AI is often described as a thinking machine, a digital mind that approaches human intelligence.
But John Nosta, an innovation theorist and founder of innovation and technology think tank NostaLab, said large-scale language models don't think like humans at all.
In fact, he calls AI “anti-intelligence.” That's because AI operates in a way that goes against the way humans reason, learn, and build understanding.
“My conclusion is that artificial intelligence is the opposite of human cognition,” Nosta told Business Insider. “I even call it anti-intellectual.”
AI doesn't understand how humans do things
At the heart of Nosta's argument is the simple but disturbing claim that AI cannot understand anything using human senses.
He says that when people think about an object, say an apple, they place it in space, time, memory, culture, and lived experience.
Nosta says that large-scale language models do no such thing. Instead, it represents words as mathematical objects in a huge hyperdimensional space and searches for statistically matching patterns, he said.
“Apples don't exist as apples,” he said. “It exists as a vector in hyperdimensional space.”
He said the distinction is important because it means the AI's output is optimized for consistency rather than understanding.
He said the system is not logically deriving an answer, but rather generating a response that best fits the patterns of the language.
Why AI is turning human thinking upside down
Nosta believes that AI is quietly reshaping the way people think, especially in the workplace.
According to him, human cognition typically follows a familiar path: confusion, exploration, tentative structure, and finally confidence. The AI will reverse that sequence.
“With AI, we start with structure,” he said. “We start with consistency, fluency, and a sense of completeness, and then we find confidence.”
The reversal creates a powerful illusion. Because AI-generated answers sound sophisticated and authoritative, people often accept them immediately without doing the more difficult work of asking, researching, and fully understanding them, he said.
“The first thing we get to the answer is a reversal of the human cognitive process,” Nosta says. “That's the exact opposite of what humans think.”
The danger of smooth answers
The danger is not that AI will outperform humans in raw computational power. Nosta said that was inevitable. His concern is how easily people can outsource the most valuable parts of their thinking.
“It is through the stumbles, roughness and friction that we arrive at observations and hypotheses that truly develop who we are as people,” he said.
Nosta said speed and fluency are being mistaken for comprehension, as some companies ask employees to leverage AI “totally” for writing, analysis and decision-making.
Using AI as a partner can enhance human thinking. When used as a shortcut, it can be quietly weakening, he said.
“Magic isn't necessarily AI,” he says. “It's an iterative dynamic between humans and machines.”
In Nosta's view, the real risk in the AI era is not that machines get smarter, but that humans learn to think backwards.
growing concern
Concerns about how AI will reshape human thinking are increasingly shared beyond theorists.
In a recent report, researchers at Oxford University Press found that while AI is increasing students' speed and fluency, it is also quietly removing the depth they gain from stopping, asking questions, and thinking independently.
A report from the Work AI Institute released last month echoed the same pattern, saying generative AI often creates the illusion of expertise, making users feel smarter and more productive, even though their underlying skills are diminished.
Mehdi Pallavi, CEO of the International Data Center Authority, which advises companies and governments on building AI-enabled data centers, said excessive and poorly designed AI use is causing “silent cognitive decline.”
“If AI writes better than you and thinks smarter than you, you'll lose confidence in yourself,” he told Business Insider.
