On Friday, the US Division of Homeland Safety introduced the formation of an Synthetic Intelligence Security and Safety Board that consists of twenty-two members pulled from the tech {industry}, authorities, academia, and civil rights organizations. However given the nebulous nature of the time period “AI,” which may apply to a broad spectrum of laptop know-how, it is unclear if this group will even have the ability to agree on what precisely they’re safeguarding us from.
President Biden directed DHS Secretary Alejandro Mayorkas to determine the board, which is able to meet for the primary time in early Could and subsequently on a quarterly foundation.
The elemental assumption posed by the board’s existence, and mirrored in Biden’s AI govt order from October, is that AI is an inherently dangerous know-how and that Americans and companies should be shielded from its misuse. Alongside these traces, the objective of the group is to assist guard towards overseas adversaries utilizing AI to disrupt US infrastructure; develop suggestions to make sure the protected adoption of AI tech into transportation, power, and Web companies; foster cross-sector collaboration between authorities and companies; and create a discussion board the place AI leaders to share info on AI safety dangers with the DHS.
It is price noting that the ill-defined nature of the time period “Synthetic Intelligence” does the brand new board no favors concerning scope and focus. AI can imply many alternative issues: It could actually energy a chatbot, fly an airplane, management the ghosts in Pac-Man, regulate the temperature of a nuclear reactor, or play an awesome sport of chess. It may be all these issues and extra, and since lots of these purposes of AI work very in another way, there is no assure any two folks on the board can be serious about the identical sort of AI.
This confusion is mirrored within the quotes offered by the DHS press launch from new board members, a few of whom are already speaking about several types of AI. Whereas OpenAI, Microsoft, and Anthropic are monetizing generative AI methods like ChatGPT primarily based on massive language fashions (LLMs), Ed Bastian, the CEO of Delta Air Traces, refers to thoroughly completely different lessons of machine studying when he says, “By driving revolutionary instruments like crew resourcing and turbulence prediction, AI is already making important contributions to the reliability of our nation’s air journey system.”
So, defining the scope of what AI precisely means—and which purposes of AI are new or harmful—is perhaps one of many key challenges for the brand new board.
A roundtable of Massive Tech CEOs attracts criticism
For the inaugural assembly of the AI Security and Safety Board, the DHS chosen a tech industry-heavy group, populated with CEOs of 4 main AI distributors (Sam Altman of OpenAI, Satya Nadella of Microsoft, Sundar Pichai of Alphabet, and Dario Amodei of Anthopic), CEO Jensen Huang of prime AI chipmaker Nvidia, and representatives from different main tech firms like IBM, Adobe, Amazon, Cisco, and AMD. There are additionally reps from large aerospace and aviation: Northrop Grumman and Delta Air Traces.
Upon studying the announcement, some critics took situation with the board composition. On LinkedIn, founding father of The Distributed AI Analysis Institute (DAIR) Timnit Gebru particularly criticized OpenAI’s presence on the board and wrote, “I’ve now seen the complete listing and it’s hilarious. Foxes guarding the hen home is an understatement.”
