@ralfmaximus<\/a> @glc<\/a> @paco<\/a> <\/p> If you didn't know the implementation of a specific model would there be any way to determine if it was capable of telling a lie?<\/p>","url":"https:\/\/techhub.social\/@ryanjyoder\/114925233013162262","status_net":{"notice_id":null}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10631461","source":"ActivityPub","repeat_of":"10626922"}},"published":"2025-07-28T06:17:27+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"title":"lxo repeated a notice by ryanjyoder","verb":"share","url":"https:\/\/snac.lx.oliva.nom.br\/lxo\/a\/c2b2104df493f85446b2e8dfd49253d9"},{"actor":{"id":"https:\/\/mastodon.online\/users\/glc","displayName":"GLC","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/288153-original-tmp20241019234211.webp","rel":"avatar","type":"image\/webp","width":350,"height":350},{"url":"https:\/\/gnusocial.jp\/avatar\/288153-96-20241204203123.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/288153-48-20241204203123.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/288153-24-20241204203123.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"288153"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/288153-96-20241204203123.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","summary":"I use CW mainly as a label as I tend to use the maximum character limit.Old man yelling at clouds. Clouds may need to mute me.","url":"https:\/\/mastodon.online\/@glc","portablecontacts_net":{"preferredUsername":"glc","displayName":"GLC","note":"I use CW mainly as a label as I tend to use the maximum character limit.Old man yelling at clouds. Clouds may need to mute me."}},"content":" @ryanjyoder<\/a> @paco<\/a> <\/p> Nope.<\/p> Just another illustration of the original point.<\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/mastodon.online\/users\/glc\/statuses\/114914889097962406","object":{"id":"https:\/\/mastodon.online\/users\/glc\/statuses\/114914889097962406","objectType":"note","content":" @ryanjyoder<\/a> @paco<\/a> <\/p> Nope.<\/p> Just another illustration of the original point.<\/p>","url":"https:\/\/mastodon.online\/@glc\/114914889097962406","status_net":{"notice_id":null},"inReplyTo":{"objectType":"note","id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114914614450553177","url":"https:\/\/techhub.social\/@ryanjyoder\/114914614450553177"}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/infosec.exchange\/users\/paco"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/techhub.social\/users\/ryanjyoder"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626919","source":"ActivityPub"}},"published":"2025-07-27T19:29:07+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/mastodon.online\/@glc\/114914889097962406"},{"actor":{"id":"https:\/\/techhub.social\/users\/ryanjyoder","displayName":"Ryan J. Yoder","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/254879-original-tmp20240410135421.webp","rel":"avatar","type":"image\/webp","width":400,"height":400},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-96-20250426221236.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-48-20250426221236.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-24-20250426221236.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"254879"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/254879-96-20250426221236.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","url":"https:\/\/techhub.social\/@ryanjyoder","portablecontacts_net":{"preferredUsername":"ryanjyoder","displayName":"Ryan J. Yoder"}},"content":" @paco<\/a> @paco<\/a> A friend sent me the story of the LLM deleting a database during a code freeze and said \"it lied when asked about it.\" I assert that a generative AI cannot lie. These aren't my original thoughts. But if you read Harry Frankfurt's famous essay On Bullshit (downloadable PDF here)<\/a>, he makes a very reasoned definition of bullshit. And this paragraph near the end of the essay explains why an LLM cannot lie.<\/p> It is impossible for someone to lie unless he thinks he knows the truth. Producing bullshit requires no such conviction. A person who lies is thereby responding to the truth, and he is to that extent respectful of it. When an honest man speaks, he says only what he believes to be true; and for the liar, it is correspondingly indispensable that he consider his statements to be false. For the bullshitter, however, all these bets are off: he is neither on the side of the true nor on the side of the false. His eye is not on the facts at all, as the eyes of the honest man and of the liar are, except insofar as they may be pertinent to his interest in getting away with what he says. He does not care whether the things he says describe reality correctly. He just picks them out, or makes them up, to suit his purpose.