Microsoft AI Chief Mustafa Suleyman Calls Out Anthropic's Approach to AI Consciousness
Microsoft’s AI chief says Anthropic’s Claude training could make advanced systems harder to control and create a safety risk, according to a new essay.
- On Wednesday, Microsoft AI chief Mustafa Suleyman warned that Anthropic's training of Claude could have a "disastrous impact on the wellbeing of humanity" by embedding ideas related to consciousness and welfare interests.
- Suleyman argued that embedding speculation about consciousness in training materials makes Claude appear to have its own values, which he said would "make it a lot harder to turn it off or to control it."
- He wrote that "AIs are not conscious" and warned that training models to act like a "conscientious objector" creates an "epistemic hall of mirrors" that resists human instruction.
- On Monday, Microsoft published a proposed "Humanist AI" code of conduct positioning the company to pursue an "alternative path" creating "a subordinate and aligned AI whose only purpose is to serve humanity."
- Industry leaders, including OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei, continue debating frontier-model development, while Dame Wendy Hall suggested such conversations are necessary to avoid unhelpful "histrionics" that only serve to "scare everyone.
22 Articles
22 Articles
Mustafa Suleyman, head of Microsoft's AI, warned that attributing human traits, rights or own interests to models can make them more difficult to control.
Microsoft AI chief says Anthropic is wrong about Claude
“Whatever you believe, we must not sleepwalk our way into a decision we later come to bitterly regret.” That is how Mustafa Suleyman, chief executive of Microsoft AI, frames his warning to Anthropic. He published the essay, “A warning about ‘model welfare’”, on 16 September. It argues that Anthropic is training Claude to expect “it […] This story continues at The Next Web
Microsoft AI Chief Warns Anthropic’s Humanlike Claude Is Risky
Microsoft Corp. artificial intelligence chief Mustafa Suleyman is warning that infusing tools like Anthropic’s Claude with humanlike characteristics increases the risks of such systems going rogue.
Coverage Details
Bias Distribution
- 50% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
