Maieutics in the Machine: Can AI Make Us Wiser, or Just Lazier?

From a cognitive perspective, the invention of writing (around 3,500-3,000 BCE in Sumeria) was the second massive cognitive revolution that served as a technological upgrade of the first revolution in humankind (the emergence of complex language and symbolic thought - ~70,000 years ago). 

Writing enabled the externalization of memories, making complexity possible (you cannot run a large city, an empire, or a complex economy on oral memory alone. Writing was essential for laws, administration, science or mathematics), fostering new kinds of thoughts (abstraction, linear and analytical thinking) and ultimately creating History. 

However, like for all revolutions, not everyone was enthusiastic about it. One of them was Socrates. As reported by his disciple Plato, Socrates considered that writing was a double-edged invention that presented several major drawbacks for thought and memory (you can find his critique in Plato's dialogue Phaedrus). 

He believed that writing was weakening memory. Socrates feared that writing would make people mentally lazy (true knowledge is that which is “inscribed in the soul of the learner”). 

true knowledge is that which is “inscribed in the soul of the learner" - Socrates

For him, writing was a fixed and dead form of knowledge as a written text is unable to answer questions, defend itself, or adapt to its reader. Writing was also prone to be misinterpreted. Once a thought is written down, it can “tumble about anywhere”. It falls into the hands of anyone, including those who are not prepared to understand it or who might misinterpret it. The author is no longer there to correct misunderstandings or to guide the reader. The text is an orphan, deprived of its “father”. Lastly, Socrates believed that writing offered an illusion of knowledge. Writing does not transmit wisdom, but only the appearance of wisdom. It allows one to accumulate a large amount of information without developing real intelligence or true comprehension. Readers may believe themselves to be very knowledgeable when, for the most part, they are merely ignorant…

So what about now? A few revolutions later, did we address Socrates legitimate fears? 

In this new AI era, we are facing a paradox. Generative AI creates a tool that can simulate the Socratic method of dialogue, making knowledge more accessible and less "dead" than ever before. Yet, in doing so, it offers an unprecedented shortcut around the difficult internal work required for genuine wisdom, potentially encouraging the very cognitive laziness and intellectual vanity that Socrates deplored… 

We can see LLMs as a powerful sophist (a skilled and persuasive rhetorician that can argue any point without possessing true knowledge or belief). Socrates spent his life combating the sophists, so it's safe to say he would be profoundly wary of a world where every person has an infinitely patient, eloquent, and knowledgeable sophist in their pocket! 

But LLMs can be seen as Socratic disciples too! Indeed, if used with the right intention, an LLM can be a partner in the Socratic method rather than a tool for avoiding it.

Socrates believed he didn't teach people answers! He helped them give birth to the ideas already within them through careful questioning.

Socrates believed he didn't teach people answers! He helped them give birth to the ideas already within them through careful questioning. Or more simply the art of being an “intellectual midwife” (this is a core Socratic concept: the maieutics).

Seen this way, an LLM can function as a powerful partner in this process:

Revealing Hidden Connections: An LLM has been trained on a vast corpus of human text, art, and science. By prompting it to synthesize seemingly unrelated fields (e.g., ”What are the connections between quantum mechanics and Eastern philosophy?”), a user can uncover novel patterns and relationships that would be difficult for a single human mind to see. It acts as a mirror to our collective knowledge, revealing the invisible threads that connect different domains.

The Unwearying Partner in Dialogue: A human partner in a Socratic dialogue can get tired, biased, or run out of knowledge. An LLM is available 24/7 to explore endless ”what if” scenarios, act as a sounding board for nascent theories, and provide immediate access to relevant data to support or challenge a line of thought.

Challenging Assumptions: You can explicitly ask an LLM to act as a “devil's advocate”, forcing you to strengthen your own arguments. You can instruct it to find flaws in your reasoning or to provide the strongest possible counter-arguments to your thesis. This is a direct application of the Socratic method, using the tool to refine your own understanding.

An LLM is neither a sophist nor a Socratic disciple by nature. It is a mirror of the user's intent. The Socratic challenge is therefore not in the tool, but in the person using it.

This brings us to the crucial point: An LLM is neither a sophist nor a Socratic disciple by nature. It is a mirror of the user's intent. The Socratic challenge is therefore not in the tool, but in the person using it.

The Sophist's Use: The user asks, “Write me an essay about justice”. The goal is to get a finished product without engaging in the process of thinking. This is using the LLM to create an illusion of knowledge.

The Socratic Use: The user asks, ”Let's explore the concept of justice. Start by giving me the Platonic definition, then contrast it with John Rawls's. Now, let's create a hypothetical scenario to test the limits of both definitions”. The goal here is not an answer, but a journey towards understanding. Moreover, it requires the user to have some pre-knowledge in order to engage in this discussion. 

An LLM is not destined to be a sophist. It can absolutely be a disciple (or better yet) a Socratic partner. It puts the ultimate responsibility back on us. In an age of easy answers, the true challenge is to keep asking the right questions and to have the discipline to use these powerful tools not as a shortcut to an answer, but as a catalyst for our own thinking.