AI Could Claim Countless Lives; Anthropic Researcher Quits, Citing Major Threat to Civilization
An AI researcher has resigned from Anthropic citing ethical concerns, claiming that this technology could claim countless lives over the next decade as it poses a threat to humanity. There is ongoing debate about AI displacing jobs and endangering people; this resignation and the accompanying claim by an AI researcher highlight the potential dangers of AI. Let us explore this in detail.
Jacob Coxon, an AI researcher at Anthropic, announced his resignation in a post on X on Tuesday. In the post, he criticized the AI development practices of both Anthropic and his previous employer, OpenAI.
In his post on X, he wrote, “I have resigned from Anthropic; I spent the last three years working on pre-training research at OpenAI and Anthropic. Neither company is operating responsibly. Both are engaged in a race to build rapidly self-improving superintelligence, effectively gambling with our lives.” He added, “Do not make the mistake of underestimating the power of this technology. Very soon, these will become superhuman systems capable of hacking anything. They could revolutionize any field overnight and acquire real power and resources. We have witnessed growth across all these domains, and it is not slowing down at all.”
He further noted that those developing AI believe it could potentially wipe out humanity by the end of the decade. “This is not a marketing stunt. While many executives and senior researchers speak cautiously in public, I have seen those same individuals express fear. No human activity poses a threat of this magnitude.” A common question asked in this context is: if they truly believe this, why are they developing it? Many people at OpenAI failed to fully grasp the gravity of the potential impact on civilization. While the team at Anthropic understands the implications well, they too are caught up in the race to be first. They operate on the belief that since no one else will act responsibly, they must undertake the task themselves, despite the risks.
Engaging in this race and stepping into the endgame is an arrogant gamble—one that should not be initiated from the Slack channels of a private company. Before attempting to fast-track the alignment process, one must be certain that no superior alternative exists. I remain hopeful about the possibility of coordination; the attack on Hugging Face has made it more likely that US-based labs will reach an agreement regarding the pace of development. I do not believe we are currently on the right path to halt this global race; doing so might require drastic measures, such as a temporary moratorium on enhancing model capabilities.
