According to Wiggers, AI “makes mistakes at unpredictable moments, cannot distinguish truth from falsehood, and has no notion of meaning.” These shortcomings, he explains, are deeply rooted in how AI systems learn — through pattern recognition in massive datasets. “The moments when it performs brilliantly and the moments when it fails completely are intertwined.”

While major tech companies portray AI as an evolving intelligence that will soon surpass humans, Wiggers challenges that optimism. “These limitations are not temporary. They will never fully disappear.”

He points to examples of AI models that make confident but nonsensical claims or draw false correlations — such as a medical system that wrongly sent pneumonia patients home. These errors, he argues, reveal that AI doesn’t reason like humans do; it merely mimics language and logic statistically.

Beyond technical flaws, Wiggers highlights AI’s real-world costs: massive data centers, energy use, rare metal mining, and low-paid workers labeling data. “AI may seem immaterial, but its impact on the world is very real.”

Wiggers calls for a more reflective and responsible approach to AI. “Responsible is a verb,” he says. “We must keep weighing our choices, experiment critically, and sometimes decide not to use AI.” He urges professionals and students alike to strengthen their critical craftsmanship: “It will become more important than ever.”