Concerns rise over AI’s potential risks to humanity
In a stark warning, a prominent safety researcher at Anthropic suggested that the rapid advancement of artificial intelligence (AI) could lead to humanity’s extinction within the next decade, prompting UK leaders to acknowledge significant national security risks associated with the technology, reports BritPanorama.
During a session in the Commons, Prime Minister Rishi Sunak faced inquiries regarding the dangers of AI following comments from Evan Hubinger, who stated there was over a 10% chance of a species-ending event occurring in the near future. This revelation underscores growing unease within governmental and scientific circles about the implications of uncontrolled AI development.
The situation was further complicated when the Financial Times reported that Anthropic had not shared its latest model with the UK’s AI Safety Institute, a key institution dedicated to assessing the risks posed by AI technologies. The decision raises questions about transparency and collaboration in safeguarding public safety.
Responding to a query from Liberal Democrat leader Sir Ed Davey, Burnham acknowledged the legitimate concerns. He stated, “He is absolutely right that AI poses risks to our national security, but it also could be the source of solutions to keeping us safer.” His remarks highlight the dual-edged nature of AI, presenting both challenges and potential benefits.
To enhance governance in this critical area, Mr. Burnham has elevated the role of AI safety minister Kanishka Narayan by providing him a seat at the Cabinet table. This move, he noted on Wednesday, has “lifted the level of the conversation about artificial intelligence within Government.”
In the face of escalating risks, Hubinger shared on social media his fears about AI’s unchecked progress, asserting, “We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade.” His comments reflect growing alarm within the tech community regarding AI’s trajectory.
Hubinger’s post followed a resignation by fellow Anthropic researcher Jacob Coxon, who expressed grave concerns. He described the competition between tech firms to develop self-improving superintelligence as “gambling with our lives” and criticized both Anthropic and its rival OpenAI for not acting responsibly.
Coxon elaborated, stating, “The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt,” and emphasized the unique dangers associated with AI technology.
In light of these warnings, former Cabinet Office minister Darren Jones has called for an international treaty to regulate AI development. In a letter to PM Sunak, UN Secretary-General Antonio Guterres, and OECD Secretary-General Mathias Cormann, he argued for a safety-first approach: “We need a new multi-national treaty for the regulated and safe development of superintelligence.” He stressed the importance of innovation but within a framework that prioritizes safety.
A Cabinet Office spokesperson asserted the UK’s leadership in AI safety, noting, “The AI Security Institute continues to collaborate closely with industry partners, including Anthropic, to make models safer.” The spokesperson emphasized the necessity of addressing these risks collaboratively and scientifically, stating, “These risks do not stop at national borders and no country can tackle them alone.”
As discussions on AI safety intensify, the balance between innovation and risk management has never been more crucial. The dialogue around effective governance remains pivotal as technological advancements continue to evolve at an unprecedented pace.
