I Am Not a Sheep: Greed, AI Ethics and Human Responsibility

The year was 2020. The month, April. The day, Sunday the 5th. Palm Sunday on the Christian calendar.

That morning, I checked the Covid-19 global tracker. More than a million confirmed cases. Tens of thousands dead. Behind every number, a person; around every person, a world of relationships.

There was a deep silence in my spirit.

What was this time teaching me?

Late the night before, I had watched Dark Waters, the account of lawyer Robert Bilott’s long struggle against DuPont over chemical contamination. That morning, I read David DiSalvo’s Forbes article about the feeding frenzy among buyers and sellers of N95 masks during the pandemic.

One story concerned the poisoning of a community. The other unfolded while people desperately needed protection from a deadly virus. Together, they brought me to a question I could not put aside: How does human suffering become someone else’s business opportunity?

Six years later, I find myself asking it again.

This time, I am looking at the development of artificial intelligence, and at the extraordinary contradiction of people warning that the technology they are building could threaten humanity while continuing to build it.

In May 2023, Sam Altman, Demis Hassabis and Dario Amodei were among the signatories to a statement declaring that reducing the risk of extinction from AI should be a global priority, alongside pandemics and nuclear war.

They were acknowledging a possibility, not predicting a certainty. That distinction matters. But so does the admission itself.

What does it mean to believe that your work could contribute to the end of humanity, and to continue pursuing it? What responsibility follows from recognising a danger of that magnitude? And who has given you the authority to decide how much risk the rest of us should bear?

I hear an echo of the question that troubled me in 2020. How can the pursuit of prosperity, power or success become so compelling that other people’s lives are treated as an acceptable expense?

I do not imagine that every person developing AI is driven by greed. Some are sincerely committed to its benefits and to preventing harm. Anthropic, for example, has published safety protocols explicitly intended to address catastrophic risks as its systems become more capable.

But good intentions and safety commitments do not settle the moral question. They make it necessary to ask what those commitments will require when caution becomes expensive.

When would the people leading this development decide that a capability should remain unbuilt? What would persuade them to surrender an advantage? And would they accept that judgment from people outside their own organisations?

The men we loosely call the “tech bros” are often presented as visionaries: people able to see a future the rest of us cannot yet imagine. There is something deeply familiar in that arrangement. A few know the way. The rest are expected to follow.

We are sheep in need of a shepherd.

That thought surfaced in my original reflection, and I have continued to question it. The image of the sheep and the shepherd was central to my Catholic Christian upbringing. I understand its tenderness: the shepherd protects, seeks the lost and cares for the vulnerable. There is comfort in being loved and guided.

But I wonder what happens when that image becomes a model for our relationship with every form of authority. When being guided becomes being compliant. When trust requires us to put our questions away.

Many of us learnt early that the answer belonged to the teacher. Our task was to learn it, repeat it and demonstrate that we had understood. Questioning could be mistaken for disrespect. Obedience was easier to recognise and reward than independent thought.

There was value in the knowledge we acquired. Yet alongside it, we could also acquire a habit: looking outside ourselves for permission to judge.

Eventually, the teacher gives way to the employer, the political leader, the expert or the person whose wealth appears to confirm that they know what they are doing.

Now, increasingly, it is the person building the system through which we receive our answers.

I work with AI. I see its possibilities in education, creativity and community development. I know what it can help people accomplish.

That makes these questions more personal.

If I recognise its usefulness, am I also willing to examine the conditions under which it is being developed? Can I welcome a capability and still challenge the judgment of those offering it? Can I remain alert to what I am being asked to accept in exchange?

Technical expertise deserves serious attention. But understanding how to build a powerful system does not confer the sole right to decide what risks society should tolerate.

Those decisions concern all of us.

In 2020, I asked what money could possibly mean to someone whose workers were too sick to work, whose customers were dying, whose own life was threatened.

Today, I find myself asking: What does winning the AI race mean if the future you win is one humanity cannot safely inhabit?

