LIM Center, Aleje Jerozolimskie 65/79, 00-697 Warsaw, Poland
+48 (22) 364 58 00

The Pentagon Urges AI Developers to Be More Transparent

The Pentagon Urges AI Developers to Be More Transparent

The Defense Department’s top artificial intelligence (AI) official, Craig Martell, has called for more transparency from AI developers. Martell believes that the department needs to know more about AI tools before fully committing to their use. He wants companies to share insights into how their AI software is built so that the department can feel comfortable and safe adopting it. Currently, AI software relies on large language models (LLMs), which use massive data sets to power tools such as chatbots and image generators. However, these services are often offered without revealing their inner workings, making it difficult for users to understand how the technology arrives at decisions or improves over time.

Martell expressed concern about the lack of information regarding the structure and data used in these models. He also highlighted the dangers posed by AI systems, which companies often fail to disclose. To address these issues, Martell is inviting industry and academics to a symposium in Washington in February. The symposium aims to determine the appropriate uses for LLMs within the Defense Department and develop benchmarks related to issues such as hallucination, bias, and danger.

Martell’s team has already identified 200 potential uses for LLMs within the department, and he emphasizes that they do not want to stop their use altogether but rather understand their benefits, dangers, and how to mitigate against them. The Pentagon has more than 800 AI projects underway, some of which involve weapons systems. Given the high stakes involved in these projects, the Pentagon will apply a higher bar in terms of how it uses algorithmic models compared to the private sector. Martell believes that allowing for hallucination or incorrect results is simply not acceptable, especially in critical situations where lives are on the line.

By seeking greater transparency and understanding of AI tools, the Defense Department aims to ensure responsible and effective implementation of AI technology within its operations.


– Bloomberg (no URL)