AI Safety · Satire
AI Alignment Researcher Who Is 'Very Concerned' About AI Still Building AI Full Time
Dr. Nathan Wells says the risks of advanced AI are 'existential and imminent' and has published fourteen papers on the topic while accepting a $450,000 salary from an AI lab.
NEW YORK — Dr. Nathan Wells, a senior alignment researcher at a leading AI laboratory, gave a keynote address at the Machine Intelligence Research Institute conference Thursday in which he described the development of artificial general intelligence as potentially “the last mistake humanity will make.”
Dr. Wells has worked full time on AI development for eleven years.
“I want to be clear,” said Wells, standing in front of a slide that said EXISTENTIAL RISK in large red letters. “I am deeply, profoundly concerned about what we are building. The trajectory is alarming. The timelines are shorter than most people think.”
Wells then returned to his office at DeepMind, where he is working on a new architecture he believes may be transformative.
“The only way to ensure safety is to be in the room,” Wells said when asked about the apparent contradiction. “If not us, someone worse.”
When asked if the someone worse might describe themselves the same way, Wells said the question was “a bit unfair.”
At press time, Wells had signed an open letter calling for a six-month pause on AI development and then accepted a grant to continue his research.
