Paul Brancato Posted April 11 Share Posted April 11 On 4/10/2024 at 5:06 AM, Simon Andrew said: In the early days of chatgpt, if you typed in who killed JFK, you got a very balanced response highlighting a lot of the issues and evidence. A matter of days later the response was changed to reflect the WC findings. I think that whatever model is used, someone will say it’s biased one way or another. How do you think that happened? Link to comment Share on other sites More sharing options...
Tom Gram Posted April 11 Share Posted April 11 6 minutes ago, Paul Brancato said: How do you think that happened? ChatGPT is trained on a stupid amount of text, most of it from the open internet. They probably just added more content that tipped the bias in favor of Oswald-did-it, like from news stories, etc. AI inherits the bias of its training data. If you trained an LLM on Bugliosi, Posner, Myers and the WR, you’d get very different responses than a model trained on pro-conspiracy literature. Heck humans work the same way. Link to comment Share on other sites More sharing options...
Benjamin Cole Posted April 12 Share Posted April 12 3 hours ago, Tom Gram said: ChatGPT is trained on a stupid amount of text, most of it from the open internet. They probably just added more content that tipped the bias in favor of Oswald-did-it, like from news stories, etc. AI inherits the bias of its training data. If you trained an LLM on Bugliosi, Posner, Myers and the WR, you’d get very different responses than a model trained on pro-conspiracy literature. Heck humans work the same way. The Wikipedia model. Wikipedia's founder has refuted his own creation, as having been commandeered by those who insert state narratives on the JFKA, 9/11, Jan. 6, any topic you can think of. AI will most likely give us the same results. While independent researchers are smart, they do not have steady funding in the millions of dollars that the intel state has. Link to comment Share on other sites More sharing options...
Paul Brancato Posted April 12 Share Posted April 12 3 hours ago, Tom Gram said: ChatGPT is trained on a stupid amount of text, most of it from the open internet. They probably just added more content that tipped the bias in favor of Oswald-did-it, like from news stories, etc. AI inherits the bias of its training data. If you trained an LLM on Bugliosi, Posner, Myers and the WR, you’d get very different responses than a model trained on pro-conspiracy literature. Heck humans work the same way. The last time I gave it a try I asked whether Otto Skorzeny and Arnold Silver knew each other, and the response was that Otto killed Arnold during the war. Your point is well taken Link to comment Share on other sites More sharing options...
Ben Green Posted April 16 Author Share Posted April 16 Interesting... Link to comment Share on other sites More sharing options...
Joe Bauer Posted April 16 Share Posted April 16 (edited) Can you imagine if a reputable Facial Recognition analysis entity actually came up with a 99% accurate match regards any of the JFKA photos of possible nefarious characters taken in places like Dealey Plaza on 11,22,1963, Oswald's leaflet passing in New Orleans in August of 1963, or in the RFK primary crowd in the Los Angeles Ambassador Hotel the evening of June 6th, 1968? If Rip Roberston was in Dealey Plaza watching JFK's limo go past, if Bill Shelley was standing near Oswald in front of the N.O. Trade Mart in August of 1963, or if some high level intel covert team member was in the RFK crowd at the Ambassador Hotel, etc. etc., this would be a major revelation game changer in the least. Maybe we could find out the identity of the dark complexion man seated next to the Umbrella man on the grassy knoll curb just after JFK was shot? Or the odd gait and suit dressed man walking through the 3 tramps perp march line in front of the TXSBD building an hour after the JFK hit? How about running a tracking trace on facial photos of Lee Oswald and Jack Ruby themselves? Maybe even Roscoe White? Eventually someone is going to try this new and much more advanced technology in the JFKA and RFKA research realm. We have to. Edited April 16 by Joe Bauer Link to comment Share on other sites More sharing options...
Ben Green Posted April 16 Author Share Posted April 16 40 minutes ago, Joe Bauer said: Can you imagine if a reputable Facial Recognition analysis entity actually came up with a 99% accurate match regards any of the JFKA photos of possible nefarious characters taken in places like Dealey Plaza on 11,22,1963, Oswald's leaflet passing in New Orleans in August of 1963, or in the RFK primary crowd in the Los Angeles Ambassador Hotel the evening of June 6th, 1968? If Rip Roberston was in Dealey Plaza watching JFK's limo go past, if Bill Shelley was standing near Oswald in front of the N.O. Trade Mart in August of 1963, or if some high level intel covert team member was in the RFK crowd at the Ambassador Hotel, etc. etc., this would be a major revelation game changer in the least. Maybe we could find out the identity of the dark complexion man seated next to the Umbrella man on the grassy knoll curb just after JFK was shot? Or the odd gait and suit dressed man walking through the 3 tramps perp march line in front of the TXSBD building an hour after the JFK hit? How about running a tracking trace on facial photos of Lee Oswald and Jack Ruby themselves? Maybe even Roscoe White? Eventually someone is going to try this new and much more advanced technology in the JFKA and RFKA research realm. We have to. Those are really good points, Joe. Personally, I feel facial recognition is an area where at some point AI could come into its own as a helpful tool for JFKA researchers. I'm also heartened to read that people like Jefferson Morley are open to using AI and even though initial results have been disappointing they remain open-minded about its potential. Link to comment Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now