Oshen Davidson

AI has already proven useful in everyday life, and I believe it could help address some of society’s hardest problems. But the future it creates is not a predictable path toward either progress or catastrophe. The outcomes will depend on what these systems become capable of, how they are built, deployed, and regulated and whether our ability to understand these risks can keep up.

Right now, capabilities are advancing faster than our ability to understand what is happening inside these systems, creating a significant gap in our ability to predict how advanced AI will behave outside controlled environments or when given greater autonomy. Understanding that gap is important, but it is only one part of the work ahead.

As a society, we need all hands on deck to prepare for risks that may be difficult to predict or reverse. That means identifying dangerous behavior early, strengthening safeguards and defenses, developing effective regulations, creating institutions capable of governing advanced AI, and building response plans for when prevention fails. It also means improving public understanding and turning what we learn about AI risk into coordinated preparation and action. There are still not enough people working on these questions, while many more remain unaware of how quickly the capabilities and risks are changing or what is at stake.

I firmly believe humanity will not be able to control advanced AI. We may be able to slow its development, limit access, build safeguards, and reduce the likelihood of harm, but we cannot assume every failure will be predictable or containable. For me, the most important goal is preventing a future in which AI disempowers humanity and weakens our ability to shape society’s direction. That disempowerment could take many forms, including giving powerful actors or hostile states new ways to undermine democratic institutions, concentrate power, destabilize societies, or contribute to wider societal collapse.

The work we do now may determine not only what a future with AI looks like, but whether there is a safe and viable future for humanity at all. I see AI safety as a shared societal responsibility, one that needs people from every field working to preserve humanity’s ability to shape its own future.

If reading this leaves you feeling overwhelmed, just know that you’re not alone. If you want to better understand the risks, learn what AI safety actually means, or figure out how you can contribute, I'd love to chat with you.