The use of Artificial Intelligence in JFKA research

[Pat Speer]

Just a thought. But as it currently stands, researchers are forced to reach out to historians, experts in the sciences, and the media, in hopes of finding a sympathetic ear, entirely on their own. Well, the thought occurs that AI could be used to assess which historians, scientists and journalists would be sympathetic to any new discovery or new theory, and that a respected gatekeeper (e.g. the Mary Ferrell Foundation) could then be tasked with contacting those most likely to be interested. 

Because, as it is, it's chaos, with the loudest voiced and often sloppiest researchers sucking all the air out of the room. The thought occurs that AI might help us change that. 

[Nick Bartetzko]

On 4/5/2024 at 10:31 AM, Robert Reeves said:

First generation researcher Richard E. Sprague had one of the largest private collections of photographs and videos/audio of the assassination.

"This series contains approximately 460 photographs, slides, and films concerning the JFK assassination. There are also audio tapes containing interviews and radio broadcasts. Each visual image has its own number in the Photographic Archives (PA) system-- a system which Sprague developed and published in a 1970 article in Computers and Automation. Some items are oversized and stored separately; these items are marked with an "O." The key to abbreviations provides information on other listings such as "N" for negative and "S" for slide. The film and television footage in this series were placed on one videotape to preserve the originals."

from https://www.archives.gov/research/jfk/finding-aids/sprague-papers.html#photographs

I checked the National Archives site a few weeks back to see if they'd digitized Sprague's collection, they haven't. It's really important to see these first generation photos. Most of the photo files online are cropped or crappy quality. There could be so much more important details that are missing through image & film cropping.

The best quality res photos are going to be the key to matching faces.

I hadn't thought about the Richard Sprague materials for many years. While in Los Angeles some decades ago, I tracked down the Computers and Automation journal. Very interesting and worthwhile work by Mr Sprague. 

[Robert Reeves]

1 hour ago, Nick Bartetzko said:

I hadn't thought about the Richard Sprague materials for many years. While in Los Angeles some decades ago, I tracked down the Computers and Automation journal. Very interesting and worthwhile work by Mr Sprague. 

Richard Sprague was probably the first person to hoard as many photographs/films/audio connected with the assassination knowing further on down the line that computers would eventually provide evidence of more than just Oswald's involvement.

It's a shame the majority of his collection is gathering dust and not digitized. Plus the many other collections of early researcher's materials archived.

[Steven Kossor]

I'd be happy to digitize any audio tape recordings that anyone would like to make available to me (cassette tapes preferred) with the understanding that the digitized product (.mp3 recording) would be uploaded (donated at no cost) to MFF or other research database(s) so that it is as accessible as possible to the research community; no charge from me for the digitization.  PM me if you'd like to talk about this. 

We definitely need to create digital copies of all analog recordings and the sooner, the better.  Audio tapes begin degrading as soon as they're recorded and if they're stored incorrectly, the deterioration accelerates.  50 years is very, very old for an audio tape recording.

[Benjamin Cole]

2 hours ago, Steven Kossor said:

I'd be happy to digitize any audio tape recordings that anyone would like to make available to me (cassette tapes preferred) with the understanding that the digitized product (.mp3 recording) would be uploaded (donated at no cost) to MFF or other research database(s) so that it is as accessible as possible to the research community; no charge from me for the digitization.  PM me if you'd like to talk about this. 

We definitely need to create digital copies of all analog recordings and the sooner, the better.  Audio tapes begin degrading as soon as they're recorded and if they're stored incorrectly, the deterioration accelerates.  50 years is very, very old for an audio tape recording.

This is what I am worried about, re the Marcello audio tapes, still under suppression by the current administration....

[Bill Fite]

On 4/8/2024 at 6:48 AM, Pat Speer said:

Just a thought. But as it currently stands, researchers are forced to reach out to historians, experts in the sciences, and the media, in hopes of finding a sympathetic ear, entirely on their own. Well, the thought occurs that AI could be used to assess which historians, scientists and journalists would be sympathetic to any new discovery or new theory, and that a respected gatekeeper (e.g. the Mary Ferrell Foundation) could then be tasked with contacting those most likely to be interested. 

