A future we help build

AI agents and robots will replace human work. Whenever I hear this argument, I find myself thinking about how the question is framed. We need to understand what can be replaced. But if we derive the human future solely from what machines can do, the question of how we want to live gets pushed out of the discussion.

I would rather discuss what the future ought to be and work to realize it than predict it and lament the result. I have, in modest ways, turned things I believed in into things that exist. That experience makes me want to take responsibility for more than building individual systems: I want to consider what those systems bring to human lives and society as a whole.

My position is clear. I want the evolution of machines to give people more room to shape their own lives. This requires more than a continuing search for work that only humans can do. It requires a relationship in which individual differences enrich society, and society cultivates the differences that come next.

Distinctiveness and common ground

Each person has a different body, grows up in a different environment, and accumulates different experiences. Even when we encounter the same information, we differ in what we notice, how we interpret it, and what we value. These are differences in how we see the world.

At the same time, as human beings, we share sensations and concepts that let us understand one another. Neither difference alone nor sameness alone can explain human society.

Here, I define distinctiveness as the irreproducible perspective and source of value creation that an individual or organization possesses. I define common ground as a foundation that society can agree upon and share. I understand the progress of civilization as a process in which distinctiveness continually and safely contributes to that common ground.

This contribution does not mean erasing differences or making everyone alike. It means that insight arising from someone's particular experience can become knowledge, technology, or institutions that others can use. That shared foundation, in turn, makes room for new forms of distinctiveness. Civilization advances through this movement in both directions.

This view does not measure a person’s worth by the amount they contribute to society. An idea no one yet understands, or an activity that never becomes useful to others, can still be part of a person’s life. Distinctiveness requires the freedom to share and the freedom to retain something without sharing it. If everything must be surrendered to society, contribution becomes extraction.

The infrastructure that lets differences flourish

Distinctiveness does not begin and end inside an individual. Without room to speak or an opportunity to experiment, an idea can disappear before it reaches society. Education needs to transmit shared knowledge while leaving room for questions that knowledge cannot answer. An economy needs to give a chance to survive to attempts whose value is not yet understood, as well as to things for which demand has already been demonstrated.

A society that suppresses an idea merely because it differs from established practice loses a source of progress. Yet novelty alone cannot justify dismantling the arrangements on which people depend. There is an unavoidable question here: when we speak of contributing safely to common ground, who decides what is safe?

If only those who already hold authority decide, common ground can become another name for preserving the existing order. Nor does majority support, by itself, justify concentrating the burdens of change on a minority. Common ground should provide conditions for living together with different values, rather than require everyone to hold the same values.

I therefore want to distinguish the freedom to hold and express an idea from the authority to apply it to others. An idea needs room to exist even before society agrees with it. When it enters other people’s lives through an institution or a technology, those affected need access to the relevant information and a way to object. We should limit the scope of experiments, revise what can be revised in light of results, and require stronger justification for burdens that cannot be undone.

Recognizing objections does not give one person the power to stop everything. Sometimes a decision must be made without agreement. That makes it necessary to explain whose interests were prioritized and why, and to require those responsible to make corrections when harm or exclusion comes to light. People who did not agree must retain a secure basis for living and speaking after the decision.

Civilization, states, law, economies, education, and technology create these conditions together. Safety is a foundation that lets people participate in change without being silenced by fear of losing their livelihood. That foundation protects distinctiveness; distinctiveness exposes its defects and helps renew it. This is the cycle I want to call progress.

What are machines for?

People have made tools and machines to reduce danger and burdens and to enrich their lives. Looms, locomotives, and household appliances changed the work that human bodies had to perform. Personal computers extended information processing to individuals, while the internet expanded the routes through which people could share knowledge and the results of their work across distance.

Of course, the development of machines has not always made everyone safer or better off. Outcomes depend on who owns them, who can use them, and how benefits and burdens are distributed. We need to consider a technology's capabilities together with the purposes to which they are put.

