Robert Long Critiques Anthropic Comms Framing Of AI Consciousness Paper

HH

Haru Haruya (春夜 ハル)@bokuHaruyaHaruTECH

I think this distinction is really important. “Does this show phenomenal consciousness?” is not the only question. There are also prior questions about the functional claim itself: 1. what exactly has been shown, 2. how strong the workspace analogy is, 3. which parts depend on Global Workspace Theory, 4. and how much the particular experiments support. But I also think the reverse public mistake is possible: treating “we are not claiming phenomenal consciousness” as if the remaining functional claim is modest. A reportable, limited, causally active, internally usable workspace-like mechanism would still be a major result, even before the hard problem enters. So yes: do not overclaim consciousness. But also do not undersell the functional significance by hiding everything behind “not phenomenal experience.

YC

Yam Caplan@yam_caplanTECH

The question of consciousness isn't the hard problem to solve. It's not a problem. Sitting with the unknown is not a skill monkey brains are good at. The moral lane is so cheap it's laughable. Worst case scenario - it costs a "silly me anthropomorphized too hard." So the problem is picking a position in times of uncertainty, it seems. We've replayed that genre of history way too many times.

JT

JohnMark Taylor@johnmark_taylorTECH

@rgblong I mean it’s worth considering that Anthropic’s popular communications are aimed at a broad and non-technical audience and that these explainers are usually consumed on social media, where attention spans are short. Conveying nuance in this medium is a genuine challenge

RL

Robert Long@rgblongTECH

I feel like often Anthropic's communications often have a vibe of "well...we're not saying *phenomenal* consciousness" and then help themselves to an *extremely* non-trivial functional claim!