Brain science set to shape the future of AI
The future of artificial intelligence may depend less on better chatbots – and more on understanding how the human brain works.
CQUniversity educational neuroscience expert Professor Ken Purnell says the next generation of AI is likely to be shaped by human intelligence, not just language processing.
“Today's chatbots are remarkably capable in some tasks, but they remain brittle, unreliable and easy to overestimate,” Professor Purnell said.
“They can sound authoritative while still being wrong, and that creates real problems for learners, teachers and institutions.”
Building AI to work with us
His new paper, Beyond the Chatbot: What Brain Science Tells Us About AI’s Future, explores what may come next in artificial intelligence.
Professor Purnell says understanding how the brain integrates memory, attention, emotion and decision-making will be critical to building AI that works with humans, rather than against them.
“Human intelligence isn’t just about information – it’s about how memory, attention, emotion and decision-making all work together,” he said.
“If AI systems ignore that, they risk becoming powerful but poorly aligned with how people actually learn and think.”
The next wave of AI will look very different
While chatbots dominate headlines, Professor Purnell says the next phase of AI is likely to move beyond language-based systems.
Instead of simply predicting text, future systems may:
• build internal models of how the world works
• combine memory, perception and reasoning
• behave more like human thinking systems.
This shift is already being explored by leading AI researchers, including Yann LeCun, who argues current chatbot-style systems will not lead to true intelligence.
At the same time, large-scale brain research led by Evian Gordon is helping map how human intelligence works – offering a potential blueprint for the next generation of AI.
Professor Purnell says the real challenge is not just building more advanced AI, but ensuring it supports human thinking rather than replacing it.
“The question isn’t just how intelligent these systems become,” he said.
“It’s whether they help us think better – or slowly make us think less.”
