3 ms·

Well, sarcasm aside, it's fairly common to dismiss a new technology that can grow exponentially, because as humans we are tuned to linear thinking. In 1985, McK

Well, sarcasm aside, it's fairly common to dismiss a new technology that can grow exponentially, because as humans we are tuned to linear thinking. In 1985, McKinsey speculated that there would be 900,000 cellphone subscribers overall globally. Now we can just dismiss McKinsey as just consultants, but guess what, any one in their position would have probably speculated around the same number because that's what the models and the Math would show. Quite similarly, almost everyone is speculating on SDCs based on existing dogma and numbers, not taking into account how rapidly deep neural nets are evolving. Just look at the Imagenet competition history. I mean today we have neural nets that are better than humans at image classification. I'm quite sure, if you had asked even the most eminent researchers to speculate in 2012, they wouldn't have predicted such an outcome. It is also easy to take the position that I am taking, that SDCs will be a reality in the next 5 years, just based on hubris and some Elon Musk cool-aid. However my view is purely based on the growth that we are seeing in deep neural nets.

mikestew 9y ago
However my view is purely based on the growth that we are seeing in deep neural nets. Ah, see, there's where our POVs differ. My tack is that we've been "almost there" on things like speech and hand writing recognition for twenty years or so, and I wouldn't say we're "there" enough that I'd want a comparably competent system driving my car down the road. Adding to that is that from what I'm reading we're not to the level of, say, Alexa yet for SDCs. OTOH, you make good points about the rates of progress in neural nets, so perhaps I'm wrong. Experience says that I'm just as wrong in one direction as the rest of the industry is wrong in the other, and reality will end up somewhere in the middle. So let's check back in ten years. :-) But I wouldn't be shorting Berkshire Hathaway just yet regardless.

mikestew 9y ago

However my view is purely based on the growth that we are seeing in deep neural nets. Ah, see, there's where our POVs differ. My tack is that we've been "almost there" on things like speech and hand writing recognition for twenty years or so, and I wouldn't say we're "there" enough that I'd want a comparably competent system driving my car down the road. Adding to that is that from what I'm reading we're not to the level of, say, Alexa yet for SDCs. OTOH, you make good points about the rates of progress in neural nets, so perhaps I'm wrong. Experience says that I'm just as wrong in one direction as the rest of the industry is wrong in the other, and reality will end up somewhere in the middle. So let's check back in ten years. :-) But I wouldn't be shorting Berkshire Hathaway just yet regardless.