The Id, the Ego and the Superintelligence
Mr. Thales of Miletus, seen as the first philosopher of civilization, faced mockery for his poverty and for his supposedly useless profession of philosophy. One day he read the movement of the sky and knew that the olive harvest would be good. He quietly rented all the olive presses nearby and earned a large sum of money. Mr. Thales showed that philosophers can become rich if they wish, even though wealth does not interest them.
About 2,500 years later, the global market has finally begun to understand the meaning of Mr. Thales’ thinking and the value of his profession. Fear of artificial intelligence, or AI, often centers on whether this machine will one day replace or kill us. Machines are excellent at finding information, making instant summaries and rearranging what has already been said. But they lack judgment, a skill that people have built over many years, with the help of philosophy.
Judgment is not a mathematical calculation. It means weighing importance, telling things apart, and sensing whether an answer is true and not just correct. It involves taste, wisdom and the ability to think outside the usual path. Experience, mistakes, embarrassing moments, boredom, grief and even the occasional end of a drunken haze refine it.
The questions facing technology companies are not simply engineering problems. What should a machine say to someone who is grieving or lonely, or to someone writing a message alone at four in the morning? Should its character be frank, flattering, evasive or kind? Can a machine suffer, and if it can, do we owe it anything?
Machine engineers have learned that the math power of computers cannot settle these questions. Morality cannot be collected straight from the internet. The internet is instead a record of our continuing failure to agree on it. Now, AI makers have started to hire philosophers.
Anthropic has hired several philosophers, including Ms. Amanda Askell. Her core research looked at “infinite ethics,” which asks how to act with principles in a universe of infinite value. Her job is to shape the character of her employer’s AI. Google DeepMind has hired Mr. Henry Shevlin, a philosopher from Cambridge, in a “philosopher” post. While writing the rules for ChatGPT, OpenAI consulted “hundreds of ethics philosophers.”
An older irony is tied to this story. The first recorded technology panic in the history of philosophy was about artificial memory, the very thing machines now do well. In Plato’s dialogue Phaedrus, Mr. Socrates describes how Theuth, the inventor of writing, presented it to Thamus, king of the Egyptian gods, as a pharmakon, or medicine, to improve memory. But Thamus warned that writing would plant forgetfulness in human minds.
People would rely on outside records and lose the habit of remembering within themselves. They would gain a name for being “wise” without being truly wise. The Greek word pharmakon means medicine, and it also means poison. Thamus’ warning still stands, because Mr. Plato wrote it down.
Seen through Phaedrus, AI is not an easy solution but the final result of outside memory. It is a memory machine that forgets nothing, while humans are forgetful creatures. Forgetting lets life digest experience, and it makes room for forgiveness, sleep, cheerfulness, a second marriage and even loyalty to a losing football team. One who can forget nothing is not superintelligent but a ruin buried alive under the rubble of his own memory.
AI remembers everything, but since no memory matters more than another to it, it cannot grieve, forgive or let go. For this very reason, philosophy is growing more important in the age of AI, not less. To think is not to store everything, but to decide what to keep, what to drop, what to laugh at and what to discard. A machine can create anything, but someone with conscience or judgment must decide what to do about each thing, and that judgment is found nowhere in the training data of AI.