<\/p> And that's a generative artificial intelligence algorithm. Whether generating video, image, text, network traffic, whatever. It has no reference to the truth and is unaware of what truth is. It just says things. Sometimes they turn out to be true. Sometimes not. But that's irrelevant to an LLM. It doesn't know.<\/p> #bullshit<\/a> #ai<\/a> #llm<\/a> #genai<\/a><\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/infosec.exchange\/users\/paco\/statuses\/114914443003343367","object":{"id":"https:\/\/infosec.exchange\/users\/paco\/statuses\/114914443003343367","objectType":"note","content":" A friend sent me the story of the LLM deleting a database during a code freeze and said \"it lied when asked about it.\" I assert that a generative AI cannot lie. These aren't my original thoughts. But if you read Harry Frankfurt's famous essay On Bullshit (downloadable PDF here)<\/a>, he makes a very reasoned definition of bullshit. And this paragraph near the end of the essay explains why an LLM cannot lie.<\/p> It is impossible for someone to lie unless he thinks he knows the truth. Producing bullshit requires no such conviction. A person who lies is thereby responding to the truth, and he is to that extent respectful of it. When an honest man speaks, he says only what he believes to be true; and for the liar, it is correspondingly indispensable that he consider his statements to be false. For the bullshitter, however, all these bets are off: he is neither on the side of the true nor on the side of the false. His eye is not on the facts at all, as the eyes of the honest man and of the liar are, except insofar as they may be pertinent to his interest in getting away with what he says. He does not care whether the things he says describe reality correctly. He just picks them out, or makes them up, to suit his purpose.<\/p> And that's a generative artificial intelligence algorithm. Whether generating video, image, text, network traffic, whatever. It has no reference to the truth and is unaware of what truth is. It just says things. Sometimes they turn out to be true. Sometimes not. But that's irrelevant to an LLM. It doesn't know.<\/p> #bullshit<\/a> #ai<\/a> #llm<\/a> #genai<\/a><\/p>","url":"https:\/\/infosec.exchange\/@paco\/114914443003343367","status_net":{"notice_id":null},"tags":[{"objectType":"http:\/\/activityschema.org\/object\/hashtag","displayName":"ai"},{"objectType":"http:\/\/activityschema.org\/object\/hashtag","displayName":"genai"},{"objectType":"http:\/\/activityschema.org\/object\/hashtag","displayName":"llm"}]},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626917","source":"ActivityPub"}},"published":"2025-07-27T19:29:07+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/infosec.exchange\/@paco\/114914443003343367"},{"actor":{"id":"https:\/\/mastodon.social\/users\/ralfmaximus","displayName":"RalfMaximus","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/202290-original-tmp20231018161838.webp","rel":"avatar","type":"image\/webp","width":192,"height":192},{"url":"https:\/\/gnusocial.jp\/avatar\/202290-96-20240328163616.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/202290-48-20240328163616.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/202290-24-20240328163616.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"202290"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/202290-96-20240328163616.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","summary":"49% Evil Is Not Half BadAtlanta, Georgia, USAMulti-fandom nerd but lots of Star Trek.Multiple careers, but centered around software & technology & public healthcare especially EH.I write sometimes, mostly science fictiony things.","url":"https:\/\/mastodon.social\/@ralfmaximus","portablecontacts_net":{"preferredUsername":"ralfmaximus","displayName":"RalfMaximus","note":"49% Evil Is Not Half BadAtlanta, Georgia, USAMulti-fandom nerd but lots of Star Trek.Multiple careers, but centered around software & technology & public healthcare especially EH.I write sometimes, mostly science fictiony things."}},"content":" @ryanjyoder<\/a> @glc<\/a> @paco<\/a> <\/p> There are billions geography facts on the internet. This includes fictional data sets too from popular SF\/fantasy.<\/p> There are millions of human<->human interactions where geography facts are discussed, some of them accurately, many of them not.<\/p> Train a LLM on all that. Ask it geography questions.<\/p> The results will be the most likely next chain of text based on all those weighted interactions. That's it. There's no lying, no truth. It's making predictions.<\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/mastodon.social\/users\/ralfmaximus\/statuses\/114915602465634268","object":{"id":"https:\/\/mastodon.social\/users\/ralfmaximus\/statuses\/114915602465634268","objectType":"note","content":" @ryanjyoder<\/a> @glc<\/a> @paco<\/a> <\/p> There are billions geography facts on the internet. This includes fictional data sets too from popular SF\/fantasy.