That sounds contradictory. AI is obviously technology: it is built through technology, distributed through technology, and accessed through technological products. But calling it a technological revolution may still be one of the most limiting ways to understand it. The description is not wrong. It is simply too small.

It encourages us to place AI beside previous generations of software and digital tools. Organizations approach it as another stage of digital transformation: introduce a platform, train employees, identify use cases, automate a few processes, and measure the time saved. That framing feels sensible because it is familiar. It may also be why we are capturing only a fraction of what AI makes possible.

The electric lights of this transition

To understand the distinction, it helps to think about electricity. When electricity first entered public life, one of its most visible uses was lighting. Candles and gas lamps could be replaced by something cleaner, brighter, and more reliable. From that perspective, electricity looked like a lighting revolution.

That was true, but it missed the point. Electric light was one early application of a much more fundamental capability. Electricity eventually became an underlying layer upon which almost every part of modern life was rebuilt. Someone who understood it only as a better lamp would not have been completely wrong. They would simply have missed almost everything that followed.

I think we may be making a similar mistake with AI. We see a better writing tool, a faster search tool, a meeting summarizer, a chatbot, or a new way to create presentations and reports. These uses are real and often valuable, but they may be the electric lights of this transition: the first visible applications of something whose deeper significance has not yet settled into our institutions.

AI is not only another category of software. It is the beginning of a new layer of accessible machine intelligence.

It is important to stress that this intelligence is still deeply imperfect. It can hallucinate, reproduce bias, misunderstand context, and encourage people to outsource judgment they should protect. Still, something fundamental has changed. People can now interact with systems that engage with complex tasks and respond to intentions rather than only fixed commands. A person can describe an objective, shape it through dialogue, and iteratively refine the outcome.

That is different from opening a conventional software program and selecting a function. It changes the relationship between a person and the work in front of them.

Digital transformation was often based on substitution. A paper file became a digital file. A letter became an email. A physical meeting became a video call. The medium changed, but the underlying activity remained recognizable. AI does not replace one clearly defined thing. It can enter almost every stage of work: understanding a problem, questioning its framing, generating options, structuring information, performing parts of the task, reviewing the result, and preparing what comes next.

This is why lists of AI use cases are both useful and limiting. They give people somewhere to begin. They can show a fundraiser how to prepare donor research, an educator how to adapt material, or a small nonprofit team how to structure years of accumulated knowledge. But use cases can also become walls around the imagination. If people are shown that AI can summarize meetings, improve emails, and create social media posts, they may leave believing that these are the boundaries of the technology.

We do not enter a room and consult a list of electricity use cases. Electricity is an enabling layer that can be drawn upon wherever it becomes useful. AI is not yet mature, reliable, or safe enough to become that invisible, and perhaps it never should be. But the analogy reveals the limitation of our current approach: we are treating a broad capability as a collection of isolated tools.

The technology arrives before the institution is ready

This helps explain a central contradiction of the AI transition. People who have learned to work deeply with these systems can experience a striking expansion in what they are able to attempt. Research that once took days can be accelerated, vague ideas can be structured, communication can cross languages, and small teams can produce work that previously required access to several specialists. The quality still depends heavily on human knowledge, judgment, and discipline, but the person’s radius of possible action can widen considerably.

And yet the measurable gains across many organizations remain surprisingly modest.

This is not an entirely new observation. In 1987, economist Robert Solow famously wrote that “the computer age could be seen everywhere except in the productivity statistics.” More recently, Erik Brynjolfsson and others have described a productivity J-curve: general-purpose technologies can produce disappointing measured gains at first, because their real value depends on complementary investments in people, processes, skills, and organizational redesign.

The technology arrives before the institution learns how to use it.

We are now introducing a new form of capability into organizations designed before it existed. The roles remain the same. The workflows remain the same. The approval structures and expectations remain the same. Employees receive access to an AI tool, perhaps attend a workshop, and then return to an environment built around the old relationship between time, expertise, labor, and output.

Imagine that an employee learns to complete an eight-hour task in five hours. That is a significant personal gain, but the organization may not become any more capable. The employee may keep the method private, particularly if they fear that sharing the time saving will only lead to more work. Their colleagues may never learn from it, their manager may continue assigning work according to old assumptions, and the saved capacity may simply disappear into an already overflowing workload.

This is one form of shadow use: AI use that exists inside an institution without becoming part of its shared capability. An organization can contain increasingly capable individuals without becoming a more capable organization.

From saved hours to mission gains

Individual productivity is not the same as organizational capability, and organizational capability is not automatically the same as greater impact. Saving three hours matters only if the organization knows what those three hours are for.

