While listening to a machine learning lecture during the Modu Study AI coaching study, I was struck by a single sentence the instructor used in passing and ended up digging into the ideas behind it.
"Language is the source of misunderstandings"
It is a line attributed to Antoine de Saint-Exupery, who disappeared in 1944 during World War II at the age of forty-four.
What we mean does not always carry the same meaning for other people . Each person has lived a different life, learns in a different way, and develops a different mental model, a different structure of thought, for what they see, hear, and say.
On a very personal level the moment I saw this sentence, the keywords semiotics and ontology came to mind.
In 2024, as humans, we are investing great effort in prompt engineering to communicate better with large language models and generative AI. Before that, we poured similar effort into ICT, communication between information and technology, through tools such as smartphones, mobile devices, and remote control systems.
Amid all of this restless progress, I still feel a deep sense of regret as a human being.
Then what, exactly, is communication for humans? It sometimes feels as though we spend our whole lives working only to spend the money later on hospital bills, or travel far away for a few days of rest after exhausting ourselves in the name of rest itself. It makes me wonder what human beings are doing now, and for what purpose.
Think of something ordinary. Consider the prompt you prepare for ChatGPT in order to get the result you want. Do we offer that same degree of care when we speak to the people beside us?
Or take a planning or UX perspective. Many people are now paying attention to generative interface tools such as Galileo AI. There too, prompts and settings matter. The more clearly you define the persona, intention, and context, the better the result becomes. Then what about people? Have human planners really handed designers only dense wireframes, without kindly explaining the larger background, intention, and expected persona the way we now explain things to generative AI?
Semiotics
In semiotics, a sign is anything that conveys meaning to an interpreter beyond the sign itself. That meaning may be intentional, as with words used to express a specific idea, or unintentional, as with symptoms that indicate a medical condition. Signals can communicate through sight, hearing, touch, smell, or taste. Two major theories explain how signs gain the power to convey information, and both understand a sign through the relationships among several elements.
In the semiotic tradition developed by Ferdinand de Saussure, the sign relation is a binary relation between the sign's form (the signifier) and its meaning (the signified). Saussure argued that this relation is fundamentally arbitrary and motivated only by social convention.
The other major theory, developed by Charles Sanders Peirce, defines a sign as something that stands for something to someone in some capacity, a triadic relation. In this view, a sign is the relation among the sign vehicle (its physical form), the object (the aspect of the world it refers to), and the interpretant (the meaning understood by the interpreter).
According to Peirce, signs can also be divided by the kind of relation that binds them: icons, indexes, and symbols. An icon signifies through resemblance, like a portrait or map. An index signifies through direct connection or causality, like smoke as a sign of fire or a symptom as a sign of illness. A symbol signifies through law or social convention.
Ontology
Ontology is a model that expresses, in a conceptual and computer-processable form, what people see, hear, feel, and think about the world after reaching shared agreement through discussion. It is a formal description that explicitly defines concept types and constraints on their use. As a form of knowledge representation, ontology allows computers to understand concepts and process knowledge. Its purpose is to help programs and people share knowledge by clearly defining and describing the concepts of the resources handled in an information system so that more accurate information can be found. As a tool for the semantic web, it is often expressed in languages such as RDF, OWL, and SWRL.
Because ontology represents agreed-upon knowledge, it is not limited to one individual but consists of concepts accepted by a group. And because programs must be able to understand it, various forms of formalization exist.