<\/p> There are millions of human<->human interactions where geography facts are discussed, some of them accurately, many of them not.<\/p> Train a LLM on all that. Ask it geography questions.<\/p> The results will be the most likely next chain of text based on all those weighted interactions. That's it. There's no lying, no truth. It's making predictions.<\/p>","url":"https:\/\/mastodon.social\/@ralfmaximus\/114915602465634268","status_net":{"notice_id":null},"inReplyTo":{"objectType":"note","id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114914997556741896","url":"https:\/\/techhub.social\/@ryanjyoder\/114914997556741896"}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/infosec.exchange\/users\/paco"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/techhub.social\/users\/ryanjyoder"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/mastodon.online\/users\/glc"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626921","source":"ActivityPub"}},"published":"2025-07-27T19:29:06+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/mastodon.social\/@ralfmaximus\/114915602465634268"},{"actor":{"id":"https:\/\/techhub.social\/users\/ryanjyoder","displayName":"Ryan J. Yoder","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/254879-original-tmp20240410135421.webp","rel":"avatar","type":"image\/webp","width":400,"height":400},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-96-20250426221236.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-48-20250426221236.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-24-20250426221236.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"254879"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/254879-96-20250426221236.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","url":"https:\/\/techhub.social\/@ryanjyoder","portablecontacts_net":{"preferredUsername":"ryanjyoder","displayName":"Ryan J. Yoder"}},"content":" @glc<\/a> @paco<\/a> @glc<\/a> @paco<\/a> @ralfmaximus<\/a> @glc<\/a> @paco<\/a> <\/p> If you didn't know the implementation of a specific model would there be any way to determine if it was capable of telling a lie?<\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114925233013162262","object":{"id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114925233013162262","objectType":"note","content":" @ralfmaximus<\/a> @glc<\/a> @paco<\/a> <\/p> If you didn't know the implementation of a specific model would there be any way to determine if it was capable of telling a lie?<\/p>","url":"https:\/\/techhub.social\/@ryanjyoder\/114925233013162262","status_net":{"notice_id":null},"inReplyTo":{"objectType":"note","id":"https:\/\/mastodon.social\/users\/ralfmaximus\/statuses\/114915602465634268","url":"https:\/\/mastodon.social\/@ralfmaximus\/114915602465634268"}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/infosec.exchange\/users\/paco"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/mastodon.social\/users\/ralfmaximus"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/mastodon.online\/users\/glc"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626922","source":"ActivityPub"}},"published":"2025-07-27T19:29:05+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/techhub.social\/@ryanjyoder\/114925233013162262"},{"actor":{"id":"https:\/\/mastodon.online\/users\/glc","displayName":"GLC","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/288153-original-tmp20241019234211.webp","rel":"avatar","type":"image\/webp","width":350,"height":350},{"url":"https:\/\/gnusocial.jp\/avatar\/288153-96-20241204203123.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/288153-48-20241204203123.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/288153-24-20241204203123.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"288153"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/288153-96-20241204203123.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","summary":"I use CW mainly as a label as I tend to use the maximum character limit.Old man yelling at clouds. Clouds may need to mute me.","url":"https:\/\/mastodon.online\/@glc","portablecontacts_net":{"preferredUsername":"glc","displayName":"GLC","note":"I use CW mainly as a label as I tend to use the maximum character limit.Old man yelling at clouds. Clouds may need to mute me."}},"content":" @paco<\/a> @ryanjyoder<\/a> @ralfmaximus<\/a> <\/p> A general comment ... from another angle, more or less parallel -<\/p> The question is whether the following is a picture of a woman or a picture of a tree. Opinions vary, and are strongly held.