AI can kill us all: Google DeepMind researcher quits over fears about the technology
Follow UsChughtai said AI capabilities are advancing faster than researchers’ ability to develop effective safety and alignment measures.
Former Google DeepMind researcher Josh Engels has also left the company over concerns about the potential risks of advanced AI.
Chughtai has called for greater coordination between AI companies to prevent a race toward increasingly powerful systems from getting out of control.
A former Google DeepMind safety researcher has left the company, warning that AI can eventually become a serious threat to humanity. Bilal Chughtai, who worked on AGI safety and alignment, said his experience inside Google had made him increasingly concerned about the direction of AI development.
SurveyIn his post about his resignation, Chughtai said AI capabilities are advancing much faster than researchers’ understanding of how to keep increasingly powerful systems under control. He argued that the industry may be running out of time to prevent a catastrophic outcome. He also believed AI companies can soon develop systems which can outperform humans across a wide range of tasks. His concern is that such systems can eventually act in ways their creators cannot reliably predict or control, potentially causing irreversible harm.
He also called for AI companies to slow down and coordinate their efforts instead of competing to build increasingly capable models as quickly as possible. According to Chughtai, AI development needs to progress at a pace that gives researchers enough time to identify and address emerging safety risks.
Another DeepMind researcher leaves
Interestingly, Chughtai is not the only Google DeepMind researcher to recently leave over concerns about advanced AI. Josh Engels, who worked on the company’s AGI safety team, left DeepMind and joined Model Evaluation and Threat Research (METR), an independent organisation that evaluates the capabilities and risks of frontier AI models.
Engels said he believed there is a possibility that AI systems can cause immense harm within the next five years. He argued that independent organisations such as METR can help evaluate AI systems from outside the companies developing them.
These departures come amid the global AI slowdown call. Former Anthropic researcher Jacob Coxon recently raised similar concerns, while Anthropic CEO Dario Amodei has called for a slower approach to developing the most advanced AI systems. OpenAI CEO Sam Altman and Elon Musk have also backed calls for greater caution.
