
Writer, researcher, and philosopher working on Claude's constitution at Anthropic
Joe Carlsmith is a writer, researcher, and philosopher who works at Anthropic on the constitution for Claude.
How media typically covers Joe Carlsmith
Referenced in coverage
“Argued that automating alignment research will be crucial to preventing smarter AI systems from causing harm; previously worked at Coefficient Giving on AI safety.”
Building AI systems capable of human-like philosophy is critical for safe out-of-distribution generalization, requiring both capability in philosophical reasoning and disposition to engage in that reasoning when needed.
“Wrote an essay series on solving the AI alignment problem, discussing the importance of building AIs that do human-like philosophy.”