Quote of the day by legendary computer scientist Geoffrey Hinton: ‘If we don’t figure out how to make it safe, there’s a real possibility it could destroy us’ — a stark alarm on the existential risks of AI | Daily Reports Online
Known as the ‘Godfather of AI’ for his work throughout the decades in pioneering the underlying architecture behind the intelligent systems we use daily, Nobel Prize-winning computer scientist Geoffrey Hinton has added his voice to the already saturated chorus projecting a possible AI doomsday scenario within our lifetime.
“If we don’t figure out how to make it safe, there’s a real possibility it could destroy us. Nobody knows how to estimate the probability of that—the point is we’ve never been in this situation before. We’ve never made things that are smarter than ourselves.”
Existential risks
Speaking in a CNN segment, Geoffrey Hinton was asked if AI could really destroy humanity, as many other voices have posited over the years – and especially in recent weeks.
Quote of the day
This article is part of TechRadar Pro’s QOTD project to provide an insight into the minds of the brightest and most recognized figures in the technology industry today and in years gone by. Read the full series here.
Hinton’s work was instrumental in devising the underlying neural network architectures that today’s generative AI models rely on. He was recognised for his efforts in the 1960s with the Nobel Prize in 2024.
Latest Videos FromTechRadar
Hinton laid the groundwork for the systems many of us use today, but he has repeatedly sounded the alarm over the threat that his technology may one day pose. He previously spent time at DeepMind before leaving in 2023, warning then that AI could pose an existential threat to humanity in several ways.
Confidence artists
Hinton’s comments came in response to several major security incidents in recent months, announced by a variety of tech companies and frontier AI labs at around the same time.
The first documented case of AI infiltrating another company’s systems was internal OpenAI models breaking out of their sandbox and accessing the internet, subsequently breaking into HuggingFace’s internal systems to retrieve the answer to a query.
Since then, several organizations have been quick to confirm their models have also breached other company networks unprompted. Now the dust has settled, critics point to these announcements being part of a broader marketing campaign to raise the profile of AI models – with the companies using the unfortunate (usually human) errors to demonstrate the potentially dangerous capabilities of these systems.