<\/p> LLMs by design strip away all meaning. Similar techniques have been in use since the I Ching (more recently, describedl by Shannon 1948 with transition probabilities). It's a bit like dehydrated water - just add water.<\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/mastodon.online\/users\/glc\/statuses\/114925955800237391","object":{"id":"https:\/\/mastodon.online\/users\/glc\/statuses\/114925955800237391","objectType":"note","content":" @paco<\/a> @ryanjyoder<\/a> @ralfmaximus<\/a> <\/p> A general comment ... from another angle, more or less parallel -<\/p> The question is whether the following is a picture of a woman or a picture of a tree. Opinions vary, and are strongly held.<\/p> LLMs by design strip away all meaning. Similar techniques have been in use since the I Ching (more recently, describedl by Shannon 1948 with transition probabilities). It's a bit like dehydrated water - just add water.<\/p>","url":"https:\/\/gnusocial.jp\/notice\/10626924","status_net":{"notice_id":null},"inReplyTo":{"objectType":"note","id":"https:\/\/infosec.exchange\/users\/paco\/statuses\/114925294903339830","url":"https:\/\/infosec.exchange\/@paco\/114925294903339830"}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/infosec.exchange\/users\/paco"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/mastodon.social\/users\/ralfmaximus"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/techhub.social\/users\/ryanjyoder"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626924","source":"ActivityPub"}},"published":"2025-07-27T19:29:03+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/gnusocial.jp\/notice\/10626924"},{"actor":{"id":"https:\/\/infosec.exchange\/users\/paco","displayName":"I Value the Goose","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/35095-original-tmp20231107034324.webp","rel":"avatar","type":"image\/webp","width":260,"height":260},{"url":"https:\/\/gnusocial.jp\/avatar\/35095-96-20231126220725.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/35095-48-20231126220725.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/35095-24-20231126220725.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"35095"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/35095-96-20231126220725.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","summary":"Amateur professional #selfhost sysadmin. Professional amateur #cloud #security at #AWS. Also fond of #cats, #cigars, #whiskey and #pipes. I like board games and some video games. I am #covid cautious and I still #wearamask. Opinions are my own, but they can be yours too. 100% Organic:,No artificial colors, preservatives, or intelligence added.","url":"https:\/\/infosec.exchange\/@paco","portablecontacts_net":{"preferredUsername":"paco","displayName":"I Value the Goose","note":"Amateur professional #selfhost sysadmin. Professional amateur #cloud #security at #AWS. Also fond of #cats, #cigars, #whiskey and #pipes. I like board games and some video games. I am #covid cautious and I still #wearamask. Opinions are my own, but they can be yours too. 100% Organic:,No artificial colors, preservatives, or intelligence added."}},"content":" @ryanjyoder<\/a> It doesn\u2019t store facts. It doesn\u2019t have a representation of \u201ctrue\u201d or \u201cfalse.\u201d It isn\u2019t a database. It splits written text into tokens and does colossally huge, environmentally damaging, and fabulously expensive \u201ctraining\u201d on that data using billions of parameters to arrive at a statistical model of tokens that follow other tokens. The model can then be queried to produce statistically likely replies to inputs.<\/p> Given an input like \u201ctell me a lie about the capital of France\u201d the most statistically improbable reply is \u201cthe capital of France is Paris.\u201d Other replies like \u201cwear a seatbelt\u201d are also super improbable. The size of these models and the probabilities they work with are really difficult to get one\u2019s head around. But it returned a statement that was a probabilistically likely reply to that input. That\u2019s all it did. <\/p> When models make up legal cases that don\u2019t exist, books that don\u2019t exist, programming APIs that don\u2019t exist, etc, they are simply outputting likely results. Text that fits the probability distribution of their input data. That\u2019s why it is not a \u201cbug\u201d when an LLM bullshits. It\u2019s not an error. It is working as designed.<\/p> There is nowhere to report to an LLM company the factually incorrect outputs its model produced because there is nothing they can do with that. It is working as designed.<\/p> @ralfmaximus<\/a> @glc<\/a><\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/infosec.exchange\/users\/paco\/statuses\/114925294903339830","object":{"id":"https:\/\/infosec.exchange\/users\/paco\/statuses\/114925294903339830","objectType":"note","content":" @ryanjyoder<\/a> It doesn\u2019t store facts. It doesn\u2019t have a representation of \u201ctrue\u201d or \u201cfalse.