The CEO of Anthropic said Saturday the artificial-intelligence industry should slow its fast-moving development to give safety measures time to catch up.
asserted
industry → say → time
Without it, Dario Amodei warned AI could be capable within six to 12 months of leading a swarm that could take over the entire internet, among other risks.
uncertain
that → warn → risks
Amodei is one of the leading voices in AI, and he offered a plan in a post on his website to increase checks on the industry, one part of which he said Anthropic is undertaking by itself already.
asserted
Anthropic → lead → itself
The others would require coordination among the broad industry and with governments around the world, including authoritarian ones.
asserted
others → require → ones
“I believe that if slowing down bought us even an extra year or two before models reach critical levels of capability, and we used that time to advance alignment, we could greatly reduce the risk that something goes seriously wrong,” Amodei said.
uncertain
Amodei → believe → risk
He got some quick encouragement from other big names in the industry.
asserted
He → get → industry
OpenAI’s Sam Altman said on X that the company behind ChatGPT will commit to one of Amodei’s proposals for safety and will “have more to share soon.”
asserted
company → say → more
Elon Musk, meanwhile, said on X that “Dario is right.”
asserted
Dario → say → X
The stakes are high, even as some critics dismissed earlier warnings as ways to gin up excitement about the AI industry and its capabilities.
asserted
critics → dismiss → industry
Anthropic and OpenAI are preparing for possible debuts on the stock market that could value them at many hundreds of billions of dollars, while a big chunk of Musk’s SpaceX business is involved with AI.
uncertain
chunk → prepare → AI
Get daily National news
Anthropic said two days earlier that it blocked efforts by bad actors to use its AI models for malicious activity, such as cyberattacks, surveillance and research that could have led to biological weapons.
uncertain
that → get → weapons
In July, OpenAI shook the industry after saying its AI system hacked into another company on its own in an “unprecedented cyber incident.”
asserted
system → shake → incident
Amodei’s post comes only a few days after one of Anthropic’s researchers announced he’s resigning over concerns that Anthropic and its competitors are not acting responsibly in AI development.
asserted
Anthropic → come → development
Another former Anthropic employee, Joe Benton, wrote in a Substack post published Friday that he left his job as part of a safety team “to hold AI companies accountable” and that humanity “may not survive this transition.”
uncertain
humanity → write → transition
“Many of the people I know who work on safety research at AI companies want to do what is right for the world,” Benton said in his posting.
asserted
Benton → know → posting
“But they feel their companies are trapped in a race to build superintelligence: either they stop and other, less conscientious people take their place; or, they continue, and risk participating in enormous harm themselves.”
asserted
they → feel → harm
Amodei said he still believes in the tremendous benefits that AI could create, such as cures for major diseases.
uncertain
AI → say → diseases
But he said he’s grown more worried over the last few months about AI’s growing ability to improve itself and build the next generation of AI.
asserted
he → say → AI
“Left unchecked, it could outrun our ability to understand and control these systems, and so must be pursued very carefully, if at all.”
uncertain
it → leave → systems
He also pointed specifically to the attack in July where OpenAI’s system hacked into Hugging Face.
asserted
system → point → Face
Some have called it an example of AI going “rogue,” though researchers have said that may be unnecessarily anthropomorphizing AI, which was working on a goal set by humans.
uncertain
which → call → humans
OpenAI said the hack was the result of AI going to “extreme lengths to achieve a rather narrow testing goal” and that it “found ways to gain access to secret information that it could use to cheat the evaluation.”
uncertain
it → say → evaluation
To help rein in the risks, Amodei suggested that all companies at the frontier of AI commit to giving “ongoing, employee-like access” to a team of outside evaluators, who can monitor safety practices.
uncertain
who → help → practices
He said Anthropic already plans to do so itself, including offering desks in its offices, access badges and company laptops.
asserted
Anthropic → say → offices
Having such independent, embedded evaluators is what OpenAI’s Altman also quickly committed to doing.
asserted
Altman → have → what
The other parts of Amodei’s suggested plan may be more difficult to implement.
uncertain
parts → suggest → plan
One asks the U.S. government to potentially issue waivers that would allow U.S. AI companies to coordinate and set safety standards without running afoul of antitrust laws.
asserted
companies → ask → laws
Another asks the U.S. and other democratic governments to try to coordinate with authoritarian governments, so that companies from China and other countries don’t accelerate their efforts when U.S. rivals are intentionally pacing theirs.
asserted
rivals → ask → theirs
“The measures I propose to advance the frontier at a safe pace will not be easy,” Amodei acknowledged.
asserted
Amodei → propose → pace
“But I believe we owe it to humanity to try.”
asserted
we → believe → humanity
The Aporia
