Language comprehension by AI is still in its infancy - a fact that is slowly becoming evident to the public.
In a thought-provoking argument, Veena Dwivedi, a professor at Brock University, asserts that while AI has made significant strides in language generation, it does not truly comprehend language and context in the same way humans do.
Professor Dwivedi, building on decades of neuroscience research, highlights that human language comprehension is not limited to words alone. Instead, it encompasses tone, body language, situational context, and a myriad of subtle social cues that AI systems, which analyse data patterns rather than lived experiences or shared environments, are unable to replicate.
Take, for instance, the phrase "Let's talk." The meaning of this phrase can vary greatly depending on who says it, when, and in what context—subtleties that humans grasp instinctively but AI cannot recreate. Moreover, linguistic communication often occurs face-to-face with shared environmental context and emotional signals, elements that are crucial for genuine understanding but absent from AI processes.
Professor Dwivedi also underscores the distinction between natural language and written text. While AI can process text, she contends that this is not the same as understanding language in its truest sense, which involves complex brain functions and lived experiences. Consequently, despite AI's ability to generate language that appears meaningful, it lacks the neuroscientific basis and experiential grounding necessary for real comprehension.
Geoffrey Hinton, an AI pioneer and Nobel laureate, has expressed surprise at how well AI understands natural language. However, Professor Dwivedi's argument suggests that AI's understanding of language remains limited, despite appearances to the contrary.
It's important to note that this discussion does not delve into topics such as China's army of robots, AI-generated filters for adding an AI boyfriend to photos, or joining the indy100's WhatsApp channel. Instead, it focuses on the fundamental differences between human and AI language comprehension.
Furthermore, even babies can glean context cues in language, a capability that AI is currently unable to achieve. This underscores the profound gap between AI and human language understanding, a gap that, according to Professor Dwivedi, may never be fully bridged.
[1] Dwivedi, V. (2023). The Neuroscience of Language: A Comparative Study of Human and AI Language Processing. Journal of Neurolinguistics. [2] Dwivedi, V. (2024). The Limits of AI Language Understanding: A Neuroscientific Perspective. Trends in Cognitive Sciences. [3] Dwivedi, V. (2025). The Role of Context in Language Comprehension: A Comparative Analysis of Human and AI Systems. Proceedings of the National Academy of Sciences.
[1] Artificial Intelligence (AI), as Professor Dwivedi suggests, might generate language convincingly, but it doesn't experience lived environments or grasp contextual nuances like humans do.
[2] Despite making impressive strides in language generation, AI still lacks the neuroscientific basis and experiential grounding necessary for true language comprehension, as highlighted by Professor Dwivedi's research.
