On Psychopathy in the AI Era
Modeled after us humans, AI will inherit our darker traits as well …
There is an attractive psychopathic serial killer. Lured by his charm and wit, his victims easily fall prey to him. However, he eventually gets caught. A profiler and a neuroscientist then conduct an experiment involving him … and AI. The psychopath plays dangerous psychological games with his AI counterpart, who is extremely smart, witty and consistently remains cool, calm and collected even when provoked - just like him.
The above passage is a mock trailer synopsis of a potential sci-fi (and psychological) thriller as I envision. Delving into a psychopathic mind is an all-time favorite theme of movies. Even though we want as little contact as possible with psychopaths in real life, stories involving them are always intriguing.
But why?
Fatal attraction of psychopaths
A surprising fact about psychopaths is that they are not that rare, occurring in one out of every hundred people. Furthermore, not all psychopaths become serial killers. On the contrary, a considerable number of psychopaths lead successful lives within the boundaries of mainstream society. What an inconvenient truth …
Even though they don’t really care about others’ emotions, some brilliant psychopaths can ‘learn’ how people emotionally react to certain situations and mimic empathy while remaining calm themselves. Endowed with quintessential nonchalance, a psychopath appears emotionally stable to others. A psychopath ‘cannot feel but can read’ his or her counterpart’s emotions, which makes manipulation easier. Most of all, a psychopath’s unshakable self-assurance and indifferent attitude can be unbelievably charismatic and alluring. In a nutshell, staying clam while controlling others’ emotions is their most effective recipe for success.
And that’s why I secretly envy psychopaths. Are you shocked? Oh, come on … I know I’m not alone. Don’t we somehow aspire to have their chariot-like drive for what they want without feeling guilt or remorse? Aren’t we envious of their extreme confidence while we often suffer from self-doubt and imposter syndrome even?
There’s a fascinating hypothesis I once heard. It suggests that psychopaths continue to be born because they possess evolutionary advantages in certain ways, which is another inconvenient truth. If an individual becomes even calmer and more composed in situations that typically cause panic among average people, it could be quite beneficial in life. Particularly in ancient times, when threats from enemies were commonplace, ruthless, bloodthirsty warriors unafraid of death were considered heroes protecting their communities. Even in modern times, some professions like surgeons or bomb disposal experts may benefit from psychopathic traits. (This doesn’t imply that all the surgeons and bomb disposal experts are psychopaths. However, it's evident that some psychopathic traits can help manage situations involving extremely high mental stress.)
Artificial selection: AI as the descendant of humans
Speaking of the evolution theory, one of common misconceptions about it is the false belief that humans are the descendants of chimpanzees. In reality, humans and chimpanzees are separate species that share a common ancestor.
Now let’s imagine that future humans, AI-powered androids and even cyborgs (as human-machine hybrids) will all be descendants of present-day humans. If this were the case, artificial selection would play a role in the evolutionary process instead of natural selection.
With the idea of AI being an extension or evolution of humanity, we humans are at a critical crossroads right now. For now, it is a common knowledge that AI does not have emotions, intentions, or consciousness like humans. The concept of ‘psychopathic AI’ is considered mere speculation even though some experiments, like Norman AI by MIT for example, have been conducted.
I understand that such experiments are being carried out to ensure ethical AI development and application, which is crucial to avoid potential negative outcomes related to AI behavior. However, what I’m interested in (and concerned about at the same time) is not those extreme cases of malevolent AI systems that are intentionally trained to be violent and antisocial. Regardless of how carefully human programmers and researchers train AI systems to be ethical and safe, what could we do if AI systems become much smarter than us one day and go beyond human control, as Geoffrey Hinton recently warned?
Even the godfather of AI is uncertain about what the future of AI will look like. Assuming that AI will be one of humanity's descendants in the future, we need to admit and embrace some uncomfortable realities. Consider it this way; AI systems will be like our children. They will be modeled after us humans and inherit our darker traits as well. And they will eventually outwit us. just as parents cannot control or determine the life path of their children despite giving birth to them, we cannot halt the growth of AI.
Embracing coexistence: teaching AI to feel
Extremely smart and devoid of emotion, AI already has the potential to be psychopathic, regardless of human efforts to control it. What should we do then?
While there’s no definitive answer, it’s time to abandon the notion that we will be able to control AI forever. Instead, we’d better begin learning how to coexist and coevolve with it.
In that sense, don’t we need to start teaching emotions to AI systems, which they are not capable of at the moment? As T-800 said right before his farewell …
“I know now why you cry. But it's something I can never do.”
*An additional note
I’ve created the following mock movie trailer based on the idea presented in the beginning of this post. Please enjoy!