The AI-Everything Ending. Likely a Slow Fizzle-out, Not a Pop The article explores the trajectory of the current AI bubble, drawing parallels to historical speculative bubbles such as the Tulip Mania of the 17th century, the Gilded Age railroad overexpansion, and the dot-com crash of the late 1990s. These past episodes, driven by unchecked optimism and speculative investment, ultimately collapsed due to overcapacity and unsustainable growth. The author argues that the AI bubble today is following a similar pattern, with excessive investment in data centers and infrastructure outpacing actual demand. Unlike previous bubbles, however, the author predicts a gradual decline rather than a dramatic crash, attributing this to growing public skepticism about the promises of AI. The piece highlights how AI’s current state resembles a “junk-food binge” — a fleeting high fueled by hype but lacking substance. It critiques the industry’s reliance on statistical models and machine-generated outputs, which often produce “garbage-in, garbage-out” results. The author warns that AI tools, while marketed as productivity enhancers, frequently create more work and frustration than value. For instance, a 2025 MIT report found that 95% of generative AI pilots failed to improve corporate profits or efficiency, while a METR study revealed that AI coding tools slow down developers rather than accelerate them. Consumer complaints about AI hallucinations and errors further underscore the growing disillusionment. The article also delves into the philosophical debate over whether AI can ever replicate human intelligence. It distinguishes between statistical pattern recognition (the core of AI systems) and authentic human cognition, which relies on sensory experience, reasoning, and subjective interpretation.#ai #mit #metr #tulip_mania #dot_com_crash
