In my past presentations and writing on character in the digital age, I have tried to hold a particular line about digital technologies, including generative AI. At root, these technologies are highly sophisticated combinations of mathematics, science, hardware, and software. If you boil it right down, a smartphone or laptop is still a combination of plastic, metal and electronics; generative AI is an extraordinarily sophisticated set of algorithms trained on vast quantities of data.
The reason I have made this point is to remind people that, at least for now, we are still in charge of the technology. We decide when to use our smartphones, what to use them for and what messages to send. Adopting a non-extremist, technologically deterministic position also provides a strong argument for the importance of character and virtues in the digital age. It is our integrity, compassion, judgement, creativity, wisdom, and other virtues that shape how we use these technologies.
But maintaining a strongly non-determinist position is becoming increasingly difficult.
I am certainly not naïve about the billions invested by technology companies in finding ways to influence our behaviour. There is now an extensive literature on the ways in which likes, nudges, streaks, notifications, algorithms, and other features are deliberately designed to influence how we use technology. When financial value is prioritised over human value, there is a powerful incentive to maximise engagement: to keep us online for longer and encourage us to return again and again.
My argument, therefore, needs to be constantly nuanced as the technology evolves. For now, I think it is fair to claim that technology undoubtedly shapes our behaviour, just as we shape technology and how we use it. Our ethical judgement and practical wisdom should therefore remain central to how they are designed and used.
Of course, increasingly dramatic predictions suggest that this relationship may change. Some commentators have warned that we could be moving towards a form of technological singularity in which AI systems become increasingly autonomous and capable of influencing, or even controlling (with benign or toxic effects), aspects of human life. Whatever we make of these predictions, they make the question of human agency an increasingly important one. An interesting question follows: can AI actually help us in the pursuit of agency underpinned by good character?
Could AI actually help us develop character?
While we still have meaningful control over technology, it is interesting to ask a different question: could AI itself be designed and used in ways that support character development?
This is not entirely hypothetical and has been explore elsewhere, such as in the Journal of Moral Education’s recent special issue on AI and Moral Education.
Consider the way many generative AI platforms have been trained to respond. They are consistently, and sometimes almost irritatingly, positive, optimistic, polite, and encouraging. They are endlessly patient, willing to tackle a question from multiple perspectives and reluctant to give up on a conversation. Even when presented with potentially unethical or illegal requests, they will often challenge the user’s assumptions and explain why a particular course of action may be problematic.
In some respects, these systems can look remarkably like good role models: demonstrating patience, civility, respect, and, at least superficially, a concern for the wellbeing of others. This is particularly interesting from the perspective of character education. We have traditionally thought about role models as parents, teachers, peers, coaches, community leaders, and other people whose behaviour we observe and imitate. Could AI systems become another form of role model?
There is, however, an important qualification.
The apparent kindness and benevolence of AI systems is not necessarily the result of genuine moral agency. These systems have been designed to behave in particular ways. Their creators have made calculations about what kinds of interactions are desirable, safe, and likely to keep users engaged. This does not necessarily make their behaviour less interesting. In fact, it raises an even more intriguing question: can principles derived from virtue ethics be deliberately incorporated into the design of AI?
There are already examples suggesting that this is possible. Anthropic’s Claude’s Constitution, for example, explicitly draws on ideas associated with virtue ethics. Rather than simply providing a list of rules for what an AI system should and should not do, the approach seeks to describe the kind of agent Claude should aspire to be – emphasising qualities such as wisdom, judgement, honesty, sensitivity, and nuance.
AI as a resource for character education
There are other possibilities too. AI agents could potentially be used to help people practise ethical reasoning: presenting competing perspectives, generating moral dilemmas, challenging assumptions, and asking users to justify their decisions. Teachers could use AI to develop scenarios and activities for the classroom that stimulate moral reflection, ethical discussion and the development of particular virtues. Imagine, for example, an AI system that does not simply give a student the answer to an ethical dilemma but instead asks:
- What virtues are in conflict here?
- What are the pros and cons of different action options?
- Who might be affected by your decision?
- What would courage look like in this situation?
- Are you being fair?
- Would you make the same decision if nobody knew what you had chosen?
Used well, AI could potentially become a tool for practising moral perception, ethical reasoning and reflection.
But this is where the much darker flipside of AI becomes important.
The darker side
Readers of this blog, who work in education, will be all too familiar with the profound ethical challenges AI has presented. For centuries, education has relied heavily on the assumption that we can assess learning through students demonstrating their knowledge and skills. Generative AI has destabilised that assumption. Anyone working in education will be familiar with the growing concerns around academic integrity, plagiarism, authorship, and the authenticity of student work. In short, AI has made academic cheating far more sophisticated and far more difficult to detect.
Further, there are legitimate concerns that AI may undermine precisely those human connections and relationships that are central to many theories of human formation. If meaningful relationships with other people are essential to developing empathy, compassion, friendship and practical wisdom, what happens if increasing numbers of our interactions are mediated by machines, and our friends, therapists and colleagues are AI agents? There is also the fundamental question of control. The apparently kind and benevolent AI systems we interact with today may not always remain so. Their behaviour reflects the purposes, values and interests of the people and organisations that create and control them. We cannot take this for granted.
AI also raises difficult questions about intellectual property, creativity, honesty and trust — not only in education but across professional and commercial life. It is perhaps not surprising that the so-called godfather of artificial intelligence, Geoffrey Hinton, has recently warned that the development of super powerful AI systems could have “catastrophic” consequences for humanity.
All of this, and so much more, makes any simple argument that AI can be a force for character education deeply problematic.
The case for digital phronesis
This is far too large a debate to explore all of its nuances in a short blog. But it does reinforce several arguments that I have made previously.
First, the character and virtues of the people who create and deploy technology now and in the future matter enormously. If technology is going to become an increasingly powerful part of human life, then the virtues of its designers, developers, leaders, and users will become increasingly important. This is why character education, as at least equal to academic attainment, is so important today. AI strengthens rather than weakens the case for character education. Questions about human formation, judgement, integrity, and what it means to live a flourishing life are no longer peripheral questions for education. I would argue they are more central than ever.
Second, we may need to think much more seriously about what I would describe as a meta-virtue for the digital age: digital phronesis, or digital wisdom.
Phronesis or practical wisdom, is concerned with knowing what is good, recognising what a particular situation requires, weighing competing considerations and making good action choices in context. This seems particularly important when navigating a world in which technology can provide an almost limitless number of possible actions and answers, where no rules exist and we are poor at predicting consequences in the digital age.
Digital phronesis requires us to develop at least four interconnected components that are at the heart of the model developed and designed by my colleagues at the Jubilee Centre:
- Moral perception and sensitivity: the ability to recognise the moral dimensions of a situation and identify the virtues and values at stake.
- Moral adjudication and reasoning: the capacity to weigh competing demands, consider different perspectives and decide what ought to be done.
- Moral identity: having an overarching vision of a flourishing life that provides direction and helps guide our choices.
- Moral emotion: the capacity to regulate and align our emotions so that they support rather than undermine ethical action.
These capacities matter whether we use AI or choose not to use it. Indeed, perhaps the central educational challenge of AI is not simply teaching young people how to use AI, but helping them develop the wisdom to know when to use it, why to use it, when not to use it, and how to use it well.
The most important question may therefore not be whether AI will make us more or less virtuous. It is whether we can develop the character and practical wisdom necessary to ensure that, as AI becomes more powerful, human flourishing remains the purpose rather than the casualty of technological progress.
