Pandora's AI box is open - should we be afraid?

Is artificial intelligence a threat to humanity, perhaps even leading to our extinction? Several of the world’s most renowned AI scholars and practitioners are raising the warning flag. On Monday, next week, I have the pleasure of welcoming one or the more prominent of these experts, professor Yoshua Bengio, as the guest of honour in a seminar, hosted by the Royal Swedish Academy of Engineering Sciences.

Sign up now for an exciting learning occasion on the most powerful technology we have seen so far.

Ever since humanity started sharing stories, we have created narratives cautioning us against venturing too far into the unknown. Warning us not to release forces we cannot control:

• Adam and Eve tasted the forbidden fruit of knowledge in the Garden of Eden, in the Bible and in the Quran, and is thrown out of Paradise.
• Pandora, in greek mythology, opens a jar left in the care of her spouse and instantly releases a set of curses on mankind.

In the midst of gallopping AI hype,

it is not far-fetched to associate some of the pertinent arguments in today’s hottest debate with the moral of these ancient tales.

The world-renowned author and historian Yuval Noah Harari definitely believes there is, as he pronounces a clear warning in a recently published extract of his new book “A Brief History of Information Networks from the Stone Age to AI” (The Guardian, Aug 24 2024):

"Never summon a power you can’t control."

In the publication, Harari refers to the statement on AI risk, signed by many notable AI experts and scholars in May 2023, and designates AI as “an unprecedented threat to humanity”.

His perhaps strongest cautioning concerns the risk posed by AI creating new life forms, altering not only the evolution of the human species but all life forms.?

Now, a couple of months before this public statement on AI risk, there was another open letter published in March 2023, signed by renowned scholars and CEO’s, requesting a 6 months’ halt on training models more powerful than ChatGPT4.

Harari signed. So did professor Yoshua Bengio of Université de Montréal.

Professor Bengio is considered one of the leading experts on AI, from his pioneering work on deep learning, which earned him the 2018 A.M. Turing Award, an award sometimes referred to as the Nobel Prize of Computing.

Professor Bengio has in several blog posts exposed his detailed concerns, often relating to losing control.

If you would like to dig deeper, here are some reading suggestions:

• “How rogue AIs might arise”.
• “Reasoning through arguments against taking AI safety seriousl.

In this context, it is interesting to note that members of the global AI expertise are divided on the matter. There are more moderate voices.

These specialists are preoccupied with safety, security and creating trustworthy user experiences, definitely. ?Seriously worried about rogue states activities, no question about it. But somehow reassured that the industry will be able to build guardrails for sufficient control, at least against AI-agents going berserk. Some of them believing that the global community will be able to design a regulation, agile enough. The majority hoping that no actor will trigger the doomsday scenario.

So, there is at this point in history no simple answer to the fear question in the headline, not one “truth”.

That is why, I believe, it is crucial that as many as possible partake in the debate. Not just computer scientists and AI experts, but nannys, lawyers, electricians, nurses, civil society activists, shop attendants, artists - and you and me.

Pandora couldn’t stop herself from checking what was in that jar – and it held both curses and blessings. The AI box is wide open and we are already seeing signs of both.

The curses, you have already experienced some of them:

• Accelerating disinformation and propaganda, hate and violence.
• Deep fake scams and election manipulations.
• Manipulative interaction design, influencing our thoughts and actions, eroding our freedom and human rights.

But the box is also releasing blessings:

AI technology offers super-human analytical capabilities, which will lead to new scientific discoveries. Perhaps we will even be able to solve the grand challenges of our time, like climate change or cancer?

Whatever road humanity decides to take, it is of utmost importance that as many as possible participate in the debate.

We must all take interest in the technology, learn, experiment – and voice our opinions.

Do you feel that data and algorithms sound too complicated for you? Think again. It is not about computer science. It is about our common future.

I hope you will be joining us for the seminar next week!

?

?