4 ms·

The only reason we see articles like this about AI stealing (or not stealing) jobs is because we have a collective fantasy that generalized intelligence is righ

The only reason we see articles like this about AI stealing (or not stealing) jobs is because we have a collective fantasy that generalized intelligence is right around the corner. It's not. But that doesn't mean we won't get there eventually. AI (and AR/VR, but that