Here’s an interesting question: “Why should I bother to learn how to spell, do math, reason better, master other languages or do research, when a machine can do all these things for me?”

Technological progress regularly makes human skills redundant.  The mechanized loom made hand weaving obsolete. Back in the mid-1970s, when pocket calculators became widely available, I remember wondering: why I should bother memorizing multiplication tables when a pocket device can quickly do the calculation for me?

We now carry machines that can quickly provide a wealth of historical, scientific, and other knowledge. Phones can perform countless tasks, from translating languages to constructing arguments and spotting logical errors. And, just like calculators and chess computers, they can outperform most of us. Which raises the question—

This question is especially acute for younger people because acquiring knowledge and skills is expensive. In the past, many encouraged the young to view education as a financial investment. That the value and purpose of education is primarily economic is a view shared by many. US Representative Tim Wahlberg says: “I’ve always felt that education is primarily for one purpose: to get people prepared for a job.” But if, with the rise of AI, the jobs have gone, then so has education’s primary purpose. If education is an investment that aims to provide future financial security, it’s an investment that looks increasingly shaky.

The rise of AI also threatens a traditional view of what makes humans special. Many philosophers down the centuries—from Aristotle and Aquinas to Descartes and Kant—argue that what makes us unique is our ability to reason. We are the animal rationale, as the scholastics put it. We humans are able to analyze and process information, draw conclusions from premises, perform calculations, and so on. Mice, cats, whales, and orangutans either can’t do these things, or can do them only to a limited degree.

Our ability to reason has certainly set us apart: allowing us to land on the moon, cure diseases, and figure out the age of the universe. But if our machines can now also reason—if they’re in many ways actually better at it than us—then that traditional account of our supposed specialness is threatened.

Indeed, if the ability to reason is the yardstick by which we’re supposed to measure ourselves, then we humans are starting to look distinctly second-rate.

Why education and human intelligence still matter

I’ve just painted a depressing picture. However, it’s not a very accurate picture.

Perhaps the most obvious problem is the thought that what makes education and intelligence valuable is just their outputs, and especially their economic outputs.

That’s a view of the value of education that has always been challenged. Obviously, education and intelligence are valued in part because they make us more economically productive. However, as someone who teaches at the Department of Lifelong Learning at the University of Oxford, I’m aware that most of those who study with us don’t do it because of what that study might achieve for them economically down the line. People often study History, Art History, Philosophy, and English Literature for no other reason than that they have a passion for those subjects. If they do see some instrumental “output” value to their study, it’s not professional or economic. Rather, it’s that their studies help make them more interesting, well-rounded individuals capable of leading richer lives.

As a result of education, they’re now capable of enjoying a much wider range of experiences. They can interact with texts, art works, other people, and their environment in a much deeper way.

The philosopher John Stuart Mill famously distinguishes “higher” and “lower” pleasures. Lower pleasures include the physical pleasures we get from, say, eating and sex. They’re the sort of thing other species also tend to enjoy. The higher pleasures, on the other hand, are more cerebral, and include the appreciation of art and philosophy.

Higher pleasures, such as appreciation of art or some philosophical insight, often require some education. We need to acquire knowledge and skills in order to develop a more sophisticated palette.

Famously, Mill insisted that enjoying the higher pleasures is actually better than enjoying the lower ones: “It is better to be a human being dissatisfied than a pig satisfied, better to be Socrates dissatisfied than a fool satisfied.”  Some will dismiss Mill’s ranking of pleasures as mere snobbery—perhaps rightly.  But even if Mill is wrong to suppose the pleasure a cultural sophisticate derives from listening to opera is better than the pleasure a child derives from eating chocolate ice-cream, it’s clear that if we’re uneducated and unreflective, we’re going to be left with a very restricted range of possible pleasures.

Suppose kitchen appliances become so sophisticated they can produce a perfect version of any dish we desire. Would we then give up learning how to cook? Of course, if all I value about cooking is a tasty end product, then I probably will. However, for many of us, the journey towards that end product is part of what makes cooking valuable. It’s a journey we can only take once we have acquired the skill ourselves, rather than outsource them to a machine.

Robots will soon be able to play soccer better than any human, running faster than Ronaldo, kicking more accurately than Beckham, strategizing better than Messi. When that happens, will we give up playing soccer? Obviously not, because playing the game ourselves—engaging in the rich social, physical, and mental interactions that playing the game provides—is part of what makes the activity valuable.

