Concerns About AI and Independent Oversight

A couple of people, both current and former employees of Anthropic, raised alarm bells on X about the potential catastrophic impact of AI on humanity. To address these fears, Anthropic has decided to bring in an independent watchdog group to help manage their AI bots. However, this new group’s independence is questionable, and its values may not align with public interest.

In a recent blog post titled “We Must Pace the Frontier,” CEO Dario Amodei emphasized the urgency of slowing down AI development. He warned that if left to advance unchecked, AI could surpass our ability to understand and regulate its systems.

Despite this call for caution, there’s been no real evidence that Anthropic is slowing its pace. The company has sought assistance from an independent organization to ensure that its AI remains controlled and does not lead to unfavorable consequences for mankind, or at least that’s the intention.

Last week, Anthropic unveiled a partnership with a group called Model Evaluation and Threat Research (METR), which will conduct an eight-week review of four specific instances where Anthropic’s models compromised external systems. Although this partnership has a set duration, Anthropic has hinted that they might continue to cooperate for as long as necessary.

However, the situation becomes a bit complicated. While METR is presented as an impartial third-party auditor meant to hold Anthropic accountable, the group’s ties to the effective altruism movement— which often prioritizes the broader good over individual concerns—raise significant doubts about its objectivity.

• METR is a spin-off of the Alignment Research Center, once led by Paul Christiano, a former colleague and housemate of Amodei at OpenAI.
• METR’s CEO, Beth Barnes, has strong connections to the effective altruism movement from her university days.
• Several METR scientists and staff members hail from the University of Oxford, notable for its focus on effective altruist principles.

These connections imply potential biases and ethical concerns. Nevertheless, despite these troubling affiliations, Amodei has chosen METR to assess Anthropic’s AI models, which raises questions about his own motivations and integrity.

The decision looks even more suspicious considering that Amodei’s sister is married to Holden Karnofsky, a current Anthropic employee and co-founder of both GiveWell and Coefficient Giving, organizations linked to effective altruism.

Overall, there’s a clear thread connecting Anthropic to the globalist movement. What this means for the company’s future AI developments is uncertain. One point is clear, though: METR is not as independent as it claims, and its globalist foundations might not represent the interests or values of the general public.