of-unfinished-books
-
What a wonderful discovery! #AI #hallucination #writingcommunity
https://halupedia.com/the-national-library-of-unfinished-books
-
What a wonderful discovery! #AI #hallucination #writingcommunity
https://halupedia.com/the-national-library-of-unfinished-books
-
What a wonderful discovery! #AI #hallucination #writingcommunity
https://halupedia.com/the-national-library-of-unfinished-books
-
What a wonderful discovery! #AI #hallucination #writingcommunity
https://halupedia.com/the-national-library-of-unfinished-books
-
John Scalzi says AI cannot be trusted.
#AI #hallucination
"As always, I remind myself: If it knows this little about something I know very well, think of how little it knows about things I know nothing about. It literally cannot be trusted with anything factual (because, one again, it doesn’t know facts, it just knows what is statistically likely to be the next word), and thinking that can be is an actual intellectual hazard and fault. Don’t be the one who does that." -
American astronomer and author Cliff Stoll takes to Hacker News to refute #AI generated report of his death…He said:
“AI hallucinations are getting ambitious. A couple people recently emailed, asking whether the Klein bottle business was still operating after my death.“Huh?” I thought. “I ain’t dead yet.”
“After some digging, I discovered the source: an AI-generated review of The Cuckoo’s Egg circulating on Facebook. Alongside the usual synthetic praise and fabricated details, it confidently announced that I had died in May 2024.
“Apparently AI has now advanced to the point where it can kill people off before they notice.
“Mark Twain once wrote, ’Reports of my death are greatly exaggerated.’ I never expected to field-test the quote personally.” #ArtificialIntelligence #hallucination
source: https://www.facebook.com/story.php?story_fbid=989939243570691&id=100076638743004 -
American astronomer and author Cliff Stoll takes to Hacker News to refute #AI generated report of his death…He said:
“AI hallucinations are getting ambitious. A couple people recently emailed, asking whether the Klein bottle business was still operating after my death.“Huh?” I thought. “I ain’t dead yet.”
“After some digging, I discovered the source: an AI-generated review of The Cuckoo’s Egg circulating on Facebook. Alongside the usual synthetic praise and fabricated details, it confidently announced that I had died in May 2024.
“Apparently AI has now advanced to the point where it can kill people off before they notice.
“Mark Twain once wrote, ’Reports of my death are greatly exaggerated.’ I never expected to field-test the quote personally.” #ArtificialIntelligence #hallucination
source: https://www.facebook.com/story.php?story_fbid=989939243570691&id=100076638743004 -
American astronomer and author Cliff Stoll takes to Hacker News to refute #AI generated report of his death…He said:
“AI hallucinations are getting ambitious. A couple people recently emailed, asking whether the Klein bottle business was still operating after my death.“Huh?” I thought. “I ain’t dead yet.”
“After some digging, I discovered the source: an AI-generated review of The Cuckoo’s Egg circulating on Facebook. Alongside the usual synthetic praise and fabricated details, it confidently announced that I had died in May 2024.
“Apparently AI has now advanced to the point where it can kill people off before they notice.
“Mark Twain once wrote, ’Reports of my death are greatly exaggerated.’ I never expected to field-test the quote personally.” #ArtificialIntelligence #hallucination
source: https://www.facebook.com/story.php?story_fbid=989939243570691&id=100076638743004 -
American astronomer and author Cliff Stoll takes to Hacker News to refute #AI generated report of his death…He said:
“AI hallucinations are getting ambitious. A couple people recently emailed, asking whether the Klein bottle business was still operating after my death.“Huh?” I thought. “I ain’t dead yet.”
“After some digging, I discovered the source: an AI-generated review of The Cuckoo’s Egg circulating on Facebook. Alongside the usual synthetic praise and fabricated details, it confidently announced that I had died in May 2024.
“Apparently AI has now advanced to the point where it can kill people off before they notice.
“Mark Twain once wrote, ’Reports of my death are greatly exaggerated.’ I never expected to field-test the quote personally.” #ArtificialIntelligence #hallucination
source: https://www.facebook.com/story.php?story_fbid=989939243570691&id=100076638743004 -
American astronomer and author Cliff Stoll takes to Hacker News to refute #AI generated report of his death…He said:
“AI hallucinations are getting ambitious. A couple people recently emailed, asking whether the Klein bottle business was still operating after my death.“Huh?” I thought. “I ain’t dead yet.”
“After some digging, I discovered the source: an AI-generated review of The Cuckoo’s Egg circulating on Facebook. Alongside the usual synthetic praise and fabricated details, it confidently announced that I had died in May 2024.
“Apparently AI has now advanced to the point where it can kill people off before they notice.
“Mark Twain once wrote, ’Reports of my death are greatly exaggerated.’ I never expected to field-test the quote personally.” #ArtificialIntelligence #hallucination
source: https://www.facebook.com/story.php?story_fbid=989939243570691&id=100076638743004 -
Embarrassing.