For a mission-driven organization, the goal cannot simply be to produce more documents, emails, and presentations. The real question is whether new capacity can be translated into the mission. Can it give social workers more time with people instead of paperwork? Can it help educators adapt material for learners who are currently excluded? Can it make knowledge available across languages? Can it strengthen fundraising, improve coordination, help a small team solve difficult problems, or make a program possible that previously sat outside its capacity?

These are not merely productivity gains. They are mission gains.

The value of AI should therefore not be measured only by how many hours it saves, but by what becomes possible because those hours and capabilities now exist. Efficiency asks how the same work can be completed faster. A more serious form of adaptation asks whether the work, the process, or even the ambition of the organization should remain the same.

When execution becomes easier, direction matters more

Reaching that point requires more than learning how to prompt. Prompting and practical use cases matter because people need accessible starting points. But teaching someone to operate an interface is not the same as helping them adapt to what the interface represents.

The deeper shift is behavioral.

Employees who have spent years completing contained tasks may increasingly need to think in wider objectives. They may be able to take an intention, divide it into parts, direct different streams of work, assess what comes back, and continue refining the result. The skills begin to resemble leadership and orchestration: setting direction, communicating context, judging quality, correcting errors, coordinating contributions, and remaining responsible for the outcome.

The essential skill is no longer only knowing how to perform every part of a task personally. It is also knowing how to direct capability.

Deep craft will remain essential. Some people create value precisely because they go deeply into one problem, discipline, or relationship, and turning every specialist into a manager of machine output would not necessarily improve their work. But alongside deep expertise, the ability to form a vision, articulate it clearly, and guide available capability toward it is likely to become a much more general part of working life.

This is what I mean by a more entrepreneurial mentality. I do not mean that every employee must become commercially driven or constantly search for ways to increase output. I mean the agency to recognize a possibility, shape an intention, mobilize the resources available, and carry responsibility for where the work leads.

Music producer Rick Rubin once described his contribution as “the confidence that I have in my taste and my ability to express what I feel.” The context was music, but the distinction reaches further. As execution becomes more accessible, judgment, direction, taste, and the ability to express an intention become more valuable.

Capability cannot remain individual

That shift cannot remain individual. An organization cannot tell people to experiment while punishing mistakes. It cannot ask employees to save time without deciding what will happen to the capacity they create. It cannot encourage initiative while preserving structures that make initiative exhausting. It cannot distribute AI tools while leaving every person to independently decide what is safe, ethical, permitted, and useful.

The organization itself has to adapt.

That means shared language, leadership involvement, responsible-use practices, and clear boundaries around privacy, verification, and human accountability. It means turning individual experimentation into collective learning and reconsidering workflows rather than simply inserting AI into every existing step.

It may also mean changing what is expected and measured. If people can produce more, the answer should not automatically be to demand more. More output is not necessarily more value. A nonprofit has to decide which additional capacity strengthens its purpose and which uses merely create more activity.

This is also where restraint matters. Some things should not be accelerated, some decisions should not be delegated, and some forms of friction are part of careful thinking rather than inefficiencies to remove. AI can extend action, but it cannot decide what action deserves to be extended. That remains a human and institutional responsibility.

The human layer of adaptation

This is where alongside is positioned.

alongside is not built around the belief that mission-driven organizations simply need more AI tools. It is built around the belief that the arrival of accessible machine intelligence requires a human and organizational adaptation layer.

The work begins with literacy, because people cannot direct, question, or refuse systems they do not understand. But literacy alone is not enough. Organizations need to understand where AI can reduce unnecessary burdens, where it can widen capability, where it introduces unacceptable risks, and where the work itself needs to be reconsidered. They need to translate private gains into shared practices and shared capability into greater impact.

The caution of mission-driven institutions should not be dismissed. It often comes from responsibility. Their slowness may sometimes be care, and their resistance may reveal human costs that a purely technological perspective ignores. The task is not to remove those qualities. It is to help organizations distinguish between the parts of their identity that must be protected and the assumptions about work that no longer need to limit them.

AI is not a technological revolution, at least not only. Technology is the form through which it is arriving. The deeper revolution concerns the widening availability of intelligence-like capability: the ability to generate, analyze, structure, communicate, plan, and increasingly act.

If we understand AI only as another tool, we will use it to complete fragments of the existing world more quickly. If we understand it as a new layer beneath work itself, we can ask more important questions. What should a small organization now be capable of attempting? How do individual gains become shared capacity? How can saved time become deeper care, wider access, or stronger programs? What judgment must remain human? What has to change so that greater capability does not simply become greater pressure?

Electricity did not transform the world because people learned to replace candles with bulbs. It transformed the world when we began rebuilding around what electricity made possible.

AI will not reach its deeper potential because organizations learn to write emails faster. It will begin to reach it when they understand that the boundaries around what they can attempt have started to move, and learn how to move those boundaries in service of their mission.

