[LUM#22] Symbolic AI is based on human reasoning.
Logical and frugal, symbolic AI has not had its final say in the face of machine learning, which is now becoming established in the field of computing. Researchers at Lirmm are betting on the hybridization of these two types of AI.
In the 20th century, AI did not need to call itself symbolic. Born in the 1950s, methods based on high-level abstract representations, known as "logical," dominated AI research. But over the past decade, with the development of computer processing power, the availability of vast amounts of data, and new algorithms, neural AI—referring to neural networks—has taken the lead. From 2012 onwards, historical AI had to differentiate itself from this new form of AI: it would be called symbolic AI, because it is based on reasoning that uses symbols. Marie-Laure Mugnier, computer scientist at Lirmm1, explains the dominant paradigm of this AI: "For a long time, work on AI was based on the assumption that to be intelligent, you have to be able to reason. Researchers rely on the representation of human knowledge in mathematical languages that allow reasoning to be automated. " The researcher gives the example of a simple deductive reasoning process: Socrates is a man, men are mortal, therefore Socrates is mortal. These three assertions are linked by a logical chain that the machine can reproduce.
The first major successes of this AI were in the medical field, where it was used to exploit medical knowledge bases and quickly extract diagnostic information. "The MYCIN expert system was a pioneer; using a knowledge base of around 600 rules modeling a doctor's expertise, this program could identify the bacteria responsible for blood infections, such as meningitis, and recommend antibiotic treatments. However, the first expert systems were fairly empirical, whereas today's knowledge-based systems are rigorously grounded in mathematical theories such as logic and probability." The logical approach is very different from that of deep learning, which relies on complex numerical calculations based on enormous amounts of data. This is a major difference, since symbolic AI is inherently explainable and therefore likely to be understandable to users.
Analogy with the human brain
"Agronomists fromINRAE came to see us at LIRMM precisely to put an end to 'black boxes'. We worked on a project with the ABSys laboratory to develop an AI system capable of helping them design new agroecological systems," says Marie-Laure Mugnier. By exploiting the plant databases built by ecologists and formalizing scientific knowledge about the relationships between plant functional traits and ecosystem services in the form of logical rules, the tool can identify species capable of providing certain ecosystem services. A crucial feature of this tool is that it can justify its results. "In viticulture, we tested in particular the identification of plants capable of fixing nitrogen, improving soil structure, or storing water, which would be of interest for grassing vineyards (read: Integrates data and knowledge to support the
selection of service plant species in agroecology, in Computers and electronics in agriculture, 2024)."
Symbolic AI remains highly effective in many areas today, for example in solving problems modeled in terms of constraint systems. "This could be solving a Sudoku puzzle, but also optimizing an industrial automotive assembly line," the researcher points out. To explain the difference between the two types of AI, she offers an analogy by comparing them to the two systems that make up the human brain, according to Daniel Kahneman, psychologist and 2002 Nobel Prize winner in economics: System 1, which is fast, unconscious, and intuitive, and is used for pattern recognition, is neural AI, while System 2, which is slower, conscious, and explicit, and is used for deduction, is symbolic AI.
Complex, high-level tasks
Neural AI has revolutionized the approach to computing in areas such as image and speech recognition, language translation, and text generation. "But symbolic AI enables complex, high-level tasks to be performed, which are still necessary for decision support, planning, and collective deliberation," points out Marie-Laure Mugnier, who notes in passing that the craze for machine learning applied to everything shows its limitations if the mass of data on which the algorithms run is insufficient: "I see a lot of students who, during internships in companies, develop neural AI tools but on data sets that are too small. And it doesn't work well." This logical approach, which relies on little data, brings another advantage to symbolic AI: its frugality. Because it is the exponential processing of ever-growing masses of data that is responsible for the ecological footprint of AI.
At Lirmm, several research projects are focusing on hybrid AI that combines both approaches. "To take the example of agroecological systems, we could identify the plants that grow naturally on a plot of land using image recognition from the Pl@ntNet app , then use our symbolic AI tool to determine their potential in terms of ecosystem services, " explains Marie-Laure Mugnier.
Find UM podcasts now available on your favorite platform (Spotify, Deezer, Apple Podcasts, Amazon Music, etc.).
- Lirmm (UM, CNRS, Inria, UPVD, UPVM)
↩︎