The Compatibility of Large Language Models with Vico’s New Science
The emergence of large language models (LLMs), such as OpenAI’s GPT-4, has revolutionized the way we process, generate, and interpret natural language. These AI-driven models have demonstrated unparalleled proficiency in understanding complex linguistic patterns and delivering contextually relevant responses.
At first glance, one might assume that LLMs, with their data-driven and positivist underpinnings, are fundamentally antithetical to non-positivist approaches like Giambattista Vico’s 300-year old New Science, a ground-breaking philosophical work that offers a unique approach to the study of history and culture. But there’s a strange compatibility between LLMs and Vico’s New Science that might provide us with a way of understanding the new modes of production that AI is creating.
A New Science for a New Age
Vico’s New Science, published in 1725, was an ambitious philosophical treatise that sought to discover the principles governing the development of human societies, cultures, and history. Vico argued that human history unfolds according to a cyclical pattern, with civilizations experiencing periods of growth, stability, and decline that allowed him to imagine historical realities like the development of Bronze Age Greek culture that would not be discovered until more than a century after the publication of his work. Crucially, Vico’s work is underpinned by the belief that human history can only be truly understood by the human mind, as it is a product of human actions, beliefs, and experiences.
In this context, LLMs, such as GPT-4, can be seen as compatible with Vico’s New Science due to their reliance on the collective knowledge, experiences, and perspectives of people to generate language and knowledge. The vast amount of data these models are trained on allows them to capture the nuances of human history and culture, potentially providing new insights into Vico’s theories. Moreover, LLMs can help scholars better understand Vico’s ideas by offering fresh interpretations of his work and revealing connections to other scholarly domains.
From this perspective, the failures of LLMs actually humanize them. A system of knowledge that sometimes lies and hallucinates but still muddles through questions of meaning and fact is something like humanity in a way that modern science, in its austerity, is not.
Model Fit
LLMs are limited by the data and programming they have been trained on. As a result, their understanding of human history and culture may be incomplete, biased, or skewed by the availability and quality of the input data. LLM-generated interpretations based on Vico’s New Science might inadvertently perpetuate historical inaccuracies or overlook critical aspects of his work. It seems hard to imagine ChatGPT, with its constraints (in design, training and censorship), would predict a missing history like Vico had.
LLMs are also inherently deterministic and lack the creative agency that is central to Vico’s conception of human history. His emphasis on the active role of human agents in shaping history and culture may be at odds with the deterministic nature of LLMs, which rely on pre-existing data and patterns to generate output. This raises questions about the extent to which LLMs can truly capture the essence of Vico’s theories.
A Post-Modern LLM?
The compatibility of LLMs with Vico’s New Science presents both opportunities and challenges. As scholars and researchers, our goal should be to harness the power of LLMs to approaches typified by Vico’s work — that is to say, using these LLMs to explore outside the world in which they were built. Such explorations at least allow us to better understand their limitations and address the concerns they raise. This entails engaging in interdisciplinary research that bridges the gap between AI and the humanities, ensuring that LLMs are trained on comprehensive and diverse data, and developing strategies to mitigate potential biases and inaccuracies in LLM-generated insights not only from a fact-based perspective but from a theory-based perspective.
The intersection of LLMs and Vico’s New Science holds promise for deepening our understanding of human history and culture. By collaborating with LLMs in a spirit of critical inquiry, we might uncover new perspectives that may have otherwise remained hidden. In doing so, we can work toward a more comprehensive and balanced exploration of the complex tapestry of human history and culture that Vico sought to reveal through his New Science.