An IAPS survey by Severin Field found 20 of 25 researchers at OpenAI, Anthropic, Google DeepMind, and Meta ranked automation of AI research among the biggest AI risks. OpenAI developer Thibault Sottiaux said Astra boosted internal productivity so much that some plans moved up six months. Anthropic claims Claude writes over 80 percent of its production code. Half of surveyed researchers expect the most powerful models to stay internal.
No score is assigned. Sources and their independence are shown in the citation chain below.