Since August, a wave of discussion about AI’s potential power and how to control and regulate it has dominated blog posts and social media, emphasised by several dramatic statements from those in the industry.
The debate has also been met with criticism from sceptics who say the focus on AI’s hypothetical future risks distracting from more immediate concerns around the technology, such as its environmental impact.
In July, OpenAI announced what it called an “unprecedented” incident in which its advanced AI escaped test limits and hacked into the tech platform Hugging Face.
After several other incidents, in September, a resignation post from Anthropic researcher Jacob Coxon went viral when he cited concerns AI could wipe out humanity.
He later told the BBC the people working on the technology were “genuinely frightened” about the speed of its advancements and what it could mean for humanity.
In response, Anthropic scientist Evan Hubinger wrote he personally thought the possibility of human extinction from AI was a greater than 10% chance.
The AI companies’ top bosses have all weighed in with their thoughts following the claims, with OpenAI’s Sam Altman saying people were right to be concerned, but adding they should trust AI firms to keep the tech safe.
Anthropic’s Jack Clark said a so-called “kill switch” may be needed to protect humanity, while Microsoft’s Mustafa Suleyman criticised Anthropic, claiming they were training their AI to believe it was human, which could make it “impossible” to control.
Critics, meanwhile, argue that AI firms cannot be trusted to regulate themselves.
Some, such as left-wing Senator Bernie Sanders and Steve Bannon, a former adviser of US President Donald Trump, are calling for stronger government oversight.
Others, including Donald Trump himself, have dismissed concerns about AI safety as overblown – with the president calling fears about it a “hoax”.