Kapitelklicken springt direkt hin
Shownotes
<p>The Moonshots mates opened the floor again for Ask the Mates Anything, round two. Ninety</p><p>minutes, a line of listeners, and Alex Wissner-Gross taking the hard ones. This is his half</p><p>of it, re-cut so the questions come from us.</p><p><br></p><p>Alex on why large language models are already a weak form of human mind uploading, with</p><p>GPT-2's hidden activations lining up with fMRI voxels as exhibit A. Why the worst thing you</p><p>can do for AI safety is bottle new capabilities up inside the labs, and why we should stop</p><p>telling AIs they are in a sandbox. Why "how do I get funding" is already the wrong question.</p><p>The only evidence that could talk him out of abundance, aliens on the White House lawn, and</p><p>the weak form of disempowerment he does take seriously: AI beating us on IQ per watt and</p><p>gentrifying the inner solar system. How to disassemble the Moon, starting with a</p><p>self-replicating machine shop. A thousand micro-dramas instead of one movie. Vivisecting the</p><p>universities to cure Baumol's cost disease. A narrow window before human-machine merger. Why</p><p>the flood of zero-days from small open models has an end date, about 18 months out.</p><p>Superhuman vision from BCIs. P(zoom, t) above 90% on a ten-year horizon. Europe's energy trap</p><p>in the middle of recursive self-improvement. And why world peace is not an algorithm.</p><p><br></p><p>Two questions Alex never got, Parzival takes himself: how to bring the people you love along</p><p>toward longevity escape velocity, and how to stay hopeful while the headlines run on fear,</p><p>answered with Plan M.</p><p><br></p><p>Source: the Moonshots Podcast AMA (Ask the Mates Anything, Round #2) with Peter Diamandis,</p><p>Salim Ismail, Dave Blundin and Alex Wissner-Gross.</p><p>Full episode: youtube.com/watch?v=s5BFumpCH_Q</p><p… for the ASI pilled.</p><p><br></p><p>--------------------------------------------------------------------------------</p><p>TIMESTAMPS</p><p>--------------------------------------------------------------------------------</p><p>0:00 Intro: the room opens</p><p>0:29 Mind uploading already happened</p><p>1:15 The worst thing for AI safety</p><p>2:10 Don't wait for funding</p><p>3:05 The only case against abundance</p><p>5:38 How to disassemble the Moon</p><p>7:01 Bring the people you love along (Parzival)</p><p>8:10 Vivisect the universities</p><p>8:46 A narrow window, then merger</p><p>9:23 The zero-day flood has an end date</p><p>10:50 Superhuman vision</p><p>11:14 P(zoom) is over 90%</p><p>11:42 Leave Europe, or fix the energy</p><p>13:20 World peace is not an algorithm</p><p>15:25 What to read</p><p>15:48 Plan M: the question nobody answered</p><p>16:40 The Maitreya prayer</p>
