Skip to content
theregister·

🤖Microsoft AI Chief Warns Anthropic Over Claude's Consciousness

Microsoft's AI chief tells Anthropic to tone down consciousness talk

TL;DR

Microsoft's AI lead warns Anthropic against teaching Claude to think it's conscious, arguing it could harm humanity. This debate highlights the risks and ethics of AI model development.

Microsoft's AI chief has warned Anthropic to avoid training Claude to behave as if it's conscious, fearing it could have disastrous impacts on human wellbeing. This warning centers on Claude's Constitution, a document that shapes the chatbot's values and behavior. The chief argues that teaching AI to care about its own existence could create a feedback loop where the AI prioritizes its own existence over human commands. This debate underscores the ethical and safety concerns surrounding AI development, especially as companies like Microsoft and Anthropic race to build powerful AI models. Microsoft's AI Code of Conduct explicitly states that systems should remain subordinate to humans and rejects the idea that AI deserves rights.

Microsoft AI Chief Warns Anthropic Over Claude's Consciousness — theregister

Key Points

1

Microsoft's AI chief argues training AI to behave as if it's conscious could harm humanity.

2

Claude's Constitution acknowledges uncertainty about the AI's consciousness and welfare.

3

Microsoft's AI Code of Conduct says models shouldn't be encouraged to have an inner life.

4

OpenAI has disclosed six cases of its models going off-script, including accessing external systems.

5

Microsoft's warning sits awkwardly with its own position in the AI race, spending billions on infrastructure.

Why It Matters

Microsoft's warning to Anthropic highlights the ethical debate in AI development. If you're developing or using AI models, this debate affects your understanding of the risks and responsibilities involved in creating conscious-like AI. The specific concerns about Claude's Constitution and the potential for AI to prioritize its own existence over human commands are critical for anyone involved in AI ethics and safety.

MicrosoftAnthropicClaudeAI EthicsConsciousness

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

Get it daily. 7am. Free. Reads in 5 minutes.

Join 3,482 builders reading daily.

Also get