It’s likely no shock to individuals who grasp the implications of cutting-edge AI tools that the U.S. government is swiftly integrating them into military activities.

But what does this look like in practice?

A few days ago, I wrote about a presentation by Colonel Tucker Hamilton regarding the technological needs of warfighters, during which we touched on the milestone of an AI operating a military drone for the first time.

I particularly liked his term “battlefield of entities.”

Today, there’s more from an interview I conducted with Sean Batir, the CTO of Maven, a company aiding the military in deploying novel AI approaches.

Three Key Takeaways

During the interview, Batir outlined some of the principles experts are applying in these innovations.

One involves improving accuracy:

“Essentially, many of our models now are capable of meeting or surpassing human levels of detection, classification, and performance standards,” he stated.

Another relates to interconnectedness and system integration.

“No system, whether kinetic or non-kinetic, operates in isolation,” he noted. “In other words, different kinds of systems are usually horizontally integrated rather than deeply vertically integrated. Thus, enhancing the interoperability of interfaces to share data, models, targets, and detections has improved, since we now have more stable and standardized methods for passing this data between diverse modern systems and infrastructure.”

The third could be termed “l(fā)ocation visibility.”

According to Batir: “We're boosting the speed at which we can locate these kinds of dynamic and mobile objects on the battlefield. … What we're seeing currently is that as AI capabilities are integrated into various military processes, these kinds of triggers are being incorporated into how imagery and video are analyzed. Essentially, if you can't see it, you can't conduct an operation. And so we're observing that adversaries are now also distinguishing objects from non-objects, which is accelerating their capacity to execute conflict operations.”

Later, Batir discussed the cultural shift involved in transitioning from a place like MIT to the Pentagon, where a very different culture prevails.

Nevertheless, he indicated, some aspects carry over.

“The critical thinking you learn in undergrad, I believe, is highly transferable,” he said.

Culture of Communication

I asked Batir about his experiences in Pentagon meetings, his daily routine, and his work environment. He mentioned that phones are prohibited in meetings to encourage attentiveness.

“(Being forced to be fully present) probably has its benefits, considering that if you're making national-level decisions, you want everyone to be completely engaged and feel like they're with you throughout,” he explained.

Central Command

Here’s a recurring theme from the interview that aligns with something I’ve been hearing a lot about and that intuitively makes sense concerning technology.

In a way, it ties back to that term: “battlefield of entities” or more broadly, “IaaS?”

In a smart system, connecting components is crucial.

“The idea is that our data and detections are fed into other platforms,” Batir said, describing Maven solutions. “These other platforms aren’t always our own agencies. They belong to those other groups you just mentioned. The aim is that through our program, we’re building a sort of tapestry of connected sensors from all those branches of both the armed forces and intelligence agencies involved, into a unified AI network. Typically, many systems remain fragmented due to their histories and evolutions. Our objective is to bring these together.”

When asked whether the company focuses on computer vision or large language models (LLMs), Batir replied that Maven does both, adopting a “multi-model” approach.

Regarding risks:

On one hand, Batir pointed out, striving for perfection isn’t always beneficial.

“Our philosophy is that perfection is the enemy of progress,” he stated. “So long as we can ensure risks are clearly identified upfront and metrics are transparent, we don’t want to hinder people from utilizing capabilities that are available but not perfect.”

However, there are distinct standards for mission-critical systems.

“In real-world scenarios where lives are at stake, there’s a whole risk acceptance framework,” Batir added. “This is when people outside the pure tech domain get involved. We have things like disclosure officers. There’s legal review. There’s still a rigorous manual process, especially when discussing potential loss of life. A process exists to incorporate careful adjudication.”

I found all of this enlightening regarding what’s happening in the Department of Defense (DoD) as we benefit from AI advancements. Watch the video for audience questions about Ukraine and other concerns.

The above is the detailed content of The Military Is Doing New Things With AI. For more information, please follow other related articles on the PHP Chinese website!