aiqbench analysis
The article primarily examines expert skepticism regarding unverified existential risk claims and the lack of empirical evidence for AI-induced extinction scenarios, placing it squarely within the domain of AI safety discourse. It highlights the scientific debate over whether specific probability estimates for "doom" are methodologically sound or merely speculative assertions lacking falsifiability. TechPulse take: Experts at the AI Now Institute argue that assigning specific probabilities to existential AI risks is scientifically invalid due to a lack of concrete examples and verifiable data, challenging the authority of such claims in policy and public discourse. This analysis underscores the ongoing tension between alarmist narratives and the demand for rigorous, evidence-based safety assessments in the field of artificial intelligence.
Could AI really wipe out humanity – six experts spell out the risks
In response to alarming claims and essays warning about AI safety, AI Now's Heidy Khlaaf pushes for evidence and concrete examples that can verify these claims. The post Could AI really wipe out humanity – six experts spell out the risks appeared first on AI Now Institute.
There have been some shocking claims in recent days about AI safety: we face a 10% chance of doom; AIs are worse than nukes; a “botnet” threatens the entire internet; it’s all a big tech psyop.
But Heidy Khlaaf, chief scientist at the AI Now Institute research body, is among experts sceptical about giving percentage chances to existential scenarios.
“These claims are neither falsifiable nor verifiable, making then unscientific by their very nature,” she said. The shortage of concrete examples of what a “superintelligent” AI might do to cause “doom” only adds to the cloudiness. A source familiar with Anthropic’s thinking acknowledges that “the exact chances of any one outcome are probably unknowable”.
“If you cannot provide such evidence, it also implies whatever numbers you produced for ‘extinction’ have no scientific basis,” Khlaaf added. Critics say such numbers borrow the authority of science without doing the work.
Read the article here.
Full text under the publisher's open licence · 151 words
