In a high-profile industry departure, Robert O’Callahan, a Google DeepMind engineer based in Auckland, New Zealand, has resigned from his position over concerns regarding the lightning-fast acceleration of artificial intelligence. While O’Callahan loved his role and the intellectual challenges it offered, he concluded that continuing to work on tools designed to optimize and accelerate AI hardware development conflicted with his personal convictions.
O’Callahan did not directly build AI models during his tenure; instead, his work focused on developing tools for hardware chip design. Initially, he considered these backend contributions relatively harmless. Over time, however, he realized that his team’s objective—making AI significantly cheaper, faster, and lower-latency—was actively speeding up the arrival of a new generation of AI chips. Facing a landscape where artificial intelligence is advancing at a frantic pace, he decided he could no longer justify contributing to its rapid expansion.
Interestingly, O’Callahan’s stance diverges from the extreme extinction rhetoric popular among some tech safety advocates. Addressing the debate surrounding Artificial Superintelligence (ASI), he noted that he is not entirely convinced a doomsday scenario is guaranteed. Instead, he views the risks of ASI as “real but uncertain”. However, he argued that moving forward with reckless speed in the face of such profound uncertainty is “inherently irresponsible”. Beyond existential threats, he also highlighted pressing societal concerns, including economic disruption, concentrated corporate power, and over-reliance on automated systems.
Leaving Google came with personal and professional costs, as O’Callahan described the job as an amazing opportunity to solve fun puzzles while being handsomely compensated. Because Google maintains no other engineering teams in New Zealand, internal transfer options were virtually non-existent. Guided by his faith as a church elder and a desire to keep his actions “unambiguously pro-human,” he opted to step down.
O’Callahan clarifies that his departure was not timed to coincide with the recent viral wave of safety-related tech resignations, but rather lined up with a long-planned personal backpacking trip. Moving forward, he does not intend to abandon technology entirely. He plans to continue maintaining debugging tools like rr and Pernosco while exploring how AI code-debugging agents function, ensuring his future contributions remain safe and beneficial for humanity.
