When we interact with a chatbot, we rarely write in the same way we would speak to a friend or colleague. We tend to be more precise, avoid overly complex phrasing and choose words we believe will be easier for the system to understand.
That adjustment may seem insignificant. However, as the use of artificial intelligence becomes part of everyday life, a broader question emerges: are we merely adapting the way we speak to AI, or are we gradually changing the way we express ourselves more generally?
A theoretical paper published in the scientific journal AI & Society examines precisely that possibility.
According to the researchers, adaptation between humans and artificial intellige nce may work in both directions.
AI systems are designed to communicate in increasingly natural ways, retain information from previous interactions and tailor their responses to individual users. At the same time, however, people may begin shaping their own behaviour in ways that make them easier for algorithms to understand.
When humans adapt to machines
The researchers use the term "robotoid humanness" to describe this potential shift.
They do not argue that people are turning into robots. Rather, they suggest that we may begin adopting forms of communication that are easier for artificial intelligence systems to recognise, classify and process.
The process may emerge gradually through repeated interaction.
A user asks a question, the system responds and, over time, both sides effectively learn from one another.
Artificial intelligence identifies patterns and adjusts its responses accordingly. The human user, meanwhile, learns which forms of expression produce better results and tends to repeat them.
For example, a system may respond more effectively to a short, clear question than to a lengthy narrative filled with hints, contradictions or ambiguity.
If users systematically remove from their language everything that makes communication more difficult for AI, communication may become more efficient, but potentially also more simplified.
The difference from human communication
Adjusting communication according to one's audience is nothing new.
People constantly modify how they speak depending on whom they are speaking to.
There is, however, one important distinction.
A human being may respond unpredictably, disagree, misunderstand or even reveal aspects of our own personality that we had not recognised ourselves.
Artificial intelligence systems operate differently because their responses are based on probabilities and patterns.
They are also often designed to make conversations proceed as smoothly as possible.
The researchers suggest that the risk may not be that AI comes to know us too well. Instead, it may construct a simpler and more limited version of our identity and then reflect that version back to us through personalised responses.
Because that version is practical and efficient in everyday interactions, we may gradually begin to see it as more representative of who we are.
For now, it remains a theory
The researchers themselves stress that this approach does not constitute proof that artificial intelligence has already changed the way people behave or express themselves.
No experiment was conducted, nor were users monitored over a long period.
Instead, the paper presents a theoretical framework that raises questions for future research.
One of those questions is whether systems designed to make life easier could, without users noticing, gradually narrow the range and complexity of human expression.
According to this line of thinking, the challenge is to develop systems that do not require people to become more "machine-like" in order to be understood, but are instead capable of handling the ambiguity, contradictions and complexity that characterise human communication.
