Hi, I recently just came across your page and I really love your content. I will like to be your video editor, creating thumbnails and getting your content ready for UA-cam.
I vehemently disagree with your whole take about AI safety polarisation. Your points essentially amount to saying that the blame on this polarisation is on the optimist side, and you give the example of Marc Andreessen. First off, what you're doing is really not that far to what you're condemning him for, that is you're putting out his name to stand out as an example of a person doing what you consider the "wrong" thing. Yes you don't call him "the enemy" but I really don't think that that single noun is what the whole polarisation issue revolves around. Whether you call people out or not, and whether you call them "the enemy" or not, that only changes how brutal you are in calling out the fact that this divergence of opinion IS there to begin with, that divergence of opinion is not created merely by the fact of pointing it out in the '"wrong"' way. But more importantly in all this, the reason why people like Andreessen came out like that is because before they did, the discussion was actually not super harmonistic and everything, it was just dominated by the negative side. And whether you like it or not, to restore balance in such situation you have to put the opposite side out there, and put it clearly and sharply. Then maybe it's not perfect, but at least people that feel we should be optimistic about it don't feel ashamed and afraid in coming out with their opinion because there's just a single movement out there and it considers their opinion taboo. Is it the most perfect and tactful way ever to do it? Maybe not, but I think it's really besides the point to insist on putting such high standard about how he comes out with his side on all this, and concluding he's to blame for ruining the whole discussion and everything. He's really not. Also, your analysis that the government's behavior will be "oh hey this guy dares to think and talk publicly about this in such extreme manner, let's just push out policies that goes against what he thinks", I think saying that that's a flawed analysis would be an understatement. Is that really your view on current democracy? That we should be underhanded in the way we voice our opinions in the face of the threat that the government will act to repress it? And how can you be that dismissive of the doomer view that was so prevalent and still is? You really think that leaving it unchallenged is the best way to to steer the government into not being too anti-technology? No, coming out strongly like he did WAS necessary, either that or have some super sophisticated meme engineered movements like e/acc, because the fact is that otherwise any movement for techno-optimism would have be entirely and utterly crushed by the dominant negative doomer program we had encumbered upon us. Last point, I think the whole analogy of AIs being like an alien invasion is really misleading. Like either AIs are an absolute transparent white box to us, we have perfect mechanistic interpretability and control over them, OR they're complete aliens and we should all call ourselves friends and comrades against this coming invasion. What kind of aliens would aliens be if we actually did have control over the data their very minds are trained with, their training algorithms, the fact that we know they're trained neural network to begin with and not some unknown entity, the fact that we can test and prod them over and over, etc. Even if success is not guaranteed, we ARE actually building them to be useful, that's not an alien invasion scenario. That's kind of Andreessen's point in all this, this sort of attitude, calling AIs freaking aliens, is acting like an "enemy" whether you like it or not, it's implicitly being hostile to progress and by consequence all the benefits that can come from it. How is that a wrong description, and how is it so offensive to merely say so? It's easy NOW to say we should've build more nuclear reactors, but the point it that the reason we didn't is because we were too apprehensive towards high tech, and exactly the same thing is happening with AI now.
Hi,
I recently just came across your page and I really love your content. I will like to be your video editor, creating thumbnails and getting your content ready for UA-cam.
I vehemently disagree with your whole take about AI safety polarisation. Your points essentially amount to saying that the blame on this polarisation is on the optimist side, and you give the example of Marc Andreessen.
First off, what you're doing is really not that far to what you're condemning him for, that is you're putting out his name to stand out as an example of a person doing what you consider the "wrong" thing. Yes you don't call him "the enemy" but I really don't think that that single noun is what the whole polarisation issue revolves around.
Whether you call people out or not, and whether you call them "the enemy" or not, that only changes how brutal you are in calling out the fact that this divergence of opinion IS there to begin with, that divergence of opinion is not created merely by the fact of pointing it out in the '"wrong"' way.
But more importantly in all this, the reason why people like Andreessen came out like that is because before they did, the discussion was actually not super harmonistic and everything, it was just dominated by the negative side. And whether you like it or not, to restore balance in such situation you have to put the opposite side out there, and put it clearly and sharply. Then maybe it's not perfect, but at least people that feel we should be optimistic about it don't feel ashamed and afraid in coming out with their opinion because there's just a single movement out there and it considers their opinion taboo.
Is it the most perfect and tactful way ever to do it? Maybe not, but I think it's really besides the point to insist on putting such high standard about how he comes out with his side on all this, and concluding he's to blame for ruining the whole discussion and everything. He's really not.
Also, your analysis that the government's behavior will be "oh hey this guy dares to think and talk publicly about this in such extreme manner, let's just push out policies that goes against what he thinks", I think saying that that's a flawed analysis would be an understatement. Is that really your view on current democracy? That we should be underhanded in the way we voice our opinions in the face of the threat that the government will act to repress it? And how can you be that dismissive of the doomer view that was so prevalent and still is? You really think that leaving it unchallenged is the best way to to steer the government into not being too anti-technology?
No, coming out strongly like he did WAS necessary, either that or have some super sophisticated meme engineered movements like e/acc, because the fact is that otherwise any movement for techno-optimism would have be entirely and utterly crushed by the dominant negative doomer program we had encumbered upon us.
Last point, I think the whole analogy of AIs being like an alien invasion is really misleading. Like either AIs are an absolute transparent white box to us, we have perfect mechanistic interpretability and control over them, OR they're complete aliens and we should all call ourselves friends and comrades against this coming invasion. What kind of aliens would aliens be if we actually did have control over the data their very minds are trained with, their training algorithms, the fact that we know they're trained neural network to begin with and not some unknown entity, the fact that we can test and prod them over and over, etc. Even if success is not guaranteed, we ARE actually building them to be useful, that's not an alien invasion scenario.
That's kind of Andreessen's point in all this, this sort of attitude, calling AIs freaking aliens, is acting like an "enemy" whether you like it or not, it's implicitly being hostile to progress and by consequence all the benefits that can come from it. How is that a wrong description, and how is it so offensive to merely say so? It's easy NOW to say we should've build more nuclear reactors, but the point it that the reason we didn't is because we were too apprehensive towards high tech, and exactly the same thing is happening with AI now.