US Department of Defense wants to rely on artificial intelligence to draw up military plans: what experts fear

American researchers believe it is necessary to master artificial intelligence tools before China or any other country, but not allow it to be weaponized.

Military experts from SCSP (US Defense Space and Science Program) believe that the level of sophistication of artificial intelligence allows creating military plans, but the main decision must always be human. The Breaking Defense portal writes about this.

SCSP argues that the Department of Defense should absolutely explore AI tools so that China or another unscrupulous rival does not get ahead of them; However, as noted in the recently published Science and Space Defense study, very careful groundwork will be required in practice. Attempt.

Yulber Bayraktari, a Defense Ministry veteran who now works as a senior advisor at SCSP, warned that in any case, there must be at least one well-trained person to check the military plan and AI analytics before implementation. , stop linking AI directly to a swarm of deadly drones, for example.

“Certain military applications require much better customization of the LLM (Large Language Model). It needs to be built specifically for military purposes, and just using existing ChatGPT is not enough,” said Justin Lynch, a former Army officer. SCSP’s defense research.

If generative AI can develop a complete military operational plan, could it also execute that plan, perhaps by relaying orders directly to a swarm of fighter jets? Lynch said it was technologically possible, but it was a terrible Terminator movie idea and not what SCSP suggested.

In short, SCSP experts envision the use of generative AI in military and intelligence at three levels.

  • Content production: For military purposes, only chatbots specifically designed for tasks such as summarizing classified intelligence reports or inventorying military logistics should be used. The more accurate and specific the datasets on which the AI ​​is trained, the more accurate and actionable its answers will be.
  • Automated curation: AI can gather information from different databases rather than using only a single data set, no matter how large. A large language model can automatically retrieve the most appropriate datasets and analysis tools for you.
  • Agent AI: This is where AI can move from recommending tasks to executing them, albeit under tight control. SCSP experts suggest that generative AI will task AI subordinates with gathering information or organizing supply convoys to fill gaps in data, for example, rather than launching autonomous death swarms.

Higher levels in this framework rely on generative AI’s ability to understand wide-ranging requests from “planning team meals for next week” to “creating a battalion logistics plan” in plain English and turning it into a detailed checklist of specific tasks.

In fact, the SCSP report suggests that between the emergence of artificial intelligence and the rise of China, humanity faces worrying “similarities to the period before the First World War,” with technological change and geopolitical instability. That’s not exactly reassuring, military officials say, but with some expert advice (from humans and AI) perhaps the world will be able to deal with the problem better this time.

Focus previously reported that the Chinese military wants to be the first army to use artificial intelligence tools in combat. That’s what will stop them.

Source: Focus

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Latest

Video: Man kills thief who exposed his family, perpetrator’s mother demands ‘justice’

Ney Chavez Santillanrestaurant owner (chifa) Villa El SalvadorV Lime, Peru, killed Gianfranco Espinoza Quispe, an alleged thief who entered his business and subjugated his...