Large Language Models Have Already Infiltrated Every Other Sector, Now They're Seeping Into the US Military

Artificial Intelligence has been a hot topic around the world ever since OpenAI released ChatGPT in late 2022.

After seeing the possibilities of AI, and especially after the v4 upgrade that came out soon after, the world became obsessed with the new technology.

The US Military is planning to give it a go as well, according to recent reports.

The large-language model, or LLM, has attracted the US Air Force colonel, Matthew Strohmeyer, who was interested in how an LLM might perform a military task.

After several prompts to the model, the colonel seemed rather impressed with its capabilities. He said,

“It was highly successful. It was very fast. We are learning that this is possible for us to do.”

LLMs are trained using massive quantities of internet data fed into the AI.

The AI then uses it for learning, predicting, and generating human-like responses to human prompts.

AI is Becoming a Part of a Daily Life

AI’s success in generating human-like responses made it extremely popular among numerous companies, many of which already adopted it for handling customer service.

The increase in interest and AI usage caused AI stocks to skyrocket, although this was concentrated in five or six leading companies. Even so, the Nasdaq 100 went up 33% YTD as a result.

The interest in AI is growing among regular users, as well. According to a survey conducted by the Pew Research Center, around 27% of Americans interact with AI on a daily basis.

In addition, another 28% of individuals at least encounter AI anywhere from several times per week to multiple times per day.

Following the rise of ChatGPT, multiple tech giants released their own AI products, including LLMs, but also AI that can generate images and even music.

For example, Google released Bard to match Microsoft, which acquired OpenAI and its ChatGPT. The company then released MusicLM, the AI model that generates music. Microsoft responded by developing an AI music generator, Muzic. Even Amazon joined the race with its AWS DeepComposer

Pentagon is Overseeing the AI Testing

The success and efficiency of LLMs have been impressive enough to even attract the interest of the US military.

Pentagon is Overseeing the AI Testing

Five such models were put through the paces as part of a broader series of Defense Department experiments. The experiments are aimed at developing data integration and digital platforms across the military.

Pentagon’s own digital and AI office, alongside the military top brass, has been running the program while certain US allies participated.

So far, the Pentagon was unwilling to reveal which LLM was used in testing. However, a San Francisco-based startup, Scale AI, revealed that its product, Donovan, was among the platforms chosen to participate in the testing.

The Benefits and Risks of Using AI in the Military

If the US military adopts an LLM solution, this would represent a major shift. At this time, very little involving the military is digitized and connected.

This has clear flaws, such as long waiting times for even basic requests for information and the potential for AI hallucinations.

Such requests often take anywhere from multiple hours to several days to be granted, as most information is gathered on phones.

According to the information that was shared, the unnamed AI solution that was tested was able to complete a request in 10 minutes.

Strohmeyer noted that this doesn’t make it ready for primetime right now, but the efficiency of AI is certainly showing great promise.

For now, the military exercises are still ongoing, and they will continue until July 26th.

Apart from showing how efficient AI can be for this purpose, they will also test whether military officials can use LLMs to generate new options that were previously not considered.

One example is using generative AI to plan the military response to an escalating global crisis. The prompts will start small and then shift into large regions, such as the Indo-Pacific region.

One big issue that was noticed is that generative AI can compound bias and relay incorrect information with great confidence.

It can still be hacked in multiple ways, such as by poisoning data feeds.

Due to these concerns, the military will have to double-check any results and calculations the product might provide, as even the AI cannot offer flawless results.

What's the Best Crypto to Buy Now?

  • B2C Listed the Top Rated Cryptocurrencies for 2023
  • Get Early Access to Presales & Private Sales
  • KYC Verified & Audited, Public Teams
  • Most Voted for Tokens on CoinSniper
  • Upcoming Listings on Exchanges, NFT Drops