“How is artificial intelligence changing the humanities?” This apparently trivial question is more relevant than ever today, in the age of artificial intelligence: AI is transforming society, including the humanities, challenging the long-held notion that understanding is uniquely human.
This blog post, inspired mainly by Professor Claus Beisbart’s presentation at the University of Bern, explores what I believe to be a crucial question: “Does AI truly achieve comprehension, or does it only stimulate human understanding?”
To gain a clear picture of this topic, focusing on AI’s impact on research and analysis from a philosophical perspective, we will address different points guiding us toward a convincing final answer.
Let’s understand what “Understanding” in the Humanities means!
The concept of understanding has been central to many philosophers for centuries, who tried to grasp the various nuances contained in just one simple word. In Aristotle’s Metaphysics, deep understanding of the world combines both the nature of things (the “what”) and the reasons to explain them (the “why”) or in other words, the integration of both theoretical and practical knowledge. Modern philosopher J.D. Trout views understanding as an expression of intellectual satisfaction, which consists of correctly explaining the existing phenomena; while J. G. Droysen focuses on the possibilities of understanding from the perspective of empathy because feelings, which belong primarily to humans, play a key role in understand the reality and the contexts. Finally, L. Wittgenstein differentiates specific categories of understanding:
- Explanatory deals with answering “how” and “why” questions, often linked to scientific inquiry.
- Linguistic examines how the meaning of words is shaped by definitions and contexts.
- Objectual concerns the mastery of specific aspects of reality, focusing on how domains, systems or aspects of our reality function.
How is artificial intelligence changing the Humanities?
Nowadays, AI is rapidly transforming various fields, including the humanities, as seen in Digital Humanities, where AI helps reveal new patterns. As AI continues to evolve, a crucial distinction emerges between:
- Weak AI, which stimulates intelligent human behaviour.
- Strong AI, which is clever and actively reproduces cognitive human processes.
These distinctions highlight two key AI paradigms:
- The symbolic AI approach, also known as GOFAI (Good Old Fashioned Artificial Intelligence), which relies on rule-based procedures and logical operations.
- The connectionist AI approach, like ChatGPT, which learns autonomously (unsupervised) from large datasets, processes complex information, and mimics the human brain’s neural structure with interconnected nodes.
However, this latter can lead to a “black-box” nature, making it difficult for researchers to understand AI’s complex reasoning beyond its conclusions. To address this, Henry Kautz contributed to the development of neuro-symbolic AI, combining symbolic reasoning with data-driven machine learning. An example of this hybrid is Wolfram Alpha , a symbolic reasoning engine launched in May 2009 by Stepehen Wolfram, which could be combined today with a widely known neural network-based system, such as: ChatGPT.
This evolution also enhances AI’s growing ability to mirror human cognitive processes, such as thinking and even speaking. For instance, today Internet of things (IoT), as Alexa, are examples of AI systems that act like humans: they interact, speak and perform tasks just like us (or even better).
In addition, as noted in the book Beyond Quantity: Research with Subsymbolic AI by Sudmann, Echterhölter, Ramsauer, Retowsky, Schroter and Waibel; nowadays, AI is designed to be explainable, meaning that its results remain under human control; even as it becomes more autonomous.
However, according to my view, human-like machines raise a profound question: “Where is the borderland between human and artificial cognition?” … keep scrolling, and we will discover some surprising perspectives on this fascinating boundary together!
The AI Renaissance in XXI century: AI as a tool for research
“An attempt will be made to find how to make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans and improve themselves.“
– Bringsjord & Govindrajulu (2022) following Russels & Norvig (2009)
Furthermore, AI has proven to be an incredibly versatile tool across various fields. Here, as suggested during the presentation, we will focus on three key areas where its impact is particularly groundbreaking:
- Natural sciences: AI can explain complex scientific phenomena and laws (e.g. planetary orbits), following codified rules, equations, paradigms and using algorithms.
- Humanistic research: AI can assist researchers in formulating hypotheses on works of art, history and literary texts.
- Language understanding: AI can effectively translate foreign languages, providing explanations of the meaning of words and producing long and challenging texts.
AI & understanding:
a critical look!
“They (the presuppositions of AI) assume that man must be a device which calculates according to rules on data which take the form of atomic facts two which alone such rules could be applied without the risks of interpretation (…).”
– Dreyfus (1992/1994)
Does AI really “understand” the way humans do? Let’s examine its limits:
- In natural sciences, AI can identify scientific laws and explanations, but proper understanding often relies on intuitions: a quality machines lack.
- In the humanities, AI is effective with well-structured systems but struggles with the complexity of human behaviours, since its theoretical systems are too simplistic. It can’t fully capture emotional depth and empathy that are key to understanding human experience.
- In language, AI manipulates words according to syntactical rules but misses semantic meanings. The “Chinese Room” paradox shows us this: a computer can associate a Chinese semiotic symbol with “banana” but it lacks its qualitative value, such as: the sweetness of it.
AI in the Humanities:
a helping hand
In conclusion, answering the initial philosophical divide, AI truly serves as a valuable complement tool, since it can be better and quicker than humans for specific codifying tasks, such as data analysis and automating repetitive processes. However, it falls short of replicating true humanistic understanding, as it lacks (until now?) the empathetic and qualitative dimensions central to comprehension.
What do you think? Share your thoughts in the comments, I am curious to hear your opinions!
Bibliography:
Slides and notes of the presentation given by Professor Claus Beisbart on the 18.11.2024 at Bern University
“Beyond Quantity: Research with Subsymbolic AI” by Sudmann, Echterhölter, Ramsauer, Retowsky, Schroter, Waibel, Volume 6, AI critique, 2024
“Hey, Siri! Is Artificial Intelligence the Ultimate Oxymoron?” by Richard M. Weber, MBA, CLU, AEP (Distinguished), 2019
“Internet of Things using Node-Red and Alexa” by Rajalakshmi, Anoja, Shahnasser, Hamid, ISCIT, 2017
“Die Maschine und die Geschichtswissenschaft – Der Einfluss von deep learning auf eine disziplin” by T. Hodel, 2022
Internet Literature:
https://nationalhumanitiescenter.org/in-our-image-ai-humanities/
Image Sources:
Video sources:
Video 1: https://nationalhumanitiescenter.org/in-our-image-ai-humanities/
OpenEdition schlägt Ihnen vor, diesen Beitrag wie folgt zu zitieren:
mariapiano (21. Dezember 2024). In Original Reality, Humans Understand; In Virtual Reality, Does AI? Einblicke in die Digital Humanities. Abgerufen am 24. Januar 2025 von https://doi.org/10.58079/12zm1
Eine Antwort auf „In Original Reality, Humans Understand; In Virtual Reality, Does AI?“
Thank you for the interesting blog post!
Following the various stages of your reasoning and arriving at your conclusion, I can say that I completely agree with the idea that AI is a truly valuable complementary tool. It is faster than humans for specific tasks, but it also has its limits, particularly when it comes to replicate true humanistic understanding. I especially appreciate that you conclude by highlighting that AI lacks the empathetic and qualitative dimensions central to comprehension- dimensions that are distinctly human.