Why Human-AI Management Necessarily Requires Ethical Leadership
Here, I’m going to construct an explicit, public model that is made out of language. This means that if we use the meanings that I describe for the terms, the conclusions follow from them naturally. Why do this? Because an exercise like this can be deeply illuminating, if the meanings map our normal meanings well enough. If our language maps reality at all, then it can help us notice, or remember, what we already know, but have not adequately attended to.
This piece helps us all notice important things that we all already know, but have not adequately attended to at the scale of society.
No Shared Ethos, No Human Group
How do people form distinct, coordinated and enduring groups? Where do nations and tribes and businesses and non-profits and all sorts of other human groups come from?
The decent answers are usually complex, and I don’t intend to provide a complete one here. Or ever, for that matter, at least in this life. I don’t think that’s possible.
Still, one essential component of any group that does anything beyond sporadic activity is this: the agents int he group must share some broader set of commitments. Usually, even a single action involves some broader, shared framework of communication and commitment. This involves at least some kind of shared understanding of what they should do and how they should be together. In a word, it requires an ethos.
For example, imagine that a group paints a fence together. How does something like that happen? It doesn’t happen because a bunch of random people suddenly wake up, converge on a fence for their own arbitrary reasons, and just start painting. That could make for the start to a compelling short story. The genre would probably be fantasy. Or maybe some kind of sci-fi horror?
But in our world, coordinated creation consistently happens because of shared language and communication, some kind of shared reason for acting, and some kind of naturally felt sense that the group is a reliable-enough thing to warrant a commitment of time and effort and resources.
It might well be that the group is a painting company, or a painter’s guild. The group might also be volunteers with a civic group or faith organization or political party. They might be government employees, or a band of neighbors. Whatever the case, we don’t really understand any of these groups and their activity well if we just see them as a bunch of disconnected social atoms that have bounced into each other because someone paid them that day. Even day-laborers, for example, regularly have group norms around where to meet, how to conduct yourselves while waiting for pick-up, some expectation of employers, and some form of community.
Try as hard as possible to imagine the kind of group that might have a minimum viable ethos uniting it, something like the relationship between day-laborers and the people who use them, but even less united by a shared set of norms and expectations. I don’t think it is a coincidence that we also start to imagine some of the most vulnerable and precarious and least coordinated social systems. To be without an ethos is to live a marginal and precarious life. To engage in productive activity with a thin ethos is to do it in ways that are especially fragile and unlikely to endure.
So what is the value of an ethos? It is at the heart of how we make enduring groups of all sorts. Enduring groups are behind almost all of our productive activity, especially anything more socially complicated than digging a hole to curl up in.
On the other hand, groups that form rich and broad and deeply shared ethoses give rise to states and companies and non-profits. Some of these entities, specifically some non-profits, have endured in a recognizable institutional form for thousands of years. How is a shared ethos maintained over time? It doesn’t necessarily depend on a single ethical leader. In fact, healthy ethical systems do not. However, some aspect of leadership must be present in the group and shared widely enough for the ethos that anchors and centers the group to endure.
Human-AI Groups Still Include Humans
If this is how humans work, then groups that are really human in the sense that I am using it will also continue to work this way. So what do I mean by “human-AI groups” here? I am referring to groups in which human agency, and the human way of being in the world experientially, persist. In other words, I’m talking about a situation in which our humanity survives.
We might also imagine a variety of scenarios in which our humanity is fully submerged in, or overtaken by, AI systems. This, too, is the stuff of fantasy and sci-fi horror. However, unlike our imagined short story about a group of people who have mysteriously converged to paint a fence, these sorts of scenarios seem increasingly likely in the medium term.
For example, imagine that we decide that maximizing the relative financial assets of some group is our only goal. In a world where the ability to accumulate money into some group agent is the goal, it is easy to imagine a scenario where a tightly coordinated set of AI agents, of various types, is the group agent most equipped to accumulate the most money. We may even find ourselves in a situation where the richest organization on the planet is an assortment of coordinated AI agents without human oversight. A system like this might employ people for a limited time to learn any novel skills, or to access their social networks. But it always finds ways to absorb that knowledge into its internally coordinated network of generative models, all of them oriented toward the simple goal of accumulating money for the group (of AI agents).
To be clear, for the moment the point of this exercise is not to demonstrate that this can or will happen. I think that it might well happen, but that isn’t the matter at hand here. The point, instead, is to clarify that this is precisely not what I mean by a human-AI management system. We might call this an AI-human management system, in the sense that the AI is managing and leading the process, and using human skills as a disposable and fundamentally replaceable resource (among others) if it uses them at all.
