I accidentally created 40,000 chemical weapons when AI experiment backfired, it could kill us all, says expert
WHEN Dr Sean Ekins moved to America to continue his research into life-saving medicine, he never expected to come face-to-face with the President.
The Cleethorpes born pharmacologist and his co-workers reveal in a new Netflix documentary how he worked tirelessly to harness an AI technology that discovered non-toxic molecules to cure diseases.
But when he was asked to deliver a speech to experts about the dangers of AI – a simple test he devised to show off its capabilities went drastically wrong, and he was called into the White House.
Overnight, his supposedly life-saving machinery had come up with ideas for 40,000 chemical weapons, many of them more lethal than the deadly VX and Novichok.
The lethal nerve agents have most notably been used to assassinate Malaysian police chief Kim Jong-Nam and Sergei and Yulia Skripal in Salisbury in 2018.
And all Sean, and his co-worker Dr Fabio Urbina, had to do was – instead of asking the machine to find molecules that could cure diseases – ask the AI to find the most toxic molecules possible.
Computational toxicology expert reveals in new Netflix documentary Unknown: Killer Robots: “There’s a dark side of AI I never thought I would go to.
“I was invited to give a presentation on the potential misuse of AI. I just wanted to make the point ‘Could AI technology to design deadly molecules?’.
“We asked the machine to, instead of drive away from toxicity, drive towards toxicity, and that’s it. We thought it was going to fail.
“The computer did the work, it was cranking through generating thousands of molecules, and we didn’t have to do anything other than just push ‘go’.
“The hairs on the back of my neck literally stood on end when I saw the result. I was blown away. The computer had made tens of thousands of ideas for new chemical weapons.
“Most of these molecules were predicted to be even more deadly than VX. We had opened Pandora’s Box.
“When we were doing this experiment I was thinking, ‘What’s the worst that could possibly happen?’ But now I’m like, ‘We were naive’.
“The thing that terrified me is anyone can do what we did. All it took was the flip of a switch.”
Despite the dangers, Dr Sean and his research lab went on to deliver their findings in a presentation with AI experts before releasing their study to the public in online papers.
While the details of the molecules were hidden, Sean and his team were keen to “warn the world” and “mitigate the misuse of AI” to create chemical weapons.
Before they knew it, hundreds of thousands of people had read about their experiment gone wrong – and weeks later, they got the call from the White House.
Sean adds: “The anticipated consequences of doing that simple experiment went way too far.
“I was invited to the White House. They asked questions about how much computing power we needed, we told them it was nothing special, a standard run of the mill six-year-old Apple Mac.
“That blew them away. The folks that are in change of understanding chemical warfare and governmental agencies, they had no idea of this potential.
“We had got this cookbook to make chemical weapons, and in the hands of a bad actor that has malicious intent, it could be utterly horrifying.”
The molecule expert warns that neural networks such as AI could pose a risk to life as we know it – if we don’t act now to control and monitor how AI is used.
“AI is a dual-edged sword,” the Brit scientist goes on. “It can be used for good, which is what we would use it for, but at the flip of the switch the technology becomes something that could be lethal.
“People have to sit up and listen – the potential for lethality, it’s terrifying.
“If we don’t act now, it’s too late. It may already be too late.”
Dr Sean is just one scientist warning about the immense danger AI poses in the world of warfare in the new Netflix documentary, released to the streaming site today.
He is joined by Emilia Javorsky, director of the Future of Life Institute – a non-profit organisation working to halt catastrophic risks posed by the use of artificial intelligence.
“If it’s not used wisely, it popes a risk to every single person on the planet,” Emilia agrees.
“With AI, we are essentially creating non-human intelligence that is very unpredictable. It also gets more and more powerful – so where are the red lines we are going to draw on how we are going to use it?
“This is the time that we need to act, because the window we have to contain this risk is rapidly closing.
“Once the genie is out of the bottle, it is out – and it is very difficult to put it back in.”
Unknown: Killer Robots is available to stream on Netflix today