TL;DR:
- Military tech leaders testified before the Senate Armed Services Committee, stating that calls for a “pause” on AI development are unenforceable and would likely benefit China.
- The witnesses emphasized the urgency for the US to act in the rapidly-evolving AI landscape and not fall behind.
- Shyam Sankar, CTO of Palantir, expressed disappointment with the military’s cautious approach to adopting new technology and suggested the DoD increase its investment in tech solutions.
- Josh Lospinoso, CEO of Shift5, highlighted the importance of using the data generated by military weapons systems to train AI systems and improve cybersecurity.
- The military’s competitive edge in AI may depend on securing data generated by US tech companies, and the experts recommended considering offensive actions such as trade restrictions or “data poisoning.”
Main AI News:
As the global race for AI dominance accelerates, military tech leaders and experts testified before the Senate Armed Services Committee this week, offering a clear message: calls for a “pause” on AI development are misguided and unenforceable. Representing military tech companies and defense contractor the Rand Corporation, these experts warned that China would likely take advantage of any pause to solidify its lead in the AI race, thereby setting international standards for AI use and development.
At this “inflection point” in AI’s evolution, the witnesses argued that it’s imperative for the US to act with urgency and not fall behind. “I think it would be very difficult to broker an international agreement to hit pause on AI development that would actually be verifiable,” stated Rand Corporation CEO Jason Matheny. Shyam Sankar, CTO of Palantir, concurred, adding that a pause in the US could have far-reaching implications, potentially enabling China’s recent regulations on AI content censorship to become the global norm. “A Democratic AI is crucial,” Sankar emphasized.
This testimony follows a recent open letter from hundreds of AI experts calling for a 6-month pause on training AI systems more powerful than OpenAI’s GPT-4. While the experts agreed that smart regulations are crucial for guiding AI’s development, they warned that a full-on pause would be detrimental to the Department of Defense, which has a history of struggling to keep pace with AI innovations.
Shyam Sankar, CTO of Palantir, expressed his disappointment with the military’s cautious approach to adopting new technology during the Senate hearing. He revealed that it was easier for his company to bring advanced AI tools to banking giant AIG than to the Army or Airforce. Sankar contrasted this sluggish adoption with Ukraine’s military, which he claimed was able to procure new software quickly to fight off Russian forces. Palantir CEO Alex Karp has previously stated that the company offered its services to the Ukrainian military.
Sankar emphasized the need for the Department of Defense to increase its investment in tech solutions, suggesting that 5% of its $768 billion budget should be allocated towards capabilities that will “terrify our adversaries.” Josh Lospinoso, Co-founder and CEO of Shift5 added that the military was missing out on valuable opportunities to use the data generated by its weapons systems.
Lospinoso argued that this data could be used to train powerful AI systems and improve the military’s cybersecurity defenses, but most of it currently “evaporates in the ether right away.” He concluded that the military’s weapons systems are “simply not AI-ready,” highlighting the need for the DoD to invest in this area.
The military’s competitive edge in AI may depend on securing data generated by US tech companies. Jason Matheny, President and CEO of the Rand Corporation, expressed concern about the potential risks of open-source AI, warning that the well-intentioned pursuit of free information could inadvertently aid military AI systems in other countries. Matheny suggested that AI tools above a certain threshold should be sold to foreign governments only with guardrails in place.
In some cases, the experts recommended that the US military consider taking offensive actions to limit the development of foreign military AI systems. These actions could include trade restrictions, sanctions on high-tech equipment, or “data poisoning.” This would involve intentionally manipulating or corrupting datasets used to train military AI models in an effort to buy the Pentagon more time to develop its own capabilities. Josh Lospinoso, CEO of Shift5, concurred with Matheny’s recommendations, underscoring the importance of maintaining the US military’s advantage in the rapidly-evolving AI landscape.
Conlcusion:
The testimony of military tech leaders and experts before the Senate Armed Services Committee this week underscores the importance of the US maintaining its competitive edge in the rapidly-evolving AI landscape. The witnesses emphasized that calls for a “pause” on AI development are misguided and unenforceable and warned that China would likely take advantage of any pause to solidify its lead in the AI race.
The experts also highlighted the need for the Department of Defense to increase its investment in tech solutions and secure data generated by US tech companies. In some cases, the experts recommended that the US military consider taking offensive actions to limit the development of foreign military AI systems. The testimony of these experts provides valuable insights into the current state of the AI landscape and the steps that need to be taken to ensure that the US remains at the forefront of this technology.