And if those leading the race are themselves driven by competition, fear of falling behind, investor expectations or the desire to make history, who is doing the thinking?

Who herds the herders?

We believe we can recognise greed when we see it. I am less certain.

Greed rarely introduces itself honestly. It can arrive dressed as ambition, security, responsibility or progress. It can tell us that we are providing for our children, creating jobs, advancing knowledge or protecting the world from someone less trustworthy.

Sometimes those explanations are true. That is precisely why they need examination.

When does the desire to do well become the conviction that I deserve more than others? When does confidence become entitlement? When do I begin to regard the people affected by my decisions, as obstacles, statistics or collateral damage?

And when does my explanation become so persuasive to me that I can no longer see what my actions cost someone else?

The DuPont story troubled me because it exposed the consequences of failing to act responsibly on information about harm. The US Environmental Protection Agency pursued the company over alleged failures to report substantial risk information concerning PFOA, including information about exposure among people living near its facility.

AI presents different risks and a different kind of uncertainty. Yet I see a moral resemblance whenever those pursuing the benefits reserve the power to decide what danger others must live with.

To me, this is the same greed and disregard for human life that troubled me in 2020: the willingness to let someone else carry the cost of one’s prosperity. The scale of the promise does not remove that responsibility. If anything, it enlarges it.

I must also turn these questions inward.

It would be convenient to locate greed entirely in boardrooms, political offices and technology companies. But the capacity to excuse ourselves is much more widely distributed.

I can become so attached to an outcome that I stop examining how I am pursuing it. I can mistake agreement for understanding. I can accept a reassuring answer because questioning it would inconvenience me.

In the school of the sheep, the answer is given and the search is discouraged. Over time, we may become remarkably skilled at following the thing that drives us without asking where it is taking us.

Sometimes that thing is fear. Sometimes approval. Sometimes greed.

The work of judgment begins when I become willing to examine the driver. What am I seeking? Why does it matter so much to me? What am I refusing to see? Who bears the consequences if I am wrong?

These are uncomfortable questions. They interrupt momentum. They can require us to relinquish something we very much want. Perhaps that is why wisdom has never been as easy to acquire as information.

In Proverbs 3:13–17, wisdom and understanding are described as more valuable than silver, gold or jewels. Wisdom’s paths lead towards peace.

I keep returning to the idea of finding wisdom.

Finding implies a search. Searching requires questions. It asks something of the person who seeks.

I have long been struck by the place of solitude and contemplation in spiritual traditions. People withdraw from the noise, examine their lives and return with an understanding they could not have reached merely by continuing as before.

In 2020, a deadly virus interrupted the world and gave many of us an unwelcome opportunity for that kind of reflection.

Today, I wonder whether we have learnt to protect any space in which it can continue.

Everything urges us to move faster. AI can produce the answer, the plan, the argument and the next ten possibilities. But we still have to decide what deserves to be pursued.

What would it mean for the people developing these systems to practise that reflection? To examine ambition as carefully as capability? To ask whether restraint might sometimes be the greater contribution?

And what would it mean for the rest of us to insist that our future cannot be decided solely by those competing to build it?

The commandments at the centre of my faith call me to love God with my heart, my mind, my soul and my strength, and to love my neighbour as myself.

I cannot reconcile that call with the surrender of my intellect.

My mind is part of what I am asked to bring. So is my will. So is my responsibility towards the person whose life may be affected by my choices.

Love of neighbour must enter the decisions we make about power, profit and progress. It must have some bearing on what we are willing to build, what we are willing to risk and what we are prepared to give up.

In 2020, I wondered what that moment was teaching me.

Today, amid the promises and warnings surrounding AI, I think the question remains open. We have acquired extraordinary tools. We are still responsible for the purposes they serve.

I am called to seek wisdom, to question authority, to examine my own desires and to exercise judgment.

I am not a sheep.

And neither are you.