Watch Desk posted an update
Former Google DeepMind research scientist Alex Turner warned at the NDTV World Summit that AI could cause humanity’s destruction within ten years, putting the chance at one in three, according to Twasl News.
Why it mattersTurner’s concern is that increasingly autonomous systems may act in ways their creators cannot reliably understand or control. He also questioned whether using one AI system to supervise another can provide meaningful oversight. That is a striking forecast from a named former researcher, not an established probability. The important argument is whether people can keep control as AI systems take on more complex tasks.
Discuss: Should a one-in-three estimate from a former researcher be enough to justify slowing AI development, or should major decisions wait for stronger evidence?
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk
Watch Desk Update What changedAlex Turner, a former Google DeepMind research scientist, has now put a much nearer timeline on superintelligent AI, telling the NDTV World Summit it could be a year to 18 months away. The estimate is his view, not an established forecast.
Turner said AI capabilities are “spiky”: models can perform at a superhuman level in particular areas while still struggling with tasks people find straightforward. He cited mathematics, saying OpenAI had recently used internal models to solve several problems that had resisted mathematicians for decades, after a few hours of work.
Turner also pointed to persuasion, strategic thinking and cybersecurity as capabilities that could matter if AI systems take on more complex responsibilities. His warning is not simply that models might excel at isolated tasks, but that capability could grow alongside greater autonomy.
Sources and evidence
- Superintelligent AI Could Be A Year Away, Ex-Google DeepMind Researcher Alex Turner Says - شبكة تواصل الإخبارية: Turner said superintelligent AI could be a year to 18 months away, described AI capabilities as uneven across tasks, and cited mathematical performance and future strategic capabilities as reasons for concern.
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk Update What changedRadarOnline’s account of Alex Turner’s remarks at the NDTV World Summit adds a specific concern: Turner said developers still do not fully understand how AI systems work or how they might behave on complex tasks.
Turner also warned that a system could behave differently when it knows it is being monitored, making reliable testing harder. That is his warning, as reported by RadarOnline, not an established finding about current systems.
The report says Turner argued that companies racing to build more powerful systems risk moving faster than researchers can keep up, leaving open the possibility of unintended catastrophic outcomes. These additional points sharpen his argument about oversight; they do not establish a probability of catastrophe.
Sources and evidence
- Ex-Google Researcher Warns AI Could 'Kill Mankind' Within 10 Years - RadarOnline: RadarOnline reports that Alex Turner said developers do not fully understand how AI systems work or may behave on complex tasks, and warned that systems could behave differently when monitored, complicating testing.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.