Reflections

  Before We Talk About AGI, We Need to Talk About Consciousness Everyone in tech has an opinion on when artificial general intelligence arrives. Lab leaders say it's close — some have spoken of AI broadly outperforming humans at almost everything by 2026 or 2027; others put it at five to ten years. Yet others claim five to twenty. On the other side, some serious researchers like Yann LeCun etc argue current architectures can't get there at all, and broader academic surveys put the median estimate for human-level machine intelligence closer to the 2040s or later.   That's not a narrow band of disagreement. That's the most credentialed people in the field unable to agree within multiple decades of each other. And underneath that disagreement sits a question almost nobody in the timeline debate is asking directly: what actually is the thing we're trying to build toward?   I come at this from an unusual angle — ·        Years of wo...