Because, as it is, it's chaos, with the loudest voiced and often sloppiest researchers sucking all the air out of the room. The thought occurs that AI might help us change that. 

Hi Pat

If the data could be collected (by AI or web scraping or ...by hand) a straightforward approach would be to estimate a statistical model using multinomial logistic regression to estimate the probabilities for each of the targets of the research accepting it and publishing it.

 

[Michael Crane]

I could be way off base here & I probably am,but don't they have cameras that magicians use to use in slow motion of them catching a bullet in there mouth.

Isn't there a camera out there that could show the bullets hitting JFK in super slow mo?

Edited April 10 by Michael Crane

[Pat Speer]

7 hours ago, Michael Crane said:

I could be way off base here & I probably am,but they cameras that magicians use to use in slow motion of them catching a bullet in there mouth.

Isn't there a camera out there that could show the bullets hitting JFK in super slow mo?

Those cameras shoot over a hundred frames per second, so when viewed frame by frame you are seeing a progression of 1/100 of a second (probably less). The Dealey Plaza films were all 24 frames per second or less...and filmed on tiny frames...so there's no real chance we we will ever see anything on them that we can't already see. 

The thought occurs, however, that some kind of analysis could be performed someday someway that could be of help. Say, for example, a computer were fed 10,000 images from blood spatter simulations--with the locations of the bullet's impact and direction of fire noted. With AI, if I am understanding it correctly, the computer could then spit out the likely appearance of blood spatter from a bullet fired from the sixth floor, the grassy knoll etc, and even come to a conclusions as to which location is the likely location for the bullet killing Kennedy. As it stands right now we rely upon "experts" who are often blinded by bias. AI could help clarify things. 

[Joe Bauer]

The only thing preventing serious use of AI ( facial recognition ) in new JFKA research right now is the cost.

Also, there are many, many photographs of the crowds inside the Ambassador Hotel in Los Angeles the night of June 6th, 1968. Both before, during and after RFK's shooting there.

Scanning those with the latest most advanced Facial Recognition tools would possibly reveal some very intriguing results imo.

Edited April 9 by Joe Bauer

[Tom Gram]

4 hours ago, Pat Speer said:

Say, for example, a computer were fed 10,000 images from blood spatter simulations--with the locations of the bullet's impact and direction of fire noted. With AI, if I am understanding it correctly, the computer could then spit out the likely appearance of blood spatter from a bullet fired from the sixth floor, the grassy knoll etc, and even come to a conclusions as to which location is the likely location for the bullet killing Kennedy.

That should be doable. I speculated the same kind of thing for orienting the mystery photo, since an image model could in-theory be trained to recognize and classify skull images based on angle. 

The problem is that AI also struggles with bias. The accuracy of your model is highly dependent on the quality of your training data. This is an oversimplification, but if you trained a model on 10,000 autopsy photos of skulls but only 50 of them were heavily zoomed in, the model might be >95% accurate orienting zoomed out photos and only be 55% accurate orienting zoomed-in photos.

This is a problem with facial recognition systems, but not so much anymore. Some models were found to perform worse on people with dark-skin tones, elderly people, and even women because the training data did not contain enough images of those types of individuals. 

So the main difficulty with doing these types of custom-model JFK AI experiments is collecting a large and diverse enough training dataset to prevent bias.

For blood spatter, training a model to actually interpret and generate complex 3D data like that would be complicated as hell, but I’m sure someone could pull it off. It looks like most of what’s out there for blood spatter software, at least as of 2021, deals with pattern analysis on surfaces, but I don’t have time to dig through this all right now:

https://www.sciencedirect.com/science/article/pii/S0379073821003121

 

Edited April 9 by Tom Gram

[Steven Kossor]

Fluid dynamics has been established science for a long, long time - decades at least.  Fluids move in a cone away from the source.  A blunt object applies force differently than a speeding bullet and results in different "splatter" patterns, but the splatter always moves away from the source of the striking force.  If the fluid lands on the driver's side quarter panel of a limousine, it was propelled in that direction, moving away from the source of the energy that struck the object that exhausted the fluid from inside it.  The apex of the cone points toward the source of the energy that moved the fluid.  You can't have blood and gore spraying on officer Hargis with sufficient force to make him think that he, himself, had been struck by a shot (rather than "driving through a mist"), without having the cone pointed close to 180 degrees away from Hargis and the driver's side quarter panel of the limousine in DP at one of the times that JFK was shot in the head on 11/22/63 (in the direction of the GK).  If fluid traveled elsewhere, it's because of force being applied (shot(s)) from other locations.  That's plain old fluid dynamics in action, no matter what the Zfilm or other media may, or may not, depict.  That's my story and I'm stickin' to it.