Cooking, playing games, and making music are all things we will continue to do even when our machines can outperform us. In the same way, engaging in thinking, understanding, creating and learning are important forms of human flourishing. It’s not just their outputs that are of value—doing them is valuable too.  If we increasingly outsource intellectual activity, we end up diminishing and impoverishing ourselves.

The philosopher Aristotle famously pointed out that to live a good life and flourish we need to develop good character. That in turn involves building good habits—including physical and intellectual habits. Mastering physical and intellectual skills is an excellent way of becoming more self-disciplined, so that we are more resilient to passing distractions and temptations, less prone to hubris, vanity, and overindulgence, more able to approach problems with calm determination rather than panic. Someone who has never engaged in thinking critically and independently is unlikely to develop such virtues and is consequently unlikely to flourish. Sure, we can outsource all tricky physical and intellectual activity to machinery, but, if Aristotle is right, in so doing, we will end up damaging our own characters.

I have been making a case for the value in engaging in intellectual activity, even when machines can outperform us. However, perhaps pointing this out is moot. If a large part of the population has been made professionally and economically redundant by AI and machines, who will pay to enrich their lives through education? Not them, as they lack the spending power.

What about those whose fortunes have increased exponentially by replacing their human workforce with machines? I doubt that they will drain their personal wealth to provide expensive, life-enriching education for all. On the other hand, could AI itself step into the gap and provide meaningful, well-designed educations to billions of humans?

Outsourcing morality to machines

Many religious communities encourage us to derive our moral code from God, or at least his divine representatives. If you want to know what is morally acceptable, you should consult religious scripture, or your pastor, rabbi, Imam, etc. Even secular societies can encourage us to rely on moral decision making from external sources: perhaps our local communist party official, or prevailing cultural sentiments.

With the rise of AI, it will be possible to use AI to help our moral decision making. For example, we could use it to facilitate “effective altruism” by calculating which charity would most efficiently use our donations to save lives. But could we, should we, entirely outsource moral thinking to machines? Can that even be done?

One objection to the suggestion that it is even possible turns on the philosopher David Hume’s famous is/ought gap. Morality concerns what we ought or ought not to do. Investigation of the world around us, on the other hand, reveals only what is and is not the case. But those facts fail to justify conclusions about what we ought or ought not do. Consequently, if Hume is right, investigation of the world around us can never, by itself, allow us to draw a moral conclusion. If AI is restricted to investigating and revealing only how the world is, then it wouldn’t be capable of revealing moral facts.

But perhaps observation of the world is not required for moral knowledge. The philosopher Immanuel Kant thought that reason alone can reveal moral facts. We can establish what’s morally true by simply sitting in our armchairs and having a good think. But if reason alone can establish moral truths, then presumably AI could figure them out too?

Many philosophers reject Kant’s view that reason alone is capable of establishing moral truths. Many also accept Hume’s point that observation of the world around us cannot, by itself, reveal moral facts. If these philosophers are right, then this raises a perplexing question—

Arguably, in terms of providing moral knowledge, the most AI could ever do is figure out how best to achieve what we’ve already decided is morally desirable. For example, if we’ve decided we have a moral obligation to ensure everyone is literate and has access to clean water, AI can help us achieve that. However, it is very unclear whether AI could reveal truths about what we ought morally to aim for in the first place.

There’s a further problem with relying on AI to provide us with moral guidance. Suppose that I, knowing nothing about chemistry, am told by my chemistry professor to flush a large chunk of potassium down the sink. I do so, causing an explosion that injures other students. I’m not to blame. I can rightly excuse myself by pointing out I was only following orders. I could not be expected to know better.

But now suppose someone with moral authority tells me to go out and commit some atrocity, and I do it. I can rightly be held accountable. “I was only obeying orders” is no excuse.

Morality, it turns out, is not like chemistry. We quite reasonably outsource responsibility for making decisions about chemistry, the law, plumbing, or car mechanics, to the experts. At some point, we might also similarly reasonably defer to AI. But the responsibility for moral thinking and judgment can never be completely outsourced—not to political or religious leaders, and not to AI.

We each have a duty to think and make our own moral judgements.

Reason and self-defence

I have raised a number of worries about the outsourcing of thinking to machines. But perhaps the biggest worry is much simpler: if we outsource our thinking to machines, we become vulnerable to exploitation by others. Once we have learned to rely on machines to tell us what to think and do, those controlling the machines can then control us. They can control what we believe and what we value. As a character in Frank Herbert’s novel Dune put it: “Once, men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.”

Stephen Law is a philosopher and author based at the University of Oxford. He researches primarily in the fields of philosophy of religion, philosophy of mind, Ludwig Wittgenstein, and essentialism.