Editor’s note: Besides tracking technological advancements and innovations, our author is a Juilliard-trained musical composer. Listen to “Othering”, an original improvisation by Howard Lieberman, performed for this column.
We usually think of THE OTHER as a person, a people, a nation, a religion, a race, or some other group that is different from our own. But I am beginning to wonder whether that definition is too narrow. Does THE OTHER even have to be human? Could it be an institution, a system, a machine, or a technology? AI is just beginning to occupy territory that once belonged exclusively to people, and this is bound to be frightening to some or even many. That makes me wonder whether, under the right circumstances, a technology could eventually cross a perceptual threshold and begin to feel less like something we use and more like something separate from us.
That perceptual threshold is what interests me. The technology itself may not have to change very much for our relationship to it to change dramatically. A tool can begin as something we
control, then become something we depend upon, then something we consult, and eventually something whose behavior we interpret almost as though it had intentions of its own. And then anthropomorphization begins.
Projection
We have been making new things look familiar for thousands of years. Howard Lieberman created this image with AI help.
Every important technology seems to acquire meanings that go well beyond what it actually does. We project expectations onto it, fears onto it, hopes onto it, and sometimes explanations for changes that were already underway before the technology appeared. But we also do something almost opposite: when something new feels too unfamiliar, we often make it look like something we already understand. There is even a word for this: skeuomorphism. The term was coined more than a century ago, but the behavior is ancient. Pottery could preserve visual characteristics of baskets or earlier materials long after those structural features were no longer necessary. In other words, when the technology changed, people carried some of the old technology’s appearance forward with it. The new became easier to accept because part of the old was still visible.
We did exactly the same thing when computers arrived. Early digital calendars looked like paper calendars. Address books looked like address books. Notepads looked like yellow legal pads. Bookshelves looked like wooden bookshelves. Buttons looked as though they could actually be pushed. None of those visual features were necessary to make the software function. They were there because familiarity reduces the cognitive distance between something we already know and something we have not yet learned. Skeuomorphism is therefore more than a decorative trick. It reveals something about how human beings absorb technological change: we often make the unfamiliar resemble the familiar until we become comfortable enough to let the new thing develop its own identity.
Anthropomorphization may be another version of the same impulse. I know people, young and old, who name their cars, talk to them, praise them when they start on a cold morning, and become annoyed when they do not behave. We have always been willing to attribute personality to objects. Usually, the object does not answer.
AI does. AI participates in the interaction. It answers questions, generates language, makes suggestions, creates images and music, writes software, analyzes information, and increasingly performs tasks that until very recently seemed to require a human mind. That does not make it human, but it makes projection much easier. With earlier technologies, we often made the machine resemble something familiar. With AI, something almost reversed may be beginning to occur: the technology itself behaves in sufficiently familiar ways that we begin to interpret it through human categories.
The more a technology behaves in ways that resemble agency, the easier it becomes for us to attribute agency to it. Once we attribute agency, motives and intentions are not far behind. And eventually we may begin attributing responsibility and identity as well. That does not mean the transition will happen. But it gives me another reason to watch the perceptual threshold between something we regard as a tool and something we begin to experience as something more.
Responsibility
We already do regulate a good many industries to minimize risk. Howard Lieberman created this image with AI help.
If we begin to perceive a technology as something more than a tool, responsibility becomes an especially interesting question. Powerful technologies really do create risks, and recognizing those risks does not require either fearing the technology or celebrating it. It requires understanding it. The greater the capability of a technology, the more important supervision becomes. We need safeguards, testing, accountability, boundaries, and, where appropriate, regulation. This is not unusual. We supervise and regulate all sorts of things that are useful, desirable, and sometimes indispensable. Food is inspected. Pharmaceuticals are tested and regulated. Alcohol is controlled. Automobiles are required to meet safety standards. Airplanes operate within elaborate systems of engineering standards, maintenance requirements, licensing, air-traffic control, and accident investigation. We do not regard these things as enemies because they require supervision. Their usefulness is precisely why we have developed systems for managing their risks. The same principle can apply to AI.
Supervision is not othering. Regulation is not vilification.
That distinction may become increasingly important with AI because the technology is evolving quickly enough that our understanding of its consequences is evolving along with it. Some capabilities will probably prove extremely valuable. Some applications will probably prove unwise. Some things that frighten us now may turn out to be relatively harmless, while problems we have barely considered may eventually become much more important. That is normal when genuinely new capabilities arrive. What makes AI particularly interesting is that, as we begin attributing more agency to the technology, we may also be tempted to attribute more responsibility to it. But technological capability does not make human responsibility disappear.
