Bert Hubert on AI warnings from Big Tech

Tech expert Bert Hubert recently characterized the warnings from executives at OpenAI, Google DeepMind, and Anthropic about 'existential' AI risks as a incoherent narrative. These leaders advocate for global action and strict regulations, while their own systems rapidly evolve. Given the urgency of the AI regulation in Europe and its implications for both government and businesses, this issue is highly relevant.

Hubert notes that the current warnings are often vague and unclear, making it difficult for regulators and policymakers to formulate effective guidelines. Crucial information regarding the risks is not clearly delineated or weighed, making it hard to take appropriate steps. The consequences are significant, as companies do not know what to expect when they purchase AI systems, creating uncertainty in the market.

Particularly concerning is the tension between the commercial interests of Big Tech and public safety. The same companies that issue warnings also create the most powerful AI models, with business models based on scaling up. Hubert argues that clear, universally applicable requirements are essential to prevent existing players from leveraging regulations to their advantage, which could lead to 'regulatory capture'.

With recent legislation surrounding AI, the call for transparency is growing louder. The goal is to ensure accessible data, audits, and a clear overview of used models. The EU imposes requirements on high-risk applications and guidelines for foundation models. This impacts not only organizations but also citizens, who have the right to clear communication about how AI is used in governmental decisions. Hubert's conclusion is that safety demands concrete and verifiable requirements, such as standardized evaluations and independent oversight. This is crucial to substantiate the promises around AI with >feeling and understanding, and to bridge the gap between policymakers and market players.

Read the full article from AI Insider.