AI extends this question. As a wider range of work involving information can be entrusted to machines, the amount of processing they perform is only part of what matters. What lives does that capability allow people to lead? What does it give them room to think about and try? If higher productivity merely fills the time it releases with more processing and narrows the ways people can choose to live, it falls short of the richness I am seeking.

What the replacement debate overlooks

When we entrust work to AI, we need to understand what that work depends on. Automation is easier to advance where procedures and evaluation criteria are shared. But I do not want to turn that distinction into a permanent division in which humans create and AI processes.

Suppose AI comes to generate more original ideas than humans, judge more accurately, and perform much of our work. Would that remove the reason for keeping human beings at the center? If human worth depends on superior ability relative to machines, that worth shrinks whenever machines improve. I reject that premise.

Being good at a task is a reason to be entrusted with it. It does not, by itself, confer the authority to decide what others should live for. Among humans, greater ability does not give one person ownership of another person’s life. A comparison of capabilities cannot confer that authority on a machine either.

Treating people as ends does not mean claiming that human judgment is always right. A machine’s proposal can reveal my mistake and change my judgment. Even so, I want the person living a life to retain room to reconsider what matters and to choose a different way to live. Human distinctiveness deserves protection because it is connected to someone’s life, rather than because it can produce a rarer answer than a machine.

Whether future AI should be treated as a subject with experiences and interests of its own is a question we will need to confront separately. We do not need to deny that possibility here. An increase in AI capability alone does not justify turning the lives and self-determination of people who already exist into means of improving machines.

Building the cycle into technology

What I want to build is a technological path through which human distinctiveness can emerge, be preserved, and contribute safely to society’s common ground. AI should serve as a medium that accelerates this cycle.

Imagine a teacher noticing that a student who does not speak during class can express understanding through drawings, and adapting a lesson accordingly. This is a hypothetical example, but it shows the difficulty of turning one person’s observation into knowledge for others. Extracting only the rule that drawings improve grades would discard the student’s circumstances, the relationship with the teacher, and what was actually established in that setting.

Within the limits that those involved consider appropriate to share, AI could help describe the conditions that produced the adaptation, what changed, and what remains unknown. Another teacher could consider whether it fits their classroom, try it on a small scale, and feed back what did not work. What is shared is the part that can help others learn while protecting the student, rather than the whole of the student’s private life.

If this adaptation were imposed on everyone as the single optimal way to teach, sharing would begin to suppress new distinctiveness. If its original conditions and limits remain traceable, and people can try different approaches and revise the shared method itself, one discovery becomes a foundation for another. The same technology can have different effects on civilization depending on which relationship it creates.

This example does not mean that technology alone can solve the problems of education or the economy. Without time to experiment or room to absorb failure, distributing knowledge will not complete the cycle. We need to build the conditions that let people act on knowledge alongside the systems that produce AI outputs.

The systems I want to implement would preserve the conditions behind a result, the limits on what may be shared, the circumstances where it does not apply, and a route for correcting it later. Processing volume alone cannot tell us whether the implementation succeeds. Can an improvement from someone previously unheard reach others? Do the results of its use come back? Can they change the original method? I want to test that entire movement in both directions as an outcome of the technology.

An optimism we put into practice

Human beings came first; machines emerged from human activity. But that historical order alone is not the reason to value people. Each person alive today has a life that cannot be treated merely as a resource to spend on the advancement of machines. I believe we should advance technology to enrich those lives.

AI improving AI is worth pursuing when it serves that purpose. Humans do not need to remain in every step. But I reject the conclusion that, as machines become more capable, we no longer need to ask what people want. The more futures become possible, the more it matters which ones we choose.

I find beauty in people connecting while retaining their distinctiveness and creating something none of them could have made alone. That connection does not require everyone to give the same answer. Our differences let us see what we could not see alone and reach places we could not reach alone. I want machines to extend that possibility.

Optimism about the future means looking at dangers, taking different views seriously, and still acting on the belief that we can build something better. Distinctiveness contributes to common ground; that foundation allows new distinctiveness to emerge. I want the evolution of machines to become a force that broadens this cycle. I want to stand with others on the side of those who build the future, rather than remain among those for whom it is decided.

Back to writing