\u201d It isn\u2019t a database. It splits written text into tokens and does colossally huge, environmentally damaging, and fabulously expensive \u201ctraining\u201d on that data using billions of parameters to arrive at a statistical model of tokens that follow other tokens. The model can then be queried to produce statistically likely replies to inputs.<\/p> Given an input like \u201ctell me a lie about the capital of France\u201d the most statistically improbable reply is \u201cthe capital of France is Paris.\u201d Other replies like \u201cwear a seatbelt\u201d are also super improbable. The size of these models and the probabilities they work with are really difficult to get one\u2019s head around. But it returned a statement that was a probabilistically likely reply to that input. That\u2019s all it did. <\/p> When models make up legal cases that don\u2019t exist, books that don\u2019t exist, programming APIs that don\u2019t exist, etc, they are simply outputting likely results. Text that fits the probability distribution of their input data. That\u2019s why it is not a \u201cbug\u201d when an LLM bullshits. It\u2019s not an error. It is working as designed.<\/p> There is nowhere to report to an LLM company the factually incorrect outputs its model produced because there is nothing they can do with that. It is working as designed.<\/p> @ralfmaximus<\/a> @glc<\/a><\/p>","url":"https:\/\/infosec.exchange\/@paco\/114925294903339830","status_net":{"notice_id":null},"inReplyTo":{"objectType":"note","id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114925233013162262","url":"https:\/\/techhub.social\/@ryanjyoder\/114925233013162262"}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/mastodon.social\/users\/ralfmaximus"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/techhub.social\/users\/ryanjyoder"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/mastodon.online\/users\/glc"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626923","source":"ActivityPub"}},"published":"2025-07-27T19:29:03+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/infosec.exchange\/@paco\/114925294903339830"},{"actor":{"id":"https:\/\/hachyderm.io\/users\/inthehands","displayName":"Paul Cantrell","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/53259-original-tmp20231130153900.webp","rel":"avatar","type":"image\/webp","width":400,"height":400},{"url":"https:\/\/gnusocial.jp\/avatar\/53259-96-20231130154634.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/53259-48-20231130154634.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/53259-24-20231130154634.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"53259"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/53259-96-20231130154634.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","summary":"Composer, pianist, programmer, professor, rabble rouser, redheadComputer Science at https:\/\/www.macalester.edu\/mscs\/(Student projects: https:\/\/devgarden.macalester.edu)Artistic Director of https:\/\/newruckus.orgFreelance dev, often with https:\/\/bustout.comMusical troublemaker https:\/\/innig.net\/music\/The heart is the toughest part of the body.Tenderness is in the hands.\u00a0\u2014 Carolyn Forch\u00e9searchable","url":"https:\/\/hachyderm.io\/@inthehands","portablecontacts_net":{"preferredUsername":"inthehands","displayName":"Paul Cantrell","note":"Composer, pianist, programmer, professor, rabble rouser, redheadComputer Science at https:\/\/www.macalester.edu\/mscs\/(Student projects: https:\/\/devgarden.macalester.edu)Artistic Director of https:\/\/newruckus.orgFreelance dev, often with https:\/\/bustout.comMusical troublemaker https:\/\/innig.net\/music\/The heart is the toughest part of the body.Tenderness is in the hands.\u00a0\u2014 Carolyn Forch\u00e9searchable"}},"content":"
Maybe this is semantics but Meta AI appears to be able to lie.<\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114914614450553177","object":{"id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114914614450553177","objectType":"note","content":"
