I keep coming back to that distinction.

I am deeply optimistic about automation. I believe it can give us a world with far more abundance: faster scientific progress, cheaper essential services and capabilities that are no longer reserved for large institutions. Small teams can already build things that would have required entire companies a few years ago.

That is extraordinary. It also changes the weakest point in the system.

Until now, most digital security was built around access. Who can open this account? Which device is allowed onto the network? Did the message come from the expected server?

Those questions still matter. But they are no longer enough.

A synthetic voice can issue a convincing instruction. A generated document can look official. A machine can receive an order and act on it before a person has time to question where it came from. The cost of producing something believable is falling quickly.

The cost of proving that it deserves trust is not.

Capability and authority are different things

This sounds obvious, but technology has a habit of blurring the line. A system may be capable of taking an action without having any reliable way to prove that the action was authorized.

That gap is manageable when software produces drafts or recommendations. It becomes much more serious when software touches money, infrastructure, machines or the physical world.

Physical consequences are harder to reverse. You cannot always roll them back, restore a previous version or ask for a second opinion after the fact.

So I think the next important layer of technology will be less about making machines more intelligent and more about making consequential actions verifiable.

Not trusted because a screen displays a green checkmark. Not trusted because one database says so. Verifiable in a way that survives scrutiny.

Trust should not depend on perfect conditions

Most systems work beautifully in a demo. Reality is less polite.

Connections fail. Organizations use different systems. Credentials age. Attackers adapt. A technology that is meant to become infrastructure has to remain useful outside a controlled environment and over a much longer period than the product cycle that created it.

This is also why post-quantum safety matters to me.

Nobody can predict the exact timeline of every computing breakthrough. That is not the point. The point is that important infrastructure often survives for decades. If we know that today’s cryptographic assumptions may not last forever, the responsible time to think about that is before a system becomes difficult to replace.

This is basic first-principles thinking: begin with the lifetime and consequence of what you are building, then work backwards. Do not start with whatever is easiest to ship this quarter.

The goal is confidence, not friction

Security is often presented as a trade-off against convenience. I do not think that is a useful long-term model.

Good infrastructure disappears into normal life. People should not need to become security experts. Businesses should not have to reveal everything about themselves. Useful automation should not stop every few seconds to ask a human for permission.

The better goal is quiet confidence: when an action matters, its authority can be checked. When it does not, the technology stays out of the way.

That is especially important if we genuinely want full automation. More autonomous activity means more decisions happening at machine speed. Human review cannot be the permanent answer to that scale.

Abundance needs a foundation

My optimism about AI and my focus on trust come from the same place.

I want automation to go much further than it has today. I want it to remove scarcity, compress years of work into months and give more people the ability to build. But the more responsibility we give to machines, the more carefully we must separate intelligence from authority.

At 4SI, we are working on that problem space: how digital systems can interact with physical reality with greater confidence. The details matter, and the hard work is in making the idea practical. But the principle is simple.

Intelligence will become abundant.

Trust must become verifiable.