Anyone can sit with an artificial intelligence (AI) program, a poem, to write a children’s story or screenplay, search as a chat. It is supernatural: the results can look quite “human” at first glance. But do not expect anything with very depth or sensory “prosperity”, as researchers explain in a new study.
He found the large language mode that currently power generative AI tools are the way humans do to represent the concept of a flower.
In fact, researchers said that LLM is not very good at representing any ‘cheese’, which has a sensory or motor component – because they launch a body and any organic human experience.
Kihui Joo, the lead author of the study at Ohio State University, US, said, “A big language model can sniff a rose, touch a daisy petals or touch the wildflover’s walk thread. Without those sensory and motor experiences, it can not really represent it as what is all its prosperity.
The study suggests that AI’s poor ability to represent concepts like flowers may explain why they have a lack of human style creativity.
“AI does not have a rich sensory experience, which AI often produces things that satisfy a child of the minimum definition of creativity, but it is hollow and shallow,” said Mark Runco, a cognitive scientist at Southern Oregon University, America, said that the study is not included.
Study was published in the journal Nature human behaviorJune 4, 2025.
Poor in concepts launched among AI poor
The more scientists rehearsed the internal functioning of the AI model, the more they are finding how different their ‘thinking’ is compared to their ‘thinking’. Some people say that AIS are so different that they are like foreign forms of intelligence.
Nevertheless, it is difficult to test the ideological understanding of AI. If computers scientists open an LLM and look inside, they will not necessarily consider what the millions of changing every second are really mean.
Zoo and colleagues targeted to test how well LLMS things could be understood based on sensory characteristics. He did by test that LLM represents words with complex sensory meeanings, measuring factors, discovering what is emotionally stimulating something or you can imagine something mentally, and movement or action-based representation.
For example, he analyzed the extent to which man experiences flowers, or experiences them using tasks from the torso, discovering a petal as touching them. Thesis idea is easy for us to understand, Bens we have intimate knowledge of our nose and body, but it is difficult for LLM, which gives a body a body.
Overall, llms represent the words well – but those words lacquer any relationship for the senses or motor functions that we experience or feel as a human.
But when it comes to the words that we see, taste or interact with using our body, it is the place where AI fails to assure human concepts.
What does ‘A Art Hollow’ mean
AI represents concepts and words by analyzing patterns from a dataset that is used to train it. This idea underlines every algorithm or task, writing a poem, to predict whether you or your neighbor’s image is.
Most LLMs are trained on scrapping text data from the Internet, but some LLMs are still trained on learning scenes from images and videos.
Zoo and Coleguse found that LLM performed some similarity with human representation in visual-related dimensions with visual learning. Those LLMs defeated other LLMs trained on the text only. But this test was limited to visual learning – it excluded other human sensibilities, such as touch or hearing.
This suggests that more sensory information for AI models is obtained as training data, better it can represent aspects.
AI keeps learning and improves
The authors said that LLMs are constantly improving and said that there was a possibility that AI would be better to capture human concepts in future.
Xu said that future LLMs are enhanced with sensor data and robotics, they may be able to actively conclude and function on the physical world.
But independent experts DW spoke to Sugsted that the future of sensory AI was unclear.
Mirco Musolsi, a computer scientist at University College London, UK, said, “It is possible for AI trained on multi -sensory information that multimodal can deal with sensory aspects without anyone.”
However, Runco said with more advanced sensory abilities, AI wants to keep quiet to understand things like flowers completely for humans.
Our human experience and memory is tightly associated with our senses-it is a brain-body conversation that is spread beyond the moment. For example, a rose or it can trigger lustful enthusiasm in your childhood joyous memories or adulthood.
AI programs do not have body, memories or ‘self’. They reduce the ability to experience the world or interact with it as animals and human-enemals, which Runco said, “AI’s creative production wants to be hollow and shallow.”
Edited by: Zulfiker Ebony
