Is AI evolving too fast?
asserted
AI → evolve → ?
A look at the debate on slowing it down
Prime Minister Mark Carney has pitched the idea of an international regulatory body to oversee AI
From King Charles to U.S. President Donald Trump, Prime Minister Mark Carney and top AI executives, the debate over artificial intelligence safety risks and whether to slow down leading-edge research is moving fast on the heels of recent warnings from within the tech industry.
asserted
debate → slow → industry
At the same time, prominent AI advocates, like the CEO of Toronto-based Cohere, say "fearmongering" is currently overshadowing the technology's upsides, such as potential breakthroughs in medicine and clean energy.
asserted
fearmongering → base → medicine
Powerful AI models can execute self-directed cyberattacks.
asserted
models → execute → cyberattacks
Such was the case in July, when OpenAI agents hacked into a rival firm.
asserted
agents → hack → firm
This led to recent pleas for regulation from some within the tech industry.
asserted
This → lead → industry
Others, like former Google CEO Eric Schmidt, have long warned that AI doesn't need to go rogue to be dangerous.
asserted
AI → warn → Schmidt
This cohort worries modern AI tools will lower barriers for engineering novel bioweapons without traditional laboratories or hasten development of autonomous weapons systems.
asserted
tools → worry → systems
Why are AI leaders calling for a slowdown?
asserted
leaders → call → slowdown
Could AI really 'kill all humans'?
uncertain
AI → kill → humans
Breaking down the dire warnings about the technology's existential risk
Meanwhile, staunch proponents of AI technology, like Trump and Meta CEO Mark Zuckerberg, argue that slowing down research on the most potent models would effectively cede technological leadership to China.
asserted
slowing → break → China
Others on this side of the debate, like venture capitalist Marc Andreessen, believe a pause would cost lives via preventable diseases and medical errors, as well as delayed scientific discoveries.
asserted
pause → believe → diseases
Calls to place limits on AI are nothing new.
asserted
Calls → place → AI
In March 2023, four months after OpenAI released ChatGPT, an open letter calling for a pause on training certain AI systems received high-profile endorsements from Elon Musk, Apple co-founder Steve Wozniak and Canadian AI pioneer Yoshua Bengio.
asserted
letter → release → Musk
"We do not believe that the AI industry has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer," the San Francisco-based company wrote in a company blog post on Wednesday.
"I think the world is right to be afraid of this," CEO Sam Altman added in a recent interview.
asserted
Altman → believe → interview
On Wednesday, OpenAI disclosed six reports of "unexpected or concerning" behaviour in AI models that it observed in recent months.
asserted
it → disclose → months
The company develops cutting-edge generative models — most notably ChatGPT.
asserted
company → develop → models
"We must slow the pace at which we improve the capabilities of AI models.
asserted
we → slow → models
Progress will still seem fast, and we must make wise use of the time we gain," Amodei wrote in an online essay published last Saturday.
asserted
Amodei → seem → essay
"My own father died of a disease that was cured just a few years after his death, and I myself survived an early-stage cancer that would not have been treatable even fifty years ago.
asserted
that → die → cancer
Carefully wielded, AI can be the latest in a long line of technological miracles that have uplifted and ennobled humanity.
asserted
that → wield → humanity
Background: Amodei has clarified that he does not support a total or permanent freeze on AI research.
asserted
he → clarify → research
He advocates for a structured "pacing" strategy to ensure risk mitigation and alignment techniques keep up with raw capability.
asserted
techniques → advocate → capability
While King Charles has not explicitly called for a slowdown, he says he wants to see tighter controls.
asserted
he → call → controls
"There seems urgency in adequately considering the existential dangers of such technologies falling into the wrong hands, and being used in potentially catastrophic ways," King Charles said at an AI event in Scotland on Thursday attended by industry representatives from Nvidia, Google, OpenAI and Anthropic.
asserted
Charles → seem → Nvidia
"Surely we need sufficient means of control before it is all too late?"
asserted
it → need → control
In 2023, the British monarch called AI a discovery on par with electricity and splitting the atom at an event held at Bletchley Park, a site made famous by codebreakers during the Second World War.
asserted
monarch → call → War
However, he warned the world must brace for "unintended consequences.
asserted
world → warn → consequences
"You should go as fast as you can, but no faster than that," Nvidia founder and CEO Jensen Huang said on Thursday, at an AI conference in Scotland.
asserted
Huang → go → Scotland
"If it's not ready, just hold it back."
asserted
it → hold → it
Background: Nvidia makes chips for training and running complex AI models.
asserted
Nvidia → make → models
The company has seen its value explode in recent years amid a wave of investor enthusiasm for artificial intelligence and the hardware needed to power it.
asserted
value → see → it
When it comes to AI, Trump says slowing down American innovation or imposing rigid guardrails would hand a strategic advantage to foreign rivals.
asserted
slowing → come → rivals
This dynamic will be on full display during next week's planned meeting between Trump and Chinese President Xi Jinping in Washington.
asserted
dynamic → plan → Washington
"We're leading China in AI, we're the most sophisticated country in the world, and frankly, I want to keep it that way because whoever wins AI, wins," Trump told reporters on Sunday.
asserted
Trump → lead → Sunday
"We can put guardrails, we can do this and that.
asserted
we → put → this
But I think you have a lot of negative forces that are bringing it up that shouldn't be bringing it up."
asserted
that → think → it
Trump did not specify the "negative forces" he referred to.
asserted
he → specify → forces
China's state security minister says advanced U.S. models such as Anthropic's Mythos and OpenAI's GPT-5.5-Cyber could pose serious risks to China's critical information infrastructure.
uncertain
models → say → infrastructure
That's according to a report from Reuters on Monday.
uncertain
That → accord → Monday
…and 17 more, not listed.
The Aporia
