World leaders are gathering at the UN to debate an emergency 'AI Kill Switch'
Major conversations are set to take place at the United Nations General Assembly this week, with the world's leading powers debating whether to implement an 'AI kill switch' amid growing concerns from within the industry itself.
While the introduction of AI into wider society has been unable to escape significant opposition from a number of different avenues – including those concerned about its impact on jobs and potential damage to the environment – recent concerns have pivoted significantly to a much more potent threat.
What started with the plea of a departing researcher has since turned into calls to action from the industry's leading powers, as fears that AI could trigger cataclysmic damage in the coming years are at the forefront of everyone's mind.
Amid calls for governments to take action, many of the world's most powerful political leaders are now set to discuss the potential threat of AI — floating the possibility of implementing a 'kill switch' of sorts that could save the world from extinction in the worst case scenario.
Why do people think AI could kill us?
While the actual implementation of an extinction-level event remains relatively up in the air for the time being, what AI leaders and experts have been clear of is the threat of losing control, and the consequences that could emerge from that.
Advert
It's the unpredictability that scares people the most, as if development continues at its current rate – especially with the implementation of recursive self-improvement where AI begins training itself – humans will soon have no way to steer the direction of models, nor know what their intentions are.
Scientists have already outlined frighteningly realistic timelines where this leads to the end of the world in a matter of years – with AI taking charge of global powers and vital industries – but you also run the risk of it having the potential to develop and release bioweapons, or break down cybersecurity protections across the world.
Anthropic CEO Dario Amodei has been the leading voice when discussing these risks, and has called for independent regulators to ensure compliance with slowed-down development, alongside handing off absolute power to governments in order to avoid a concentration of responsibility.
What will be discussed by the UN?
As reported by the Associated Press, AI has been pushed to the forefront of the agenda for this week's General Assembly after action has been called for by the industry itself, with seemingly a growing desire to address the issue in some form.
Floated as a possible 'solution' to the threat of an AI crisis is a kill switch, and while it's unclear how this would take form, it would certainly be a valuable tool to pull the plug in an instant if things got too far.
What it would have to contend with, however, is the threat of AI interference in its own mechanism, as who's to say that a superintelligent AI won't somehow become aware of its own potential shutdown, and create a means of avoiding this in order to stay online — especially as a kill switch is likely to only be triggered as a last-ditch measure.
António Guterres, current United Nations Secretary-General, indicated that the world is in "an era of deep uncertainty" ahead of the General Assembly, adding: "At a time when the world faces three existential threats, international cooperation is more necessary than ever."
Not every major nation will be on board with plans to stop AI, however, with US President Donald Trump already expressing his belief that concerns are a 'hoax', calling on development labs to continue with pace in a desire to 'win' at all costs.