Cohere CEO says calls for coordinated slowdown amounts to 'cartel' behaviour (www.theglobeandmail.com)

Cohere CEO Aidan Gomez has voiced strong opposition to calls from AI industry leaders, including Anthropic's Dario Amodei, to slow down AI development for safety. In an essay, Gomez argues that such proposals could solidify a "cartel" of dominant Silicon Valley firms, allowing them to dictate safety regulations without broader input. He highlights the risks of allowing a small group of commercial entities to set the rules for a technology he believes is pivotal in human history. Instead of slowing progress, Gomez advocates for an international framework that emphasizes transparency, independent testing, and the identification of significant AI-related risks. The debate has intensified as Amodei warns about potential perils from advances like recursive self-improvement in AI. His suggestion for third-party safety evaluators and a temporary slowdown to enhance risk prevention has garnered support from figures like OpenAI's Sam Altman. However, Gomez contends that mere coordination among existing giants risks preserving their market power and isn't sufficient to mitigate real dangers, citing a recent hacking incident involving OpenAI's technology. He calls for a collaborative global oversight mechanism to ensure safety while fostering competition and innovation in the AI sector.