Depends on specific cases, I have on good authority of how in few "bleeding edge" ones they essentially repacked/wrapped YOLOv3. Purpose was specifically tracking in adversarial conditions (smoke, including smokescreen, obstacles, etc)
For realtime on the edge the YOLO series is pretty good, I don't think anyone would disagree. Most of the really advanced stuff like Vision Language models all require a lot more compute and power budget.
I skimmed through the article. I didn’t understand what role AI supposedly plays in this case for tracking aid deliveries. For tracking you need sensors and connectivity from the mode of delivery, location information, some analytics and databases. What does this AI do for tracking? I can understand a sales pitch that says AI decides where to provide aid, how much, when, etc. But tracking deliveries? It’s a head scratcher for me.
The article just describes how they're ingesting the data, some Palantir rep is watching the deliveries remotely and what sounds like manually entering delivery data.
From there I'm guessing the "AI" part is an LLM interface is offered to ask questions about the deliveries?
This wouldn't be the first product where it just provides what a database already does just fine / more efficiently...
I'm not the expert and the details that the company would put out are obviously obtuse, but: image detection/identification, predictive policing, and planning. The latter sounds to me like they'd have some system where people enter reports in natural language and an LLM assembles the information together and then proposes some plan of action. As opposed to having to have a more structured data entry and the friction that comes with it. It's all in the article if you read between the lines, really.
As an American this is the first post I have ever feared replying to because the company discussed is both powerful in government and does active online monitoring and reporting to the government and could impact me forever. So that might in itself include an answer. I wanted to post and highlight direct quotes from this company's leadership but honestly think there is a risk of personal blowback in some way.
"Palantir's AI" is Anthropic Claude.
Depends on specific cases, I have on good authority of how in few "bleeding edge" ones they essentially repacked/wrapped YOLOv3. Purpose was specifically tracking in adversarial conditions (smoke, including smokescreen, obstacles, etc)
Why YOLOv3 and not something more recent? Like YOLO26 or something
Why YOLOv3 and not something more recent?
For realtime on the edge the YOLO series is pretty good, I don't think anyone would disagree. Most of the really advanced stuff like Vision Language models all require a lot more compute and power budget.
A bunch of what they were contracted for very much required realtime tracking at the edge
Dehomag.
What has been will be again, what has been done will be done again; there is nothing new under the sun.
I skimmed through the article. I didn’t understand what role AI supposedly plays in this case for tracking aid deliveries. For tracking you need sensors and connectivity from the mode of delivery, location information, some analytics and databases. What does this AI do for tracking? I can understand a sales pitch that says AI decides where to provide aid, how much, when, etc. But tracking deliveries? It’s a head scratcher for me.
The article just describes how they're ingesting the data, some Palantir rep is watching the deliveries remotely and what sounds like manually entering delivery data.
From there I'm guessing the "AI" part is an LLM interface is offered to ask questions about the deliveries?
This wouldn't be the first product where it just provides what a database already does just fine / more efficiently...
I'm not the expert and the details that the company would put out are obviously obtuse, but: image detection/identification, predictive policing, and planning. The latter sounds to me like they'd have some system where people enter reports in natural language and an LLM assembles the information together and then proposes some plan of action. As opposed to having to have a more structured data entry and the friction that comes with it. It's all in the article if you read between the lines, really.
why draw the line on palantir? why not involve microsoft amd intel who provides their computers, or car manufacturers that provides their vehicles?
been seeing lots of these attacks on defense companies without providing a better alternative and a concrete plan they can execute
As an American this is the first post I have ever feared replying to because the company discussed is both powerful in government and does active online monitoring and reporting to the government and could impact me forever. So that might in itself include an answer. I wanted to post and highlight direct quotes from this company's leadership but honestly think there is a risk of personal blowback in some way.