Suppose automation changes a profession. There is no need to begin by deciding whether the technology is good or bad. We can ask what actually changed. Which tasks disappeared? Which remained? Which new ones became possible? What skills became less valuable and which became more valuable? What did the technology make possible that was impossible before? What new risks appeared along with those possibilities? What should we preserve, what should we change, and what should we learn? These are not questions about finding a villain. They are questions about understanding a changing system well enough to participate intelligently in what happens next. Perhaps that is what responsibility increasingly means in a technological world: not maintaining control over everything but maintaining responsibility for how we respond to what we can no longer completely control.
Truth Costs
Finding out truth is considerably more expensive and difficult than making up a story. Howard Lieberman created this image with AI help.
Unfortunately, understanding is expensive. The truth is more expensive than lying. A simple explanation is inexpensive. It can be communicated quickly, remembered easily, and repeated endlessly. A truthful explanation often requires evidence, context, uncertainty, revision, and the willingness to discover that several apparently contradictory things are simultaneously true. Technology can eliminate jobs and create jobs. It can reduce some forms of human interaction while enabling others. It can concentrate power and distribute capability. It can automate creativity while also placing creative tools into the hands of people who never had access to them before. It can solve problems and create new ones. That is annoying. We would generally prefer things to be good or bad, safe or dangerous, helpful or harmful. Unfortunately, most important technologies refuse to cooperate with those categories.
AI may make this particularly difficult because, as we have already seen, it is unusually easy to perceive agency in something that talks back. Once that happens, complicated systems can acquire deceptively simple explanations. It took the job. It wrote the music. It made the decision. It cannot be trusted. Each sentence may contain some truth, but each can also conceal a much larger system of people, organizations, software, incentives, rules, economic pressures, and individual decisions. Perhaps someday sufficiently autonomous systems will make some of these simple statements considerably more accurate. I do not know. But today the relationships among human beings and the technologies we create remain enormously intertwined. Assigning agency to the technology does not make those relationships disappear.
That is why I am interested in the possibility of technology becoming THE OTHER before I am convinced that it actually has. Once something becomes THE OTHER, understanding it may become more difficult precisely because the simple explanation becomes so attractive. We begin responding not merely to what something does, but to what it represents. Evidence that supports our existing picture becomes easier to notice, while evidence that complicates it becomes easier to dismiss. And complexity is exactly where much of the truth lives. It would therefore be useful to recognize that perceptual transition before it occurs rather than afterward. If understanding is expensive, perhaps noticing the transition early is one way of reducing its cost.
Adaptation
Human and machine are standing together looking toward the future, rather than confronting one another. Howard Lieberman created this image with AI help.
Human and machine are standing together looking toward the future, rather than confronting one another. Howard Lieberman created this image with AI help.
I have spent much of my life around technology and innovation, and one thing has repeatedly surprised me: predicting direction is much easier than predicting timing. I have correctly anticipated technologies that took decades longer to arrive than I expected. Something can clearly be coming while remaining just over the horizon for an astonishingly long time. Then enough incremental changes accumulate, a threshold is crossed, and suddenly the environment really is different. At that point, wishing for the previous environment rarely accomplishes very much. The useful questions become different ones: What is still valuable? What needs to be preserved? What has genuinely become obsolete? What new capabilities should we incorporate? What new risks require safeguards? What do we need to learn that we did not need to know before?
Those are questions of adaptation, and adaptation is one of the oldest capabilities living things possess. Environments change, and organisms, populations, institutions, and cultures either respond to those changes or eventually discover that strategies which once worked no longer work well enough. Human beings have an extraordinary additional advantage: much of our adaptation can be intentional. We do not have to wait for biological evolution. We can learn, invent, reorganize, regulate, educate, preserve, discard, and change direction. Technology itself is one of the mechanisms through which we have been doing this for thousands of years. Adaptation therefore does not mean accepting every new technology enthusiastically, assuming that innovation automatically equals progress, or surrendering human judgment to machines. Quite the opposite. Adaptation requires judgment. We have to decide what to accept, what to reject, what to regulate, what to modify, what to preserve, and what to invent next.
This may ultimately be why I am interested in the possibility of technology becoming THE OTHER. Artificial intelligence may never cross that perceptual threshold. Perhaps the entire idea will prove wrong. But AI has made the question possible in a way that earlier technologies did not because, for the first time, millions of people are routinely interacting with machines that converse, create, advise, analyze, and appear to think. We are therefore adapting not only to what a technology can do but potentially to a changing relationship between people and technology itself. Recognizing that possibility early gives us more choices. We can investigate without vilifying, supervise without fearing, regulate without requiring an enemy, preserve what matters, adopt what helps, reject what does not, and continue changing our minds as we learn more. The future does not require us to love every change. It does require us to respond to the world that actually exists. Understanding that world gives us far more leverage than blaming it. Because whether THE OTHER eventually turns out to be another person, another group, an institution, a machine, or a technology, one principle remains remarkably durable: you cannot blame your way to success.