"I won't lie to you – there are real dangers," said Anthropic CEO and cofounder Dario Amodei about the development of artificial intelligence.
asserted
Amodei → lie → intelligence
"And I think for too long the industry lied to people about the fact that this technology had risks.
asserted
technology → think → risks
"
And the biggest risk could be the end of humanity, former Anthropic researcher Jacob Coxon warned on social media this past Tuesday, saying Anthropic and OpenAI are
with the development of self-improving superintelligence.
uncertain
Anthropic → warn → superintelligence
"It doesn't look that different from, say, 'Terminator' or from science fiction films," Coxon told CBS News Thursday.
asserted
Coxon → look → News
"It will be smart enough to kill us."
asserted
It → kill → us
In response, Amodei told "Sunday Morning" yesterday, at Anthropic's headquarters in San Francisco, "I don't think I fully just appreciated what it would actually be like when the progress was as fast as it was."
asserted
it → tell → what
He said the development of AI has been on an exponential curve, which is starting to get steeper.
asserted
which → say → curve
Asked what that means, Amodei said, "It doesn't mean we need to panic today.
asserted
we → ask → what
It doesn't mean we need to shut it all down.
asserted
we → mean → it
But I would say it's a warning sign.
asserted
it → say → ?
It's a warning sign that we need to slow down."
asserted
we → need → ?
In an essay published Saturday, Amodei wrote that the entire tech industry must work together to "slow the pace" of AI capabilities.
asserted
industry → publish → capabilities
Asked if Anthropic is going to stop releasing more advanced models, Amodei replied, "It doesn't mean that.
asserted
It → ask → that
What it means is that we need to make sure that every generation of models that we release is properly tested."
asserted
we → mean → that
To attempt that, Amodei says Anthropic will now give independent evaluators permanent "employee-like" access to their models.
asserted
Anthropic → attempt → models
"Whenever anyone builds an AI model, there's a third-party evaluator who can verify whether that company is following the safety practices that they're committed to – it's like a food inspector," he said.
asserted
he → build → inspector
But AI experts say this is hardly enough, and Amodei agrees, claiming he supports federal regulation.
asserted
he → say → regulation
So far, proposed legislation ranges from an outright ban on superintelligence, to a mandatory "kill switch" that could shut down risky AI.
uncertain
that → propose → AI
"The AI kill switch could be a good idea," Amodei said.
uncertain
Amodei → say → ?
"But I don't think banning it, completely banning it, is the right approach.
asserted
banning → think → it
One, the idea that we would never get the incredible benefits, never get the medical cures, that doesn't sit right with me.
asserted
that → get → me
Second, this technology will be built in other countries, including authoritarian countries."
asserted
technology → build → countries
Asked if the U.S. should slow down AI development if China would not, Amodei replied, "This is the toughest dilemma.
asserted
This → ask → development
He looks to the nuclear arms agreements with the former Soviet Union as a potential model, but cautions even his expectations are low.
asserted
expectations → look → model
"The more long-term thing would be working together to put a speed limit on the rate of AI progress," he said.
asserted
he → work → progress
"I think that's going to be very difficult, because the incentives to pull ahead in the military advantage that you get from that are so large.
asserted
you → think → that
And honestly, I don't know if it's possible, but we should try.
asserted
we → know → ?
With the fate of humanity seemingly at risk, how much personal responsibility does Amodei feel about where we are going?
asserted
we → feel → responsibility
"My view here is, it has always been very strange that this technology is being built by a private company," he said.
asserted
he → build → company
"People ask me that question all the time: Why isn't this being built by government?
asserted
this → ask → government
And the strangest thing about it is, I agree with them.
asserted
I → agree → them
Asked if he is willing to give up the technology to the government, Amodei responded, "To the right combination of governments.
asserted
Amodei → ask → governments
The worry I have – I am concerned that one single government could abuse this technology just as easily as a single company could.
uncertain
company → have → technology
But I think a combination of democratically-elected governments, I don't know about handing over, but some kind of oversight, some kind of joint governance.
asserted
I → think → governance
The question is no longer whether AI will become powerful enough to require constraint, but who in power will take action.
asserted
who → become → action
"If we build in the right way, I think the probability of something bad happening is very low," he said.
asserted
he → build → way
"If we build in the wrong way, the probability of something bad happening is very high."
asserted
something → build → way