-
ICYMI: Italy forces DeepSeek, Mistral and Nova AI to warn users about hallucinations: AGCM closed three AI probes requiring DeepSeek, Mistral, and Nova AI to add permanent hallucination disclaimers on chat interfaces and registration screens in Italian. https://ppc.land/italy-forces-deepseek-mistral-and-nova-ai-to-warn-users-about-hallucinations/ #IntelligenzaArtificiale #Hallucination #Tecnologia #AI #Chatbot
-
ICYMI: Italy forces DeepSeek, Mistral and Nova AI to warn users about hallucinations: AGCM closed three AI probes requiring DeepSeek, Mistral, and Nova AI to add permanent hallucination disclaimers on chat interfaces and registration screens in Italian. https://ppc.land/italy-forces-deepseek-mistral-and-nova-ai-to-warn-users-about-hallucinations/ #IntelligenzaArtificiale #Hallucination #Tecnologia #AI #Chatbot
-
ICYMI: Italy forces DeepSeek, Mistral and Nova AI to warn users about hallucinations: AGCM closed three AI probes requiring DeepSeek, Mistral, and Nova AI to add permanent hallucination disclaimers on chat interfaces and registration screens in Italian. https://ppc.land/italy-forces-deepseek-mistral-and-nova-ai-to-warn-users-about-hallucinations/ #IntelligenzaArtificiale #Hallucination #Tecnologia #AI #Chatbot
-
ICYMI: Italy forces DeepSeek, Mistral and Nova AI to warn users about hallucinations: AGCM closed three AI probes requiring DeepSeek, Mistral, and Nova AI to add permanent hallucination disclaimers on chat interfaces and registration screens in Italian. https://ppc.land/italy-forces-deepseek-mistral-and-nova-ai-to-warn-users-about-hallucinations/ #IntelligenzaArtificiale #Hallucination #Tecnologia #AI #Chatbot
-
ICYMI: Italy forces DeepSeek, Mistral and Nova AI to warn users about hallucinations: AGCM closed three AI probes requiring DeepSeek, Mistral, and Nova AI to add permanent hallucination disclaimers on chat interfaces and registration screens in Italian. https://ppc.land/italy-forces-deepseek-mistral-and-nova-ai-to-warn-users-about-hallucinations/ #IntelligenzaArtificiale #Hallucination #Tecnologia #AI #Chatbot
-
My animal rights library hallucinated by an AI (Pro version).
Can you guess the actual titles and authors?
-
My animal rights library hallucinated by an AI (Pro version).
Can you guess the actual titles and authors?
-
My animal rights library hallucinated by an AI (Pro version).
Can you guess the actual titles and authors?
-
My animal rights library hallucinated by an AI (Pro version).
Can you guess the actual titles and authors?
-
My animal rights library hallucinated by an AI (Pro version).
Can you guess the actual titles and authors?
-
How do we make LLM output more trustworthy? A short survey note on three lines of recent work covering five papers: conformal-prediction coverage guarantees, behavioral calibration of the model's prose, and sample-disagreement detection. All three pay the same multi-sample inference tax; the choice is about what you want back.
https://benjaminhan.net/posts/20260505-llm-uncertainty-survey/?utm_source=mastodon&utm_medium=social
-
How do we make LLM output more trustworthy? A short survey note on three lines of recent work covering five papers: conformal-prediction coverage guarantees, behavioral calibration of the model's prose, and sample-disagreement detection. All three pay the same multi-sample inference tax; the choice is about what you want back.
https://benjaminhan.net/posts/20260505-llm-uncertainty-survey/?utm_source=mastodon&utm_medium=social
-
How do we make LLM output more trustworthy? A short survey note on three lines of recent work covering five papers: conformal-prediction coverage guarantees, behavioral calibration of the model's prose, and sample-disagreement detection. All three pay the same multi-sample inference tax; the choice is about what you want back.
https://benjaminhan.net/posts/20260505-llm-uncertainty-survey/?utm_source=mastodon&utm_medium=social
-
How do we make LLM output more trustworthy? A short survey note on three lines of recent work covering five papers: conformal-prediction coverage guarantees, behavioral calibration of the model's prose, and sample-disagreement detection. All three pay the same multi-sample inference tax; the choice is about what you want back.
https://benjaminhan.net/posts/20260505-llm-uncertainty-survey/?utm_source=mastodon&utm_medium=social
-
How do we make LLM output more trustworthy? A short survey note on three lines of recent work covering five papers: conformal-prediction coverage guarantees, behavioral calibration of the model's prose, and sample-disagreement detection. All three pay the same multi-sample inference tax; the choice is about what you want back.
https://benjaminhan.net/posts/20260505-llm-uncertainty-survey/?utm_source=mastodon&utm_medium=social
-
Semantic Entropy (Nature 2024) detects LLM confabulations by clustering sampled answers by meaning and computing entropy over the cluster distribution. "Paris" and "It's Paris" cluster together, so paraphrase noise doesn't inflate the signal. Cost: it only catches hallucinations that vary across samples. If the model is consistently wrong, all samples cluster and the detector says "confident".
https://benjaminhan.net/posts/20260505-semantic-entropy/?utm_source=mastodon&utm_medium=social
-
Semantic Entropy (Nature 2024) detects LLM confabulations by clustering sampled answers by meaning and computing entropy over the cluster distribution. "Paris" and "It's Paris" cluster together, so paraphrase noise doesn't inflate the signal. Cost: it only catches hallucinations that vary across samples. If the model is consistently wrong, all samples cluster and the detector says "confident".
https://benjaminhan.net/posts/20260505-semantic-entropy/?utm_source=mastodon&utm_medium=social
-
Semantic Entropy (Nature 2024) detects LLM confabulations by clustering sampled answers by meaning and computing entropy over the cluster distribution. "Paris" and "It's Paris" cluster together, so paraphrase noise doesn't inflate the signal. Cost: it only catches hallucinations that vary across samples. If the model is consistently wrong, all samples cluster and the detector says "confident".
https://benjaminhan.net/posts/20260505-semantic-entropy/?utm_source=mastodon&utm_medium=social
