Are People Inherently Good? Will AI Be Good?
The belief or lack of belief in the inherent goodness of people, is often the jumping off point between how different utopian political views form. There is indeed, utopian thinking that springs up from both a “yes” and “no” answer. To those who think the answer to this question is inherently “yes”, it’s common to see a predilection for political anarchism or libertarianism. To those who think the answer is “no”, their utopian visions often include more fascistic or authoritarian arrangements.
If your view is that people are inherently good, then your sentiments will tend towards believing that minimal or no intervention of authority over individuals is morally justifiable. In fact, if everyone is inherently good, then any interference with individual choices would seem to be inherently bad. This makes sense.
Conversely, if your view is that people are inherently bad, then your sentiments are likely to lead you to taking the stance that hierarchy and strong social structures to enforce morality are necessary and good. This also makes sense.
Both of these positions have a certain logical coherence, if you accept these premises about human nature. The problem that lies at the heart of these constructions, is the fact that we’re all taking for granted what “goodness” is. I will contend that it’s almost impossible to define, and is in fact, at the heart of why totalizing political ideologies ultimately always tend towards tyranny.
What does it even mean to say someone is good? It’s one of those things that seems obvious. When you say that someone is a good person, it might evoke traits like generosity, kindness, integrity, or other such character traits. But when you state a claim like “people are inherently good”, as many people do, one must ask the question: if people are inherently good, how does that explain narcissists, sociopaths and sadistic people?
One common explanation that I hear is that such people are victims of bad incentives. That they have been corrupted by malign influences, and if they had been left to their own devices, free from the influence of manipulative and broken institutions, we wouldn’t expect to see people growing into malevolent individuals. Which begs the question: why did inherently good human beings create corrupt institutions to begin with? Why are the elites in society the uniquely bad ones, making other people bad? What is it exactly, that undermines the inherent goodness in people?
I’ve taken a shot in my brain at trying to hypothesize what these things could be. Such as the unintended consequences of well-meaning actions, the corrupting influence of power, or maybe the complexity of large-scale social organization leading to emergent negative properties. But all of these things appear to have obvious counterpoints to them. For instance, unintended consequences are unavoidable by all of us, because none of us have a crystal ball and can see the future. Power structures in society seem inevitable. In fact, all attempts at building egalitarian societies have led to humanitarian disasters. Modern, advanced democratic countries are also the wealthiest, the healthiest and least-violent they’ve ever been in human history. So is the instinct that we can unleash the inherent goodness of people grounded in some historical facts that I am unaware of, from a more peaceful, less violent past, where weaker or non-existent governments, without their corrupting influence unleashed the unmitigated goodness of people?
I’m entertaining these arguments at a surface-level here, with all my culturally-embedded biases of what goodness in a person even means. But we have to take a journey around the world, and look at different cultures, where the standards of what a good person is, don’t exactly line up. Western cultures value individual achievement and self-reliance as good, where as many Eastern cultures prioritize group harmony and filial piety as virtuous. In some cultures, open emotional expression is seen as honest and good, where as in other cultures, stoicism and emotional restraint are signs of maturity and goodness. Some societies emphasize equal distribution as fair and good. Others view merit-based distribution as more just. Some cultures see dominion over nature as progress and good. Others value living in harmony with nature as virtuous. In some cultures, extreme generosity to guests is considered essential. Others may view such behavior as excessive or intrusive. Punctuality is highly valued in some cultures. Others have a more fluid concept of time and don’t see strictness as necessarily good. Some cultures value nuclear families. Others see extended family living as ideal. I could go on. But the point is that what makes a good person is inherently subjective and intersubjective. None of these features appear inherent, since they are not universal.
David Hume, the 18th century Scottish philosopher, offers a perspective — that I’m partial to — that complicates our understanding of human nature and morality. Hume gave birth to moral skepticism, which challenges the very foundations of how we conceive of goodness and moral reasoning.
Central to Hume’s philosophy is the “is-ought problem,” which is also sometimes referred to as Hume’s Guillotine. It asserts that we cannot derive prescriptive (also known as normative statements) — “what ought to be” — from descriptive statements — “what is”. This suggests that even if we could come up with a perfect description of human nature, we would be unable to use that description alone to determine how humans should behave or how society should be structured.
Hume also emphasized the distinction between facts and values, arguing that moral judgements are fundamentally different from factual claims. He proposed that moral judgements are based more on sentiment or feeling than on reason. This view leans towards moral subjectivism, suggesting that moral values are not objective features of the world, but merely projections of human sentiments and attitudes. From this perspective, the question of inherent human goodness, seems to be a suspect claim. It suggests that not only is goodness difficult to define across cultures, as I discussed, but the very concept of moral goodness is a human construct rather than an inherent quality.
This realization doesn’t lead directly to nihilism. But it does mean that the foundations of our moral beliefs and the nature of moral reasoning exist in a purely intersubjective domain.
Emotivism, often associated with philosophers such as A.J. Ayer and C.L. Stevenson, builds on the Humean view, and suggests that moral statements are merely expressions of emotional attitudes. That our emotions, not our reason, are the source of our morality. According to this view, when we say something like “stealing is wrong,” we’re not describing a feature of the act of stealing, but are instead expressing our disapproval of it. We’re saying “boo to stealing!”
This position has profound implications for how we understand moral discourse and disagreement. In the emotivist framework, moral disagreements are not factual disputes that can be resolved through reason or evidence, but rather clashes of attitudes and emotions. When people disagree about a moral issue, they’re expressing conflicting feelings rather than arguing about objective moral truths, which challenges traditional foundationalist notions of moral reasoning and ethical debate. This doesn’t mean that moral discourse is meaningless, but it does suggest that its nature and purpose might be different than commonly assumed.
From the emotivist perspective, the question “are people inherently good?” takes on a different character. Rather than being a factual inquiry about human nature, it might be seen as an expression of one’s emotional attitude towards humanity, and the various political ideologies that stem from answers to this question could be understood as elaborate expressions of fundamental emotional orientations rather than rational constructions based on objective facts about human nature.
This all has deep implications for artificial intelligence, and actually is at the heart of the potential dangers of AI. It’s at the heart of what is known as the Alignment Problem — ensuring that AI systems are aligned with human values. If morality is based on emotions rather than objective facts, the task of encoding these values becomes far more complex. We’re not simply teaching AI to recognize and adhere to universal moral truths, but attempting to replicate and respond to human emotional responses. Once again, the cultural differences of goodness outlined above, raises questions about what emotions or values should be prioritized in AI systems intended for global use. If moral reasoning is ultimately based on emotions rather than logic, how can we program an AI model to make ethical decisions? Traditional approaches to machine ethics often assume that ethical decision-making can be reduced to a set of rules or principles, but emotivism challenges this assumption.
The “is-ought problem” described above becomes particularly relevant here. AI systems are particularly excellent at processing “is” statements — or factual data about the world. But if “ought” statements are fundamentally different, based on emotions rather than facts, how can an AI system bridge this gap? Worse, if moral decisions are ultimately based on emotions, this could complicate efforts to create AI systems with transparent and explainable decision-making processes, especially for ethical choices. The very existence of differing meta-ethical views such as emotivism versus moral realism presents a challenge in the main, as well. Should AI systems be designed based on one particular meta-ethical framework, or somehow account for this uncertainty?
If the emotivist view is true, as I think it is, then this problem is at the heart of the danger we face with the rise of AI. It raises questions about the serious danger to human agency that these systems present. To the extent that we do not even all agree on what human goodness even is, how to define it, and where it comes from, how are we so sure that our AI creations will be good?
