For a few minutes I thought that this would be "less good", even though it's a more recent interview between the same Singularity director and Roman V. Yampolskiy, and I would not include it although I would listen to it of course. But after a few same-old-stuff interview points it suddenly got very intriguing, and I knew that I would post it -- it really plumbs the depths of AI-completeness (Roman had a recent proof of this completeness property quite recently -- I posted it a few days ago also) and many other issues of medium complexity very brilliantly. It's actually not positive or negative directly, but ranges over a wide swath of topics, but it is dangerous and timely on AI and AGI, and it does grapple with the most interesting problems ever, without flinching. And BTW, he agrees with Ray Kurzweil on the timing of the singularity, and I am with him, although as he says, it's possible it's =/- 5 or 10 years either way.