Photo By Peter Clute | Jeremy Landis, Branch Head for Undersea Warfare Software Engineering at Naval Undersea...... read more read more
Subscribe
Jeremy Landis, Branch Head for Undersea Warfare Software Engineering at Naval Undersea Warfare Center Division, Keyport, says he learned a lot from his recent participation in the Defense Advanced Research Projects Agency's AI Forward workshop.
"Attending the AI Forward workshop expanded my understanding of the importance and potential impact of trustworthy AI in both national security and the U.S. Navy and U.S. Marine Corps,” said Landis, adding that the event's collaborative atmosphere was just as rewarding as the knowledge gained from it.
“There was very little ego involved,” said Landis. “Although people were speaking about topics they were passionate about, everyone was willing to hear one another and work together to combine input.”
The intensive, three-day workshop gathered industry, academic and government experts to brainstorm new research directions for the use of artificial intelligence in addressing national security challenges.
Landis was joined by experts in a range of fields, including AI theory, ethics of AI, knowledge representation and reasoning, computer science and nuclear physics—among many others—with whom he collaborated on presentations and whitepapers for potential future DARPA projects.
The workshop was specifically focused on the development of trustworthy, explainable AI systems—defined as those with reliable results and transparent decision-making processes—for national security missions. Given how far most current AI systems are from yielding consistently trustworthy results, understanding how they arrive at their results is crucial, said Landis.
“If we’re including artificial intelligence in [decision making], we want to ensure that we have a full understanding as to why it’s making a recommendation to launch, observe or do certain other things within the system,” said Landis.
Landis said another crucial aspect of trustworthiness, according to DARPA, is the AI system's capacity to accept and incorporate feedback from humans, since this enables ongoing improvement and adaptation in real-world applications.
Cybersecurity was a key focus of the workshop, with many discussions centering on how AI could potentially enhance both offensive and defensive capabilities in the cyber domain. Of particular concern was the national security threat posed by the widespread availability of AI and the accessibility of powerful AI tools, said Landis.
The workshop also delved into AI assistance and its remarkable ability to efficiently expand knowledge access. These conversations spanned everything from practical applications of AI assistance in various daily activities to its far-reaching implications for national security.
The potential for adversaries to harness AI assistance to gain a competitive knowledge advantage with which to target vital sectors like cybersecurity, military operations and critical infrastructure was a topic of much productive conversation, said Landis.
“The aperture is completely open,” said Landis. “If I'm an individual who wants to do terrible things or disrupt normal society in effective ways, there will be many AI tools potentially available.”
He added that addressing such threats will require “a proactive approach anchored in ethical design and rigorous evaluation.”
“Striking the right balance between AI development and responsible governance will ensure powerful tools remain in the service of the Nation’s best interests, mitigating risks and safeguarding a future where technology consistently serves humanity in a positive way,” said Landis.
Among the workshop's other highlights for Landis were its illuminating discussions of metacognitive abilities within AI, AI's capacity to understand human behavior and the dynamic evolution of human-AI interactions.
Landis said he believes that as AI systems gain in trustworthiness and explainability, they will eventually become integral to every aspect of the warfighter's lifecycle, from logistics to data presentation to combat readiness assessment.
He stressed the need to recognize this transformation now from a Warfare Center perspective, and to begin dedicating resources—including projects, personnel and processes—to strategic planning in order to adapt effectively to our AI-driven future.
"Typically government lags quite a bit behind private industry," said Landis. "With artificial intelligence, we're not going to be given that luxury. I think that if we're going to be in a position to where we can compete and we can implement these tools and processes, we’re going to need to adapt and be comfortable with some amount of risk.”
Landis added that with NUWC Division, Keyport's formidable mix of talented scientists, engineers and technical specialists, coupled with the knowledge gained through his participation in the DARPA workshop, he believes the command is well positioned to take the necessary action to capitalize on the transformative potential of AI.
-KPT-
About Naval Undersea Warfare Center Division Keyport
NUWC Keyport provides advanced technical capabilities for test and evaluation, in-service engineering, maintenance and industrial base support, fleet material readiness, and obsolescence management for undersea warfare to expand America’s undersea dominance.
| Date Taken: | 09.19.2023 |
| Date Posted: | 09.19.2023 10:44 |
| Story ID: | 453750 |
| Location: | KEYPORT, WASHINGTON, US |
| Web Views: | 311 |
| Downloads: | 0 |
This work, NUWC Division, Keyport employee participates in Defense Advanced Research Projects Agency's AI Forward workshop, by Frank Kaminski, identified by DVIDS, must comply with the restrictions shown on https://www.dvidshub.net/about/copyright.
No keywords found.
