AI Bubble Pop: Here’s When, And How Many Jobs Destroyed | John-Clark Levin

Summary

John-Clark Levin, Head of Research at Kurzweil Technologies, maps AGI timing to 2029 and argues a market correction is likely before AGI because demo capabilities outpace reliability. He is constructive on AI chip demand, humanoid robots, and smart glasses as inference demand drivers. He also says Claude currently leads ChatGPT on his private benchmarks and discusses how AI films, labor, and education may change.

• Levin defines AGI as AI performing at least at educated/trained human level across all measurable cognitive tasks.
• His AGI timeline centers on 2029 due to compute price-performance doubling.
• He warns that even 4% prime-age job losses would be unprecedented, but sees double-digit unemployment by 2028 as too soon.
• He expects a market correction before AGI driven by the gap between demo peak capabilities and reliability.
• He argues inference demand from humanoid robots and smart glasses will keep AI chips busy despite efficiency gains.
• He places Claude ahead of ChatGPT on a plurality of tasks, with OpenAI, Google DeepMind, and Anthropic close.
• He says US AI models lead Chinese models by about 10 months after adjusting for benchmark overfitting.
• He sees AI-generated films weighing on traditional film production jobs and distributing through platforms like YouTube.