Unmasking AI: Pentagon Seeks Transparency From Top LLM Companies To Confront Ethical Challenges

The U.S. Defense Department is reportedly ramping up the pressure on AI firms, insisting they unveil the secrets shrouded in their enigmatic technologies. 

At the forefront of this bold move is Craig Martell, the Pentagon's chief digital and artificial intelligence officer, according to a report by Bloomberg. 

"We're just getting the end result of the model-building — that's not sufficient," Martell said in an interview. The Pentagon is reportedly in the dark and has no idea about these AI models' construction and data foundations.

Pentagon's Skepticism on AI 

Martell highlighted that companies also need to explain what dangers their systems could pose.

Precisely, the Pentagon is advocating for a balance where companies can disclose essential information about their AI software without compromising their intellectual property, the report added. 

AI's current "black box" nature, where outcomes are visible but the process remains hidden, is a significant concern for the Defense Department, Bloomberg noted.

AI Meeting By Pentagon

To address the concerns, the Chief Digital and Artificial Intelligence Office scheduled a symposium in Washington in February 2024, inviting AI industry experts and academics. 

The goal is to explore the potential applications and associated risks and threats of large language models (LLMs) in defense, aiming to establish a comprehensive understanding and set of guidelines for their use.

AI Titans: Leading LLM Companies Shaping the Future

The Defense Department is intrigued by the potential of LLMs, having identified over 200 possible applications within the sector, the note added.

The leading LLM companies that are in the headlines for actively developing AI strategies for their products are:  

Alphabet Inc. GOOG GOOGL: Google is currently working on integrating its Bard artificial intelligence chatbot with several of its popular products, including Gmail, YouTube and Google Docs. GOOG shares jumped 37.1% in the last year. 

Apple Inc. AAPL: The company has a slew of significant announcements lined up for 2024, including "Apple GPT." The tech behemoth plans to enter the generative AI race with "Apple GPT," and a new LLM. AAPL shares climbed more than 23% in a year's time.

Microsoft Corporation MSFT and Meta Platforms, Inc META expanded their AI partnership with Llama 2 on Azure and Windows in July. Llama 2 is a family of pre-trained and fine-tuned large language models (LLMs) ranging in scale from 7B to 70B parameters. Over the last year, shares of MSFT rose 35.6%, while shares of META nearly doubled. 

Also Read: Microsoft Considered Selling Bing To Apple As Google Rival, But iPhone Maker Remained Skeptical

Amazon.Com, Inc. AMZN: On Sept. 20, the company previewed the future of Alexa, powered by an LLM specifically optimized for voice interactions. The company is stepping into the generative AI arena with a new product called Bedrock, which offers its Amazon Web Services (AWS) clients the ability to use and adapt generative AI tools for their needs.

AI in Defense

Following a private artificial intelligence discussion at Capitol Hill recently, Alex Karp, CEO of Palantir Technologies Inc, openly expressed that tech firms should actively cooperate with the U.S. government in the realms of AI development and prospective regulations surrounding it. Earlier this week, Palantir Technologies was awarded a $250 million contract from the US Department of Defense. 

Notable pure-play military-technology firms that are actively leading the way in U.S. defense are Lockheed Martin Corporation LMT, Northrop Grumman Corporation NOC, RTX Corporation RTX, L3Harris Technologies, Inc. LHX and more. 

AI's Ethical Frontier

Martell's office is pivotal in guiding the Defense Department through the intricate landscape of over 800 AI projects, ensuring stringent safety and efficacy standards, the report said. 

The Pentagon's approach surpasses private sector norms, especially where lives are at stake, leaving no room for AI errors or "hallucinations," Martell told Bloomberg.

This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Photo: Shutterstock