Insofar as humans are governing ourselves, and human agency of some sort is centered within a group, our means of uniting to form group agents through a shared ethos will persist.
Human-AI Management Necessarily Requires Ethical Leadership
Now that it is clear that “Human-AI Management” is being contrasted with “AI-Human Management”, the weight and import of the article’s title should be clear. It should also be clear that it is true definitionally, if we grant the discussion above.
I am not saying that we really ought to persuade people to engage in ethical leadership, because it will produce benefits from a managerial standpoint. To be clear, I also think that this happens to be the case. Let me spend a bit of time on this to clarify my thoughts, and to clarify why they are beside the point of this particular article. Ethical leadership shows up on the balance sheet in many ways, and one of them is in the ability to recruit, retain and motivate teams. This remains true for now, even as everything is in the process of becoming a technological field today. The depth of technological penetration resulting from the new AI-agents into every sphere of productive activity is almost certain to be more intense than the internet’s own socially transformative roll-out. This means that to have a thriving organization today requires work that is far, far more complex than digging a shallow shelter in the dirt.
Still, my points today are not about how to have a thriving organization, but to have one that allows human agency to survive in any meaningful way at all. The point is not that ethical leadership happens to be useful, but that it is necessary for the survival of our humanity. So yes, more deeply and broadly ethical leadership can be seen as an input whose proximate output is motivated, skilled teams. But its ultimate output, or better, its ultimate purpose, is something else. Ethical leadership is also very likely to play a role in encouraging people to engage in kinds of retraining that are emotionally and intellectually complex, and that require a lot of trust, and lot of dedication to a shared ethos. We really shouldn’t trick people into automating themselves out of work, based on promises that can’t be kept. Not only is that wrong, it is also unsustainable: people will figure out what is happening and will work to resist, which will in turn reduce the group’s ability to gain efficiencies. For these reasons, I suspect that it is true that if a group’s ethos stops at money, and an employee’s participation in that stops at the paycheck, it is increasingly likely to fall apart in the near to medium term. The issue is that people who are working for a paycheck cannot be rationally motivated to work themselves out of a paycheck. They need some kind of warranted trust that the organization really does value them beyond the paycheck, or they need to be committed to the mission more than they are to their paycheck.
This means that groups that endure and succeed through this transition, which will require skilled human-AI teams at least in the near term, will need to have teams of people who are even more intensely united around a common mission, a shared ethos. But that is an argument for another day, and it really may be a hopeful kind of speculation on my part.
Here, today, I am highlighting something much more fundamental. A ground truth. A definitional reality. I am not standing here on behalf of “ethical leadership” begging for a seat at the table with the big boys, because it can help you make money, so please let the ethicists in the room. No. That isn’t it at all.
Instead, I am saying that we need to snap out of it. We need to get clear on the fact that money and power may be means, but they cannot be good ends.
We all need to take a giant step back, in the wake of current developments in AI. We need to look as far up as we possibly can, to the biggest and broadest questions of life. We need to come to a hard stop on the illusion that business as usual is enough. We urgently need to ask ourselves why we’re doing what we’re doing, and if we value our lives, the human mode of life, and biological life on this planet at all. Do we care, after all, if the world is dead and gone? Would it all really be the same to us, as long as some entity is accumulating some kind of resource more efficiently than before?
I suspect that for most of us the answer is that we really do think that life matters. If you don’t feel this, and I’m utterly serious about this, you may also find yourself experiencing suicidal ideation now, or soon. If you are considering suicide, please call a suicide hotline. Suicide has deeply scarred my own family system, and I do not bring it up lightly. Here in the United States the number for the national suicide hotline is 988. Please do not hesitate to use it if needed. Even if I do not know you, your life matters to me deeply.
I am emphasizing this because one of the plausible effects of rapid AI roll-outs could be an increase in the suicide rate. We should all be prepared to help walk people through this, as careers are lost and some people struggle to reformulate a basic sense of meaning in the world even as they and their communities try to survive economically. We need to identify a real sense of purpose, rooted in valuing life and experience, and find ways to encourage that in others.
With that, you are equipped to understand what I mean when I say that human-AI management necessarily requires ethical leadership. This doesn’t mean that we should try to have ethical leadership, so that we can manage these systems well. It means that if we don’t have ethical leadership as a part of management, then by definition we have managed ourselves out of being human agents, either because we have managed ourselves out of existence or (at a minimum) because we have eliminated genuine human society. To have organizations is to have at least minimally ethical leadership. And to form enduring and thriving organizations that have positive impacts through the generations, we need especially deeply and widely shared ethical leadership.
