Dear fellow humans,

As I was walking yesterday afternoon, lost in thought (as usual), a simple example came to mind, capturing part of the essence of the problem we face. The average person accepts without a second thought that the sky has 400 billion stars, but when you tell them that the bench is painted, they check with their hand. This same person, I realized, will soon be called upon to judge whether the results given to them by Artificial Intelligence (AI) are correct. And this, in an era when even experts struggle to fully understand how AI arrives at its conclusions.

As I continued my walk, this thought began to unravel. I realized that as AI becomes more and more advanced, this discrepancy between the complexity of the technology and human understanding will widen at an exponential rate. AI systems will arrive at conclusions through processes that will be essentially opaque to the average user, and perhaps even to their creators.

This confronts us with a fundamental dilemma, I pondered. How can we trust and rely on a technology that we cannot fully understand? How can we ensure that AI decisions align with human values and ethical principles when its decision-making processes are a black box to us?

These questions become even more pressing when we realize the speed at which AI is evolving. Each leap in the intelligence of AI systems brings us closer to a reality where human judgment will be insufficient to evaluate the correctness of AI results. Just as the average person blindly trusts scientific statements about the stars, they may soon be forced to blindly trust the conclusions of AI, without the ability to truly verify.

As I neared the end of my walk, a dark thought crossed my mind. Mathematically, it seems we are heading towards an impasse. The complexity of AI is increasing exponentially, while human understanding remains linear at best. If we don’t address this mismatch, we risk losing control of the technology we create.

The solution, I concluded, is not easy, but it is necessary. We must invest significantly in research not only for the development of AI, but also for its understanding and control. We must develop new tools and methods to make AI systems more transparent and interpretable. And we must cultivate a new generation of experts who can bridge the gap between technology and society.

But above all, I realized, we must start an honest and open discussion about the role we want AI to play in our society. We must decide what boundaries we want to set and how we will ensure that AI is used for the benefit of all humanity.

Returning from my walk, one thing was clear: The road ahead is full of challenges, but we have no choice but to walk it. The future of humanity depends on our ability to understand and manage the technology we create. If we don’t, we risk becoming like the person who blindly trusts the stars but doubts the bench – lost in a world we can no longer understand or control.

Sincerely,

Georgios