On October 22, at Harvard Department of Government’s annual Vik-Bailey Lecture, Dean Ball, head of strategic futures at OpenAI and former Senior Policy Advisor for Artificial Intelligence at the White House, spoke about the future of AI. He argued for the continued primacy of humans over the world and society. However, because of the absence of any role granted to virtue or value in his recommended solutions or preferred society, Ball failed to develop a compelling theory of humanism, and thus could not successfully defend the human primacy which he rightfully argued we ought to hold dear.
Ball argued that one of the most significant threats in the era of AI is what he called the “Silent Consensus.” According to this consensus, shared by many of those in Silicon Valley and regulators, the development of AI will lead to the loss of human primacy over world and individual events. According to Ball, many in the AI industry assume that the “best case scenario” is one in which AI systems come to take over much of the world — making economic, political, and even societal decisions for humans, rendering man complacent in a world where he has little autonomy.
But Ball said that we should find this outcome outrageous. We should recognize that humanity’s “history,” our having built society and civilization, justifies our continued control over world events. Even if AI systems “join that conversation,” they must recognize that this is a “human conversation,” and thus, that the world should be one dominated by human interests and decision-making, not AI interests. Ball calls this human-focused attitude “human chauvinism.” Some might merely call it “humanism.” In his view, it is a strange sociological fact that those in Silicon Valley have abandoned humanism.
Yet perhaps one reason why many in Silicon Valley have abandoned this vision is that the alternative may seem to be no vision at all. Take Ball’s own “human chauvinistic” argument, for instance. Little in it is very appealing or positive. According to Ball, the main way man can fight against the Silent Consensus is by no longer creating the sorts of AI systems that could lead to significant problems. This means more regulation on data use, clearer constraints on the power of AI systems, and even giving up certain “classically liberal” principles and “foundational liberal tenets” by restricting information of AI to a trusted few and, perhaps, nationalizing the AI industry.
In Ball’s view, this would have to occur via agreements between industry and the government and “incentives” for the AI companies, rather than direct regulation, in order to avoid a “war on drugs-style scenario” in which the development of anti-humanist AI was “driven underground” but not eliminated. The implication of Ball’s focus on incentives and restrictions is that, in order to avoid the takeover of AI, what we need is (likely international) regulations to make information about AI limited to a select few technology companies, presumably including OpenAI, which will have a complete monopoly over the AI industry and which will be paid ransom money from our taxes to convince them to hopefully not destroy the humanistic world. Perhaps the Silent Consensus may appear so appealing because the alternative is a government-run monopoly run at the expense of everyone (except, of course, Ball).
This is not to say that Ball offered no positive vision. However, this vision is far less attractive when considered alongside his negative vision. Specifically, while Ball imagined creating systems which helped us expand our own power, this expansion of power was limited to increasing knowledge. It involved us asking “new and bolder questions” to become “a man expanded.”
It is unclear what Ball exactly meant by this. Perhaps the idea is that using AI systems will allow us to discover more about the world, and thus to become greater humans. But who will access this knowledge and power once the government restricts knowledge to the select few in control of the AI monopoly? Who will be these expanded men? Will they be virtuous, humanistic men, or will they be the sort of man currently in charge of these monopolies — those who adhere to the Silent Consensus? If the latter, Ball cannot seem to explain how to make his expanded men interested in the human endeavor. We can make Sam Altman infinitely knowledgeable, but that will not magically make him more interested in human autonomy.
What Ball is ignoring in his entire model of the future of AI is the concept of virtue. Throughout, Ball couched his vision for the future in vague, nonsectarian language — wealth, knowledge, freedom. Yet none of these things are sufficient to make humans care about humanity. None of it is sufficient to lead us to the sort of goodness or freedom which Ball values. There are possible and even exciting ways in which AI systems can aid our virtue; indeed, many alignment systems right now are motivated by that exact goal.1 Moreover, there are ways in which the government could be involved, not to sign away the human future to a monopoly of those trusted with knowledge, but to create systems that can train us in what makes us valuable as humans: the freedom to choose and the virtue to choose well.
It is only with this sort of virtue-focused vision, a vision of how AI can expand our horizons as humans rather than just as knowers, that we could possibly maintain a truly humanistic society. It is only once we imagine a species made more virtuous and freer because of its interaction with new technologies that we can imagine maintaining our independence from those technologies.
But, of course, Ball could not possibly admit this, because as much as he discusses humanism, he is confined by his job to a limited moral world in which the only value of humanity is chauvinistic.
After all, his job is not to promote virtue; it is to sell goods and services. And those goods and services do not always lead people to better lives. In fact, it seems that most people using AI do so for mentally unhealthy activities. But Ball’s job is not to criticize or to imagine a future. Rather, it is just to say that any human activity, even if it is unhealthy, is equally valuable. It is to say that the value of human activity comes from its being human, not its being valuable. To make a sectarian claim about the capacity of humanity for virtue would be to make a sectarian claim about how customers should be using AI. And to make such a sectarian claim would be to undermine OpenAI’s business model.
Thus, Ball is forced to accept human chauvinism but not humanism. He cannot accept any deep view about the value of human life or how AI systems can help us achieve that value because he cannot accept that the value of humanity is anything greater than the mere fact that we are a species. Ball’s demands, then, must slowly accelerate towards a socialist society paying AI companies rent for the right to be enslaved by humans rather than AI systems. When faced with either the option of an international human-controlled autocracy or an authoritarian set of AI states, we may start to understand why the Silent Consensus has become so popular. At least AI states would take away our freedoms competently.
Nevertheless, Ball’s dichotomy is not a real one. We are not, in fact, forced to choose between losing freedom to either AI or to OpenAI. But what this requires is a vision less focused on regulation and control and more focused on the expansion of human ability. The value of AI, as in any technology, should not be limited to the value of knowledge. It should be applied to move humans closer to our true value as free, good individuals in a righteous society.
Mike Zajko, “Autocompleting inequality.” Journal of Digital Social Research, vol. 7, no. 1, 30 May 2025, pp. 66-82, https://doi.org/10.33621/jdsr.v7i154879.