Gary Marcus

Rank 12 of 47

Score 83

“Our most advanced AI technologies need to rigorously comply with the same standards we expect of human officials: They should obey the law.”

💯 agree. Unfortunately, I see no evidence that LLMs will ever be able to do this reliably.

If we want safe AI, we are going to need

Lawfare

@lawfare

265d

Artificial intelligence agents—which allow for end-to-end automation of computer-based tasks—are the next big frontier in general purpose AI. Cullen O'Keefe and @ketanr argue that if these agents are to be deployed by the government, they need to be trained to follow the law.

5/23/2025, 10:12:23 PM

The statement emphasizes the importance of AI technologies adhering to legal standards, especially when deployed by the government. It expresses skepticism about the current capabilities of large language models (LLMs) to reliably follow laws, highlighting a concern for safety and compliance.

Principle 1:
I will strive to do no harm with my words and actions.
The statement aligns with the principle of doing no harm by advocating for AI compliance with laws, which is crucial for safety. [+2]
Principle 3:
I will use my words and actions to promote understanding, empathy, and compassion.
It promotes understanding and empathy by addressing the need for AI to be safe and lawful, which benefits society. [+1]
Principle 6:
I will use my influence for the betterment of society.
The statement uses influence to advocate for responsible AI deployment, aiming for societal betterment. [+2]