Plato, AI and humility

Our world often outsmarts us. The way we see the reality is just one perspective, and usually not the most effective. True humility might be one of the best qualities to have. In this article, I was inspired by the argumentation
here, here, here, here, here, here, and here.

Misology

Misology is defined as the hatred of reasoning; the revulsion or distrust of logical debate, argumentation, or the Socratic method. The hatred of reasoning is as old as Socrates, and possibly the first mention of the word appears when Socrates is about to be executed in Plato’s work, Dialogues. Why would anyone hate reasoning? Just as people can often become boring and annoying, so can become pure reasoning. Asking complex questions undermines our confidence and threatens our ego. We cannot solve the philosophical paradoxes, and they threaten our existence.

Two sorts of intelligence

The centipede’s dilemma is a strange and poetic situation when a centipede stops to think how it walks, and cannot walk anymore due to the perplexion. In our daily activities, we strongly rely on the automatic behaviors of the body and the brain. The automatic behaviors are very fast and very energy-efficient. They are often executed by the tightly packed neurons in the cerebellum. Deliberate thinking is much more complex and energy expensive. It is very slow and activates almost all the brain.

Automatic patterns are biased

The automatic behavior is subject to various cognitive distortions since we can change our decision based on totally arbitrary factors, like the temperature of the glass we hold in our hand or exposure to the holy bible. It is not very clear how different senses and experiences mess up our automatic thought processes. The research in convolutional neural network shows that often changing a single pixel in an image can have a strong and unpredictable effect on an entire neural network. Quite possibly our brain has similar sensitivity.

Think more to err less? Maybe not…

We intuitively expect that the more we think about the subject, the less will we err. This is partially true. If we are aware of certain cognitive biases, we can sometimes fight their effects. We see some similar patterns in generative adversarial networks. However, we can get trapped in our own logic. We can start questioning the things that we took for granted to the point that we will not be able to perform the activity. Moreover, the effects of deliberate thinking are grossly overrated. The first 11 seconds or less often define what we think about a person, a game or a video, even a musical masterpiece. The later thinking often serves to find reasons to justify our initial impression, rather than consider additional perspectives and ideas.

Projections

There is a well known allegory in Plato’s Republic. Plato has Socrates describe a group of people who have lived chained to the wall of a cave all of their lives, facing a blank wall. The people watch shadows projected on the wall from objects passing in front of a fire behind them, and give names to these shadows. The shadows are the prisoners’ reality. Plato expected the philosophers to get free of that reality and embrace different perspectives and dimensions. Does it really happen? Depends whom you ask. Philosophers are free to search for the truth until they embrace paradigms. Once a philosopher embraces a paradigm, he will probably reject all pieces of evidence that contradict the paradigm. This is very human, and it happens all the time. Remember Einstein’s quote “God does not play dice with the universe.”? To our best understanding, until proven otherwise, Einstein was wrong. Einstein was possibly the smartest philosophically trained person to walk the earth.

Place the bets

We do not really know about the god and the dice. So it might be reasonable to place bets. Pascal argued that it is best to bet that the god exists since otherwise, the penalty would be infinite. Voltaire is quoted to say “If God did not exist, it would be necessary to invent him.” Occam’s law of parsimony was used both by believers and atheists to support their views. I think it is much easier to place strange, logically sophisticated and amusing bets than simply accept that our knowledge has its limits.

Beyond reasoning

Physicists argue that nobody can truly understand the quantum mechanics and some other fundamental laws of the universe. Philosophers argue that to some extent knowledge is nothing more than a justified belief. Maybe knowledge is a way of formulating thoughts. I quote:

A case from Plato’s Meno offers an intriguing example that cuts across some of the modern categories of cognitive biases. Socrates is attempting to provide an adequate definition of color. He gives his first definition in simple language, and Meno rejects it. Without improving the definition, Socrates rephrases it in grandiose terms, and Meno is instantly satisfied. Socrates explains the change in Meno’s response like this: ‘that’s probably because it’s in the manner to which you are accustomed … And so you prefer it.’

Illusion of confidence

People with limited skills assess their abilities much higher than they truly are. This is called Dunning – Kruger effect, by the names of psychologists that discovered it. Students that have limited achievements tend to overestimate their abilities. Curiously, students with higher abilities tend to deal with more complex subjects, and as a result, underestimate their abilities. Here the humility comes from knowledge and experience. In Apology, Plato relates that Socrates accounts for his seeming wiser than any other person because he does not imagine that he knows what he does not know. The humility to see the limits of one’s knowledge is the super-skill characteristic to Socrates. A question leads to another question until the paradox appears, and there is no apparent way to go beyond the paradox.

Go with your guts

Herbert Simon who got a Nobel prize in physics argues in his work “bounded rationality”: humans are “bounded” by either the limited information they receive or the limited capacity to process vast information, therefore end up “satisficing” their decisions, choosing the best alternative on what they later identify as “gut feeling.” The decision making theories I know include simple computations based on probabilities of different events and estimates of the results of such events. Yet, in real life, we usually do not have the probabilities of the outcomes and the numerical value attached to each outcome. At some point, we start to guess, and eventually, we have no accountability and transparency. Try running a complex convolutional neural network and analyzing why it decided what it did. In the tasks critical to human lives or huge amounts of money, less efficient rule-based systems are used simply because we can understand their reasoning. Trained experts see many examples, and eventually, their gut feeling is more accurate than a rigorous analysis would be. If the performance is more important than accountability, we use our trained intuition.

Be positive within limits

Philosophers acknowledge the limits of our knowledge, while psychologists argue that our beliefs shape our reality. If we lack faith in our ability to tackle certain tasks, we are less likely to succeed. Confidence is essential, even to begin undertaking risky and challenging tasks. While success is never guaranteed and failure is a real possibility, selecting tasks with a reasonable probability of success and either high rewards or minimal penalties for failure can lead to achievement through perseverance. It is important to exercise humility and avoid tasks with very low chances of success, high costs, or severe consequences. Our chances of success are often comparable to those of other equally qualified individuals. It is all too common, especially for mathematicians, to invest a lifetime in a difficult task with little progress to show.

Conclusion

This article is a bit long and complex. I want to summarize my thoughts. Plato knew about the limitations of human knowledge almost as much as we do now. Our knowledge is limited. At some point, we act automatically, trust our guts or place bets. We may feel that we are likely to win, yet we cannot know this. It is great to be positive, but we need some humility when choosing the challenges and trusting our beliefs.