Technology experts from the private sector and academia have called on the US government not to “abdicate” its responsibilities and leave the future path of artificial intelligence solely to Big Tech.
This call was issued by Aleksander Mądry, the Cadence Design Systems Professor of Computing at MIT and director of the MIT Center for Deployable Machine Learning, during a Congressional panel held last week.
At the hearing - Advances in AI: Are We Ready for a Tech Revolution? - before the House Subcommittee on Cybersecurity, Information Technology, and Government Innovation, Mądry, a leading expert on explainability and AI, emphasised the importance of government regulation of corporate algorithms.
Mądry argued that the government should ask questions about these algorithms' purpose and explainability to ensure that they align with society's goals. He expressed concern for the future of AI if the government fails to take action.
Other witnesses at the hearing included former Google CEO Eric Schmidt, IBM Vice President Scott Crowder, and Centre for AI and Digital Policy Senior Research Director Merve Hickok. In her opening remarks, Subcommittee Chair Rep. Nancy Mace referenced the book The Age of AI: And Our Human Future by Schmidt, Henry Kissinger, and Dan Huttenlocher, and a related op-ed in The Wall Street Journal that discussed ChatGPT.
Mace also revealed that her formal opening remarks had been entirely written by ChatGPT.
AI has moved beyond science fiction
In his prepared remarks, Mądry raised three main points. He noted that AI has moved beyond science fiction and is now in the world, posing benefits and risks. He also expressed concern about the potential for AI to undermine human analytical skills by causing people to unquestioningly believe what AI produces.
Mądry warned against regulating AI as if it were human and emphasised the importance of addressing problems resulting from the AI supply chain.
Mądry's concerns about the AI supply chain centred on the fact that AI systems are built on top of each other, with general systems like ChatGPT at the base and task-specific AI systems layered on top. This layering raises policy concerns, as the entire AI system is subject to the vulnerabilities and biases of the large system at its base.
The interaction of AI systems is not well understood from a technical standpoint, making it difficult to predict or explain their results. Mądry also pointed out the difficulty of assigning legal liability when problems arise from the mix of AI tools.
Mądry concluded that AI technology is not well-suited for deployment through complex supply chains, even though this is precisely how it is deployed.
“We are at an inflection point in terms of what future AI will bring,” says Madry. “Seizing this opportunity means discussing the role of AI, what exactly we want it to do for us, and how to ensure it benefits us all. This will be a difficult conversation, but we do need to have it, and have it now.”
You can watch a video of the proceedings below, including testimonies from Eric Schmidt, Scott Crowder and Merve Hickok.