The A.I. Bubble is Bursting with Ed Zitron [video]

Reactions to the video and its critic of AI/big tech

• Some find the critic’s work compelling: tech has “lost its charm,” produces user-hostile, low‑value products, and search quality (esp. Google) has deteriorated.
• Others say he’s a PR “influencer,” not technically deep; accuse him of misrepresenting facts to cater to anti–big-tech sentiment.
• Several comments note that criticizing crypto was easy and accurate, but applying the same snarky, surface‑level frame to AI makes these takes feel shallow or “unhinged.”

Is there an AI bubble?

• One camp: no bubble, just a boom in genuinely useful tech; compares skeptics to early crypto or social‑media bubble callers who were repeatedly “wrong.”
• Another camp: sees clear bubble dynamics—massive capital, weak business models, hype about AGI and “digital gods,” lots of me‑too product integrations.
• Middle view: like the dot‑com era—foundational tech is real, many current companies will fail, and payoffs may come later after infrastructure and experimentation.

Current utility vs. overhype

• Many report strong personal and professional value: faster coding, prototyping, scripting, documentation, debugging, and domain learning without deep prior expertise.
• Non‑coders describe automating workflows (transcription, image pipelines), fixing machines, and accelerating everyday problem‑solving.
• Others say models are still “just very good chatbots”: great at summarizing, rewriting, trivial code, but poor at hard problems, deep logic, or complex math/physics.
• Some argue only true AGI can justify current investment levels; otherwise foundation‑model providers may never reach profitability.

Economic, environmental, and sector impacts

• Several see rapid margin improvements inside organizations already using AI, but these use cases are mostly undisclosed “trade secrets.”
• Health‑care examples are debated: proponents foresee AI in intake, charting, imaging, and instructions; critics call this a chatbot arms race and potential nightmare.
• Concerns raised about energy use and “boiling the oceans”; others downplay or say it could be amortized over long‑lived models.

Education, reasoning, and data limits

• Heavy student use is noted; some worry about erosion of critical thinking, others compare it to calculators/Wikipedia and suggest shifting to editing/fact‑checking skills.
• Sharp disagreement over whether LLMs “really reason” vs. merely pattern‑match; both sides produce examples to support their view.
• The video’s “we’ll run out of data” claim is contested: commenters argue architecture and data quality, not raw volume, are now the main levers for improvement.