What’s behind the AI industry’s latest warnings of doom?
techcrunch ·
TechCrunch’s Equity podcast discussed the AI industry’s latest warnings about existential risk. The debate began after researcher Jacob Coxon said he resigned from Anthropic because leading AI companies were gambling with lives, and Anthropic’s alignment lead posted that AI could kill all humans and put the chance at more than 10% within the next decade. Anthony Ha said he was skeptical of many doomer narratives and criticized the vague “we” and made-up percentage, while crediting Coxon for acting on his stated beliefs. Kirsten Korosec asked whether such warnings were a strange way to show off how advanced a company’s AI model is, especially as companies prepare to go public, and Sean O’Kane wondered how it would affect Anthropic’s S-1 filing for its IPO. The panel also noted recent events like a Hugging Face hack involving OpenAI’s internal model and new models from Anthropic and OpenAI with Astra. Korosec suggested that in current markets, danger and capability might even be seen as beneficial to valuation, and the group discussed ControlAI’s Connor Leahy on control efforts. Ha said the doomer focus can distract from more immediate labor, environmental, and climate harms, and that AGI and superintelligence language can crowd out other discussion. They recorded before Anthropic CEO Dario Amodei published a plan for more cautious AI development.
techcrunch