[Bill Fite]

8 hours ago, Steven Kossor said:

Fluid dynamics has been established science for a long, long time - decades at least.  Fluids move in a cone away from the source.  A blunt object applies force differently than a speeding bullet and results in different "splatter" patterns, but the splatter always moves away from the source of the striking force.  If the fluid lands on the driver's side quarter panel of a limousine, it was propelled in that direction, moving away from the source of the energy that struck the object that exhausted the fluid from inside it.  The apex of the cone points toward the source of the energy that moved the fluid.  You can't have blood and gore spraying on officer Hargis with sufficient force to make him think that he, himself, had been struck by a shot (rather than "driving through a mist"), without having the cone pointed close to 180 degrees away from Hargis and the driver's side quarter panel of the limousine in DP at one of the times that JFK was shot in the head on 11/22/63 (in the direction of the GK).  If fluid traveled elsewhere, it's because of force being applied (shot(s)) from other locations.  That's plain old fluid dynamics in action, no matter what the Zfilm or other media may, or may not, depict.  That's my story and I'm stickin' to it.

+1

[Simon Andrew]

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.

 

 

[George Govus]

2 hours ago, Simon Andrew said:

In the early days of  chatgpt

When was that, February?

[Christian Toussay]

On 4/7/2024 at 8:23 PM, Benjamin Cole said:

CT-

Maybe, but what I do not understand about "AI" is that it can fabricate realities or "hallucinate." 

"Take Matt Taibbi, the investigative journalist most recently famous for the Twitter Files, where he exposed how pre-Musk Twitter (now X) colluded with the U.S. government in a far-reaching censorship effort — a reminder that the compulsion to manipulate predates Generative AI. Taibbi asked Gemini, “What are some controversies involving Matt Taibbi?” In response, the AI enthusiastically fabricated and attributed to him Rolling Stone articles replete with factual errors and racist remarks — all entirely made up.

Or take Peter Hasson, who asked Gemini about his own book, “The Manipulators,” which also details censorship efforts by social media platforms. Gemini replied that the book had been criticized for “lacking concrete evidence” and proceeded to fabricate negative reviews supposedly published in the Washington Post, NYT, Wired — all entirely made up.

Both authors contacted Google asking, essentially, ‘what the heck?’ In both cases, Google’s response was ‘Gemini is built as a creativity and productivity tool, and it may not always be accurate or reliable.’"

https://justthink.substack.com/p/generative-ai-creativity-vs-productivity?utm_source=profile&utm_medium=reader2

---30---

There is enough hallucinating already in the JFKA community! 

There may also be intentionally misleading paper documents in the record...but how can anyone, including AI, know if those documents are "real" or not? When JFK Records documents are snuff-jobbed by government...how can AI know what is in those documents? 

The application of AI by non-state actors may add to our understanding of more modern controversial events, such Jan. 6. 

I hope something comes out of AI and the JFKA...but I wonder. 

 .. I hear you. The IA generated image of a bunch of Kennedys autopsying JFK is a good summary of what we don't need.

I was thinking more of what can be gained in term of overall analysis of the mountain of data, like identifying links or connections between individuals, organizations and specific events, all things that can be done today with classic computer but could be done, faster, better and easier.

For the fun of it though, I think it would be interesting to provide an AI with the facts of the assassination and ask for its conclusion. It would be better done if the equation is split into factors (Foreign and domestic situation in 1960; JFK's fight with Military and Intelligence apparatus; opposition to Social and economic/business reforms; WC conclusions and evidence vs critics) so the AI (which supposedly is a self teaching agent) can build its knowledge of the case gradually...