The limits of Generative AI: A Canadian professor explains

Generative AI has captured global attention with its ability to produce text, images, and even code. Yet, Fenwick McKelvey, a communications professor at Concordia University, Canada, cautions that the technology may not work as well as many expect.

AI’s limitations in context

McKelvey points out that while generative AI can simulate understanding, it does not truly comprehend the content it produces. “These systems can give the impression of intelligence without grasping nuance, context, or cultural subtleties,” he argues. This can result in outputs that are superficially convincing but ultimately flawed or misleading.

Biases in the system

Another concern he raises is the risk of bias. AI models are trained on large datasets sourced from human-generated content, which inevitably contain biases. McKelvey suggests that this can lead to reinforcement of stereotypes, partial perspectives, or misinformation, especially when the AI is applied in sensitive or high-stakes contexts.

Technical and practical hurdles

McKelvey also highlights technical limitations. Generative AI often struggles with long-form reasoning, specialized knowledge, and rapidly changing information. Its outputs may be inconsistent or inaccurate when dealing with complex or novel scenarios, making it unreliable for tasks requiring critical judgment.

Ethical and societal considerations

Beyond technical challenges, McKelvey emphasizes societal and ethical concerns. The widespread adoption of AI could influence public discourse, media, and decision-making in unpredictable ways. He notes that policymakers, institutions, and users must consider these impacts, even if the AI seems impressive on the surface.

Conclusion

While generative AI offers transformative possibilities, Fenwick McKelvey urges caution. According to him, it is not a fully reliable substitute for human judgment, expertise, or contextual understanding. Instead, he presents it as a tool with potential—but one that comes with significant limitations that society must address.