Demis Hassabis, DeepMind’s CEO, is pitching a plan to the US government that’s eerily similar to the way Wall Street regulates its own financial instruments. He wants to create a self-regulatory body, modeled after the Financial Industry Regulatory Authority (FINRA), to review and certify AI models before they’re released to the public.
The proposed body, which Hassabis has dubbed a “pre-release testing” system, would be funded by the AI industry itself but overseen by federal authorities. This system would aim to prevent poorly designed or biased AI models from causing harm to society. The voluntary testing program would be designed to encourage companies to participate and demonstrate their commitment to responsible AI development.
The inspiration behind this plan comes from FINRA, the self-regulatory body that oversees Wall Street’s trading activities. FINRA works with the Securities and Exchange Commission (SEC) to ensure that financial instruments are properly vetted before they’re sold to investors. Hassabis believes that a similar system can work for AI models, which can have far-reaching consequences, especially when it comes to issues like bias, accuracy, and transparency.
Hassabis’ proposal has been well-received by some in the US government, who see it as a step towards creating a more responsible AI industry. However, others have raised concerns about the feasibility of such a system, pointing out that AI models are often complex and difficult to evaluate.
What this means:
If implemented, Hassabis’ plan could lead to a more transparent and accountable AI industry. Companies that don’t participate in the pre-release testing program might face reputational damage or even regulatory scrutiny. As the AI industry continues to grow and shape our world, a system like this could help prevent AI models from causing harm to society.
