For anyone who thinks AI is just a big scam or whatever..not so fast.
Christmas is coming early this year
The Anduril and OpenAI strategic partnership will focus on improving the nation’s counter-unmanned aircraft systems (CUAS) and their ability to detect, assess and respond to potentially lethal aerial threats in real-time. As part of the new initiative, Anduril and OpenAI will explore how leading edge AI models can be leveraged to rapidly synthesize time-sensitive data, reduce the burden on human operators, and improve situational awareness. These models, which will be trained on Anduril’s industry-leading library of data on CUAS threats and operations, will help protect U.S. and allied military personnel and ensure mission success.
That is cool until the LLM start hallucinating crap and soldiers start dying. The reality is there is real AI in the military since the 90s or even before, TERCOM, DSMAC, automated remote sensing AI software, Automated Targeting and so on and works great, algorithms that do one thing and that thing really good. Like google Alphafold, it does one thing and do that thing really good, what this crypto bros stooges is AGI hype, multimodal models that do a lot of things mediocrely. One area that LLMs could help is probably in code generation to reduce costs, maybe document review and legal stuff, automated mailing and chatbots to spread fake news automated.For anyone who thinks AI is just a big scam or whatever..
First announcement: OpenAI collaborates with defence startup company Anduril
OpenAI's operator(computer use) tool can be used to ease the burden on humansAI has a lot of real world applications, but I'm not sure how much openAI chatbot is really going to help you defend against incoming missiles. But never stop the tech bros from overhyping their AI to get more funding.
Anduril is a real defence company. And they state as clearly as they can on how they are going to use OpenAI productsas I said in my tweet today, Sam Altman is a professional bullshitter with super human power of tricking people into overvaluing what he has to sell.
AI has a lot of real world applications, but I'm not sure how much openAI chatbot is really going to help you defend against incoming missiles. But never stop the tech bros from overhyping their AI to get more funding.
the partnership aims to improve the nation’s defense systems that protect U.S. and allied military personnel from attacks by unmanned drones and other aerial devices.
The Anduril and OpenAI strategic partnership will focus on improving the nation’s counter-unmanned aircraft systems (CUAS) and their ability to detect, assess and respond to potentially lethal aerial threats in real-time.
As part of the new initiative, Anduril and OpenAI will explore how leading edge AI models can be leveraged to rapidly synthesize time-sensitive data, reduce the burden on human operators, and improve situational awareness.
These models, which will be trained on Anduril’s industry-leading library of data on CUAS threats and operations, will help protect U.S. and allied military personnel and ensure mission success.
“Our partnership with OpenAI will allow us to utilize their world-class expertise in artificial intelligence to address urgent Air Defense capability gaps across the world. Together, we are committed to developing responsible solutions that enable military and intelligence operators to make faster, more accurate decisions in high-pressure situations.”
Yes, I know they are a real defense company and I know they are a hype machine from SV.Anduril is a real defence company. And they state as clearly as they can on how they are going to use OpenAI products
1- Neural network (in this case a "transformer") trained on 99.9% general mostly crappy data and 0.1% specific mission data vs a neural network (whatever the type is required) trained on 90% mission specific data or better yet a AI Silicon?OpenAI's operator(computer use) tool can be used to ease the burden on humans
This applies to all domains
Models are now trained on multi modal data. The model should be able to generalize if it is trained on enough data
We are in the very early stages. Let's see how it plays out.