The History Channel's latest 'The Proof is Out There' episode investigated the story of Bigfoot. More specifically the famous or infamous Bob Gimlin and Roger Patterson video from 1967 that claimed to have captured the creature in northern California. The program "enhanced" the video using AI and came up with details it said made the film more compelling and possibly "real".
The problem with this is the very use of AI to "enhance" the images may have actually altered the film to make it more believable. AI does not enhance a video or image as most people think. Instead, it adds or subtracts details depending on its training. So the detail of the muscle definition offered as evidence of genuineness may in fact be proof of the video's falseness. If the AI was trained on naked humans walking it would add the definition of musculature as a result. Likewise the "butt crack". As to the way the foot moved, if the AI was trained on apes and gorillas walking it would have interpreted that and added it to the details of its "enhancement".
The "evidence" that the fur moved naturally could also be the AI's interpretation of the way fur is supposed to move based on its training. Again AI does not bring out detail it adds it, subtracts it or interprets it based on its algorithm and training.
The most compelling evidence of the untrustworthiness of this process is the inability of the AI to fill in details of the face. While this was blamed on graininess, it was more likely that the AI was never trained on people walking in a gorilla suit wearing a mask. If it was trained on hominid faces it could have tried and failed to interpret the face as belonging to one of them and removing detail to make it fit.
Just like in Oz we are drawn to the display of power in AI while paying no attention to the machine behind the curtain. The truth is we are no further along in proving whether the video is real or fake than we were before. But now we have a better version made by AI like it does with Deep Fakes.
Dr Tim Lynch
Psychsoftpc
Dr. Tim Lynch, President of Psychsoftpc, received his Ph.D. in Psychology of Computers and Intelligent Machines from Boston University. Shortly thereafter, Omni Magazine named him the first Robopsychologist or Computer Psychologist. He was then written up as a computer psychologist, or psychologist who studies how computer interaction effects personality and how to make computer interfaces more user friendly, in the Wall Street Journal, Psychology Today, the New York Times, the Washington Post, the Atlanta Journal and Constitution, the London Sunday Times, Computer World, and many other publications. He was the first Psychologist to recognize and study tech addiction and the insulating effect of tech mediated communication like the social media we have today. As part of his Doctoral Dissertation on the Effects of Computer Use on Personality and Social Interaction Patterns, he created one of the first neural net based Artificial Intelligence Natural Language software programs which was subsequently the basis of programs used by NIH division of AIDS Research and the United Nations, among others. Dr Tim Lynch consulted on the look and feel of visual url link and image based web browsing leading to the development of the modern internet. Dr. Lynch was an editor for the first Journal of Psychology of Computers. He taught graduate level courses and has written numerous journal articles on Ethics in computer science, Psychology of computers and how interacting with computers and intelligent machines effects people. He consulted on technology in the classroom and helped develop the plan for the 21st Century Classroom for the US Department of Education and was a grants reader for DOE on technology implementation in the classroom. Dr Tim Lynch is the recipient of the Albert Nelson Marquis Lifetime Achievement Award by Marquis Who's Who and has been endorsed by Marquis Who's Who as a leader in Technology, Artificial Intelligence, Computer Psychology and High Performance Computing.