Maybe this is semantics but Meta AI appears to be able to lie.<\/p>","url":"https:\/\/techhub.social\/@ryanjyoder\/114914614450553177","status_net":{"notice_id":null},"inReplyTo":{"objectType":"note","id":"https:\/\/infosec.exchange\/users\/paco\/statuses\/114914443003343367","url":"https:\/\/infosec.exchange\/@paco\/114914443003343367"}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/infosec.exchange\/users\/paco"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626918","source":"ActivityPub"}},"published":"2025-07-27T19:29:07+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/techhub.social\/@ryanjyoder\/114914614450553177"},{"actor":{"id":"https:\/\/infosec.exchange\/users\/paco","displayName":"I Value the Goose","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/35095-original-tmp20231107034324.webp","rel":"avatar","type":"image\/webp","width":260,"height":260},{"url":"https:\/\/gnusocial.jp\/avatar\/35095-96-20231126220725.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/35095-48-20231126220725.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/35095-24-20231126220725.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"35095"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/35095-96-20231126220725.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","summary":"Amateur professional #selfhost sysadmin. Professional amateur #cloud #security at #AWS. Also fond of #cats, #cigars, #whiskey and #pipes. I like board games and some video games. I am #covid cautious and I still #wearamask. Opinions are my own, but they can be yours too. 100% Organic:,No artificial colors, preservatives, or intelligence added.","url":"https:\/\/infosec.exchange\/@paco","portablecontacts_net":{"preferredUsername":"paco","displayName":"I Value the Goose","note":"Amateur professional #selfhost sysadmin. Professional amateur #cloud #security at #AWS. Also fond of #cats, #cigars, #whiskey and #pipes. I like board games and some video games. I am #covid cautious and I still #wearamask. Opinions are my own, but they can be yours too. 100% Organic:,No artificial colors, preservatives, or intelligence added."}},"content":"
If an llm can't lie, then it seems true by only a technical definition. I can ask Meta ai basic questions about geography and get correct answers reliably. And I can ask it to lie about basic geography and I can reliably get incorrect answers.<\/p>","generator":{"id":"tag:gnusocial.jp,2026-07-07:notice-source:ActivityPub","objectType":"application","status_net":{"source_code":"ActivityPub"}},"id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114914997556741896","object":{"id":"https:\/\/techhub.social\/users\/ryanjyoder\/statuses\/114914997556741896","objectType":"note","content":"
If an llm can't lie, then it seems true by only a technical definition. I can ask Meta ai basic questions about geography and get correct answers reliably. And I can ask it to lie about basic geography and I can reliably get incorrect answers.<\/p>","url":"https:\/\/techhub.social\/@ryanjyoder\/114914997556741896","status_net":{"notice_id":null},"inReplyTo":{"objectType":"note","id":"https:\/\/mastodon.online\/users\/glc\/statuses\/114914889097962406","url":"https:\/\/mastodon.online\/@glc\/114914889097962406"}},"to":[{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/infosec.exchange\/users\/paco"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/person","id":"https:\/\/mastodon.online\/users\/glc"},{"objectType":"http:\/\/activitystrea.ms\/schema\/1.0\/collection","id":"http:\/\/activityschema.org\/collection\/public"}],"status_net":{"conversation":"tag:gnusocial.jp,2025-07-27:objectType=thread:nonce=dd11fc5722b2dea4","notice_info":{"local_id":"10626920","source":"ActivityPub"}},"published":"2025-07-27T19:29:06+00:00","provider":{"objectType":"service","displayName":"GNU social JP","url":"https:\/\/gnusocial.jp\/"},"verb":"post","url":"https:\/\/techhub.social\/@ryanjyoder\/114914997556741896"},{"actor":{"id":"https:\/\/techhub.social\/users\/ryanjyoder","displayName":"Ryan J. Yoder","status_net":{"avatarLinks":[{"url":"https:\/\/gnusocial.jp\/avatar\/254879-original-tmp20240410135421.webp","rel":"avatar","type":"image\/webp","width":400,"height":400},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-96-20250426221236.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-48-20250426221236.webp","rel":"avatar","type":"image\/webp","width":48,"height":48},{"url":"https:\/\/gnusocial.jp\/avatar\/254879-24-20250426221236.webp","rel":"avatar","type":"image\/webp","width":24,"height":24}],"profile_info":{"local_id":"254879"}},"image":{"url":"https:\/\/gnusocial.jp\/avatar\/254879-96-20250426221236.webp","rel":"avatar","type":"image\/webp","width":96,"height":96},"objectType":"person","url":"https:\/\/techhub.social\/@ryanjyoder","portablecontacts_net":{"preferredUsername":"ryanjyoder","displayName":"Ryan J. Yoder"}},"content":"
All models work fundamentally in the same way: predicting a series of output tokens based on a series of input tokens. If you don\u2019t understand this basic implementation mechanism, the rest of this conversation is inaccessible.<\/p>
All models work fundamentally in the same way: predicting a series of output tokens based on a series of input tokens. If you don\u2019t understand this basic implementation mechanism, the rest of this conversation is inaccessible.<\/p>