The AI conversation has become hard to trust.

That is probably the most honest place to begin, not because the topic is unimportant, but because so much of the language around it has become difficult to take seriously. Too much of it sounds like a sales pitch, too much of it sounds like panic on one side and hype on the other, and too much of it speaks in a flood of productivity language about people as if they were problems to be optimized, jobs to be automated, or inefficiencies to be removed. A reasonable person learns to stop listening, so I fully understand why many people are tired of these discussions.

That skepticism is not ignorance. Often, it is justified. When a topic becomes this noisy, tuning out can become a way of protecting your own judgment.

But there is also a second reason many people stay out of this conversation: they do not feel entitled to be in it. AI sounds like a technical subject, so people assume it belongs to technical people. If you cannot explain how a model works, who are you to have an opinion about it?

I think that assumption could become very costly.

You do not need to understand the technical mechanics of a system to have a legitimate stake in what it does to your work, your students, your community, your attention, or your future. You do not need to understand the full engineering of a car engine to speak seriously about the consequences of car accidents, because those consequences are human before they are technical. The same is true of AI. A shift of this scale needs engineers, of course, but it also needs psychologists, sociologists, philosophers, teachers, social workers, and people who understand what technology does once it enters real life. The systems may be built by technical people, but what they do to people cannot remain only a technical conversation.

The direction is not decided

Whether we like it or not, AI is becoming part of the environment in which work, education, institutions, communication, and social impact will happen. It will not have the same meaning in every context, it will not always make things better, and it should not be accepted without resistance. Still, it is becoming increasingly present. It is already entering the background of how people and organizations write, learn, organize, search, decide, communicate, and make sense of the world around them, and we are most likely still at the beginning of this process.

That presence may soon become hard to avoid, but the direction is not decided.

This is the distinction I keep coming back to. It may become increasingly difficult to halt AI from becoming part of the systems around us, but it is still very much open how it is distributed, understood, governed, questioned, and used. It is still open who gains power through it, who remains dependent on systems they did not shape, and whether it mostly strengthens those who are already powerful or also becomes a layer of capability for people and institutions working toward human good.

The future of AI will not only be shaped by the technology itself. It will be shaped by the people who understand it, the institutions that integrate it, the values that guide it, and the capacity people have to use it with judgment. A powerful tool does not automatically create a better world. It multiplies what it is attached to. It can multiply extraction, bureaucracy, dependency, shallow content, and existing power. But it can also multiply care, education, translation, access, coordination, social work, public-purpose institutions, and the reach of people already trying to serve others. The most important question is therefore not only what AI can do, but whose work it strengthens.

This is why I do not think the responsible answer is blind adoption. To simply rush toward adoption in every case would be careless. But I also do not think the responsible answer is moral withdrawal. Distance can feel clean, but it does not necessarily protect anything. If the people who care most about education, dignity, justice, health, democracy, and human agency step away from this transition, the transition will not stop. It will simply be shaped without them.

The most responsible thing we can do now is to build literacy, judgment, and responsible implementation, not in the abstract and not only among specialists, but especially among the people and institutions whose work already serves humanity.

I do not mean that as a slogan. I mean it as a way of staying oriented. I am not interested in defending AI as if it were good in itself, and I am also not interested in rejecting it as if refusal alone were enough. What matters most is whether this transition strengthens or weakens human agency. The questions that matter are not only technical; they are human questions. Does AI help people understand more deeply, or does it make them more dependent? Does it help institutions become more capable without losing sight of why they exist? Does it widen participation, or concentrate power? Does it protect judgment, or replace it with fluent output? Does it help the people doing meaningful work, or mainly accelerate those who were already ahead?

These questions lead to a simple conviction: if AI is becoming part of the environment, then literacy is no longer a side issue. It is one of the first conditions of agency. People need to understand these systems well enough to question them, direct them, refuse them, and use them responsibly.

Access is not capability

Before we even get to literacy, though, there is another divide we have to name clearly: access is not equal. We sometimes speak about AI as if it will arrive everywhere at once, in the same form, with the same quality, and with the same usefulness, but that is not how technology spreads. Infrastructure, language support, education, connectivity, devices, and the freedom to experiment safely are not equally distributed.

This matters especially for under-resourced countries and the global majority. If AI becomes a new layer of work, knowledge, administration, education, and public life, then unequal access is not just a technical inconvenience. It becomes a social and political problem. Communities that are already under-resourced could once again be placed in the position of adapting to systems and values built elsewhere, trained elsewhere, governed elsewhere, and priced elsewhere. The first divide is therefore access. But even if access became broader and fairer, the more fundamental problem would not disappear. Access is not the same as capability.

This may be one of the easiest mistakes to make. We assume that once people have a tool, the main problem is solved. But a tool does not create power by existing. It creates power only when people know how to use it with judgment. A saxophone is a beautiful instrument, a marvel of engineering even. And yet in untrained hands, it is mostly noise. The instrument itself may be extraordinary, but its value depends on the person holding it: their practice, their taste, their control, and their ability to know when to play and when silence would be better.

AI is similar. The capability of the system matters, but the capability of the human holding it matters just as much, if not more. A person can have access to AI and still not have agency over it. A student can use it and become less capable, not more. A team can adopt tools and still lack confidence, strategy, ethical clarity, or institutional support.

The divide is therefore layered. There is the divide between those who have access and those who do not. And then there is the divide between those who have access and those who can turn that access into agency.

This is what I mean by literacy. I do not mean prompt engineering, knowing the newest tools, or collecting a technical badge for specialists. I mean the ability to understand what these systems can and cannot do, to ask better questions, to judge outputs critically, to notice bias, to protect sensitive information, to understand limits, to know when not to use the tool, and to never mistake fluency for truth. Without literacy, access can become dependence.

There is also a difference between adoption and adaptation. Adoption is adding tools. Adaptation is building the human and institutional capacity to use them wisely. Access does not automatically become enablement, and individual use does not automatically become organizational maturity. The real work is helping people and institutions integrate AI into their workflows, culture, ethics, judgment, and mission without losing themselves.

Closest to need, furthest from capacity

This is especially urgent for the people and institutions whose work already serves human need. The people closest to human need are most often not the people closest to technological capacity. Teachers, social workers, community workers, local organizers, nonprofit teams, civil-society organizations, health workers, educators, and small mission-driven institutions stand very close to the places where the world needs our collective effort.

If AI becomes part of how work is done, then organizations that cannot build capability will not simply remain where they are. Their staff may use tools informally without guidance. Their leadership may hesitate because the risks are real and still being uncovered. But if they are not enabled, the future will be shaped elsewhere, often by actors with more money, speed, and technical confidence, but not necessarily more responsibility toward the vulnerable.

There is harm in misuse, but there is also loss in missed use. Missed use is the good that does not happen because the right people were not enabled in time. It is the teacher who could have adapted material for different learners but did not know how, the social worker buried in documentation that could have been reduced, and the small organization whose knowledge never becomes useful because no one has the time to structure it.

I want to stress one point: this is not about replacing people. It is about protecting human attention for the work only humans can do. Some things should remain human and some things should remain slow. A student should not outsource the struggle that builds understanding, and an organization should not automate judgment and call it progress. But there are also burdens that do not make the work more human. Administration can consume care, translation barriers can limit access, repetitive reporting can drain small teams, and knowledge can stay unused because no one has the time to structure it. If responsible use can reduce some of that weight, then missed use matters.

We need to ask where and how it can enable the right people to do more of the work they are already here to do. That is why I founded alongside.

Where alongside lives

alongside is not built on the belief that every organization needs more AI. It is built on the belief that humane institutions need the capacity to meet AI with clarity. Its purpose is to stand in the space between concern and capability, between powerful tools and the people who should not be left behind by them, between the noise of the AI conversation and the quieter, harder work of helping humans remain capable.

Our job is to translate possibility into practice. That means listening before prescribing, understanding the mission before choosing the tool, building literacy in plain language, and helping teams decide not only what they can do, but what they should do.

This project started with a question: whose enablement will create the most positive impact for humanity as a whole? That is why alongside strategically sets its focus on mission-driven institutions. They already hold trust, relationships, and responsibility. They already serve communities and work on urgent human problems, so their becoming capable can multiply existing structures of care and action. From there, the intention is to widen the work to frontline professionals such as teachers, social workers, educators, health workers, and community organizers, because they are closest to actual lives and carry trust where abstract strategies cannot go. In the long run, the mission is intended to widen into education itself, because children cannot be expected to simply pick up this technology while remaining fully capable of questioning, directing, and responsibly using it.

Of course, literacy work is only one layer of a much larger puzzle that includes governance, safety, regulation, public infrastructure, and global access. But I believe it is one of the most necessary layers, because even the best policies will fail if people do not understand the systems around them, even the best tools will fail if they land in institutions without capacity, and even the best intentions will fail if they cannot become practice.

This is where alongside lives.

We are at a point in time where we still have a window of opportunity to shape how this is implemented, and we should do everything in our power to ensure the future gets shaped by people who care what it becomes. The people who care about equality, education, democracy, justice, and the long list of global issues we are facing cannot afford to remain outside the transition. The future deserves responsible and enabled hands.

That is why this work matters to me. We do not need to worship the machine, and we do not need to run from it either. We need to look it in the eyes, understand the potential, and make sure the people who care about the future are not the last to become capable inside it.