Recent research published in Quanta Magazine reveals a growing consensus among computational neuroscientists and artificial intelligence (AI) researchers: AI, despite its name and inspiration, is fundamentally unlike the human brain—but that’s not a flaw, it’s an opportunity for new frontiers in both technology and neuroscience. This divergence, explored in the article “AI Is Nothing Like a Brain, and That’s OK” (Quanta Magazine, 2025), is now informing efforts to both make AI more efficient and gain deeper understanding of our own minds.
Thai readers, who have witnessed the rapid integration of AI tools in everything from healthcare to language learning, may be surprised to learn just how far apart today’s neural networks and actual brains remain. While AI has become a buzzword in Thai technology and innovation circles, this new research suggests that its powers—so impressive in some respects—are built atop simplifications that don’t begin to capture human cognition, emotion, or memory.
The architecture of the brain is vastly more complex than any AI program yet designed. Human brains have around 86 billion neurons, each a living cell with its own DNA and specialized structures, communicating via a symphony of chemical and electrical signals in a dense web of trillions of connections. These biological neurons can excite or inhibit, communicate analogously (not just in digital binary), and vary in size, shape, and function—even among the same neuronal type. By contrast, the artificial “neurons” powering even the most advanced AI—such as those in large language models or image classifiers—are mathematical abstractions, little more than nodes connected in layers, computing weighted sums and triggering binary outputs.
AI’s origins in neuroscience are real: The first learning algorithm, the “perceptron,” was inspired by the brain’s ability to change with experience (“neurons that fire together, wire together”). But today’s neural networks operate through linear algebra on silicon chips, not the dynamic, self-repairing, energy-efficient pathways sculpted by millions of years of evolution. As computational neuroscientist from the University of Minnesota notes in the report, “We are in the process not of re-creating human biology, but of discovering new routes to intelligence.”
Both neuroscientists and computer scientists agree that the two fields have grown apart, but remain in dialogue. Research teams, including those at Newcastle University and the University of Illinois Urbana-Champaign, are pushing biology-inspired approaches: adding diversity in artificial neuron firing, infusing algorithms with abstractions of neuromodulators, or even wiring real neurons to electronic components. Preliminary studies show that such tweaks can boost learning and efficiency in AI—though the biological complexity can be costly in computational demands.
The brain’s superior energy efficiency remains a major benchmark: An adult brain runs on about 20 watts (comparable to a dim LED bulb). No artificial network yet invented approaches this frugality; deep learning models consume orders of magnitude more electricity. As a researcher at IBM highlights, debates continue about whether AI and brains are even comparable—are we measuring against a lifetime of experience, or an entire species’ evolution?
Nonetheless, modern neural networks, especially in language and vision, can already outperform humans at certain tasks—playing Go, classifying images, or predicting protein folding. Still, as leading neuroscientists quoted in the article stress, AI “does not understand anything.” Neural networks excel at recognizing statistical patterns in immense datasets, but lack the kind of environmental interaction and embodiment that makes a maggot—or a human—knowledgeable in a meaningful sense.
For Thailand, where AI is increasingly used in education, healthcare, and content creation, this has profound implications. The Ministry of Higher Education, Science, Research and Innovation has championed AI literacy among students and medical professionals (source), yet these findings caution against equating AI “intelligence” with human thinking, empathy, or ethical reasoning. Thai AI startups, often backed by university partnerships, are currently working on language models tailored to the Thai script and medical image classifiers for regional hospitals. Recognizing the strengths and the crucial limitations of current AI systems will help guide realistic expectations and responsible deployment.
Culturally, the Thai approach to science often emphasizes harmony and adaptation. In that context, the article’s message—that studying both the contrasts and the occasional similarities between biological and artificial cognition can help both fields—resonates. Already, AI is accelerating neuroscience, helping researchers analyze protein structures, genomic data, and even offering new tools for studying thought and perception. As neuroscience fellow from the Flatiron Institute points out, we are entering an “exciting time” when models learned from AI can be used to better understand the biological basis of the mind itself.
Globally, researchers at MIT, IBM, and other top institutions are now exploring questions of “universality of representation”—whether, despite different architectures, both brains and AI can arrive at similar ways of processing information. The field is still rife with “black boxes,” as both AI and the brain are hard to interrogate from the outside. But as a systems neurobiologist from the University of Sydney says, “It’s actually more sensible to think of them as a really different information-processing object—one that’s extremely interesting in its own right.”
For Thai readers, the future of AI will depend on understanding both its powers and its limits. As this new research suggests, AI may never become a replica of the human brain, but that’s all right. By learning from both, we can develop smarter technologies while gaining ever-deeper insight into the amazing complexity of our own minds. For Thai educators and policymakers, this means continuing investment in both neuroscience and computer science, fostering dialogue between the disciplines, and ensuring that ethical, human-centered values remain at the core of the country’s technological innovation.
Recommended actions for Thai readers would include seeking out cross-disciplinary education in neuroscience and AI, supporting local research that bridges these fields, and fostering informed discussions about the promises and the ethical limits of artificial intelligence. Recognizing that intelligence, whether artificial or biological, takes many forms, will serve Thai society well as it navigates a future increasingly shaped by both.
Sources: