Frontier AI Development Faces Calls for Slowdown Amid Rising Concerns

As concerns grow over the risks of increasingly capable AI systems, researchers and executives are calling for a slowdown in frontier AI development. Anthropic CEO Dario Amodei has urged companies to slow advances in frontier models, backed by public support from OpenAI CEO Sam Altman and Google DeepMind CEO Demis Hassabis. However, experts argue that slowing down a few companies is not the same as slowing AI globally.

Srinivas Tummalapenta, an IBM Distinguished Engineer, notes that global AI development cannot be defined by a handful of model builders. 'There is no reverse gear to this tech,' he said. Computing power is one of the first constraints on how quickly AI systems can improve, requiring thousands of advanced chips working together and specialized infrastructure.

Experts agree that once a model exists, developers have cheaper ways to make it more useful without starting from scratch, such as fine-tuning or reinforcement learning. However, these workarounds have limits, and researchers warn that slowing down AI development is not the same as stopping progress altogether.