The United Kingdom’s prime minister’s adviser on the artificial intelligence task force estimated that before AI gets overly strong, humans have around two years to govern and regulate it.
Matt Clifford, who also holds the position of chief of the government’s Advanced Research and Invention Agency (ARIA), emphasized in an interview with a local U.K. media site that current systems are “more and more capable at an ever-increasing rate.”He went on to suggest that the systems will become “very powerful” in two years if officials don’t take safety and laws into account right away.We have two years to have a structure in place that will make controlling and regulating these enormous models much easier than it is at the moment.In the short and long terms, Clifford cautioned that there are “a lot of different types of risks” associated with AI, which he described as “pretty scary.”
The interview followed the publication of an open letter by the Center for AI Safety, which featured 350 AI professionals, including Sam Altman, CEO of OpenAI, who said that AI should be viewed as an existential threat comparable to that posed by nuclear weapons and pandemics.They are discussing what would occur after we successfully construct a new species with an intelligence greater than that of humans.These risks posed by AI, according to the AI task force adviser, might be “very dangerous” and “kill many humans, not all humans,” only based on where we anticipate models to be in two years.
Clifford contends that learning how to regulate the models and then putting restrictions into effect globally should be regulators’ and developers’ top priorities.He stated that for the time being, his biggest concern is that we won’t fully comprehend why AI models act the way they do.The creators of the most advanced systems publicly acknowledge that they are unable to fully comprehend how [AI systems] behave as they do.Clifford emphasized that a large number of the executives of companies developing AI also concur that effective AI models need to go through some sort of audit and evaluation procedure before deployment.
Regulators around the world are currently working hard to comprehend the technology and its implications while attempting to develop rules that both protect consumers and promote innovation.On June 5, representatives from the European Union went so far as to propose requiring all AI-generated content to include a disinformation-prevention badge. A front-bench Labour Party member in the U.K. backed the Center for AI Safety’s letter in arguing that technology needs to be governed similarly to how nuclear power and medicine are.