How do we build artificial intelligence and c systems that people can trust?
DARPA wants to answer this question with the help of academic, government, and industry experts through its AI Forward initiative.
To kickstart AI Forward, DARPA will host two workshops in summer 2023 – a virtual event on June 13-16 and an in-person event located in Boston on July 31-August 2. During each workshop, participants will have an opportunity to brainstorm new directions towards trustworthy AI with applications for national security.
“Though we’ve seen great progress in the field, AI still requires significant investment and advancement to develop technology that reliably operates, interacts appropriately with people, and meets the most pressing national security and societal needs in an ethical manner,” said Dr. Kathleen Fisher, director of DARPA’s Information Innovation Office. “We are calling on all relevant researchers and experts to join us as we define the future of today’s most consequential technology.”
DARPA experts estimate that research in the following areas will be essential to creating trustworthy technology:
- Foundational theory, to understand the art of the possible, bound the limits of particular system instantiations, and inform guardrails for AI systems in challenging domains such as national security;
- AI engineering, to predictably build systems that work as intended in the real world and not just in the lab; and
- Human-AI teaming, to enable systems to serve as fluent, intuitive, trustworthy teammates to people with various backgrounds.
Factors such as reducing costs to produce and train state-of-the-art machine learning models and improving their resiliency to adversarial attacks will be equally important.
“By participating in AI Forward, you’ll have the opportunity to help chart the course for capabilities that will be essential to keeping our nation and our allies safe for decades to come,” said Fisher.
Each AI Forward workshop will include guided discussions with industry, academic, and government leaders on current and future research directions. At the end of each workshop, DARPA will ask participants to produce white papers that will inform rapid explorations of select topics to be funded by the agency.
Participation will be limited to approximately 100 people per workshop and requires submission and acceptance of an application to join one of the two events being held. Interested individuals must submit their request to participate by Monday, March 20 via the application website: https://apply.knowinnovation.com/darpaaiforward/
Those with questions about the submission process may email [email protected].
More information on trustworthy AI can be found at the following replay of Fisher’s presentation from the 2022 DARPA Forward event series: https://www.youtube.com/watch?v=wOrMZdyRlL0&list=PL6wMum5UsYvY2odwylYg81JoQhu3P7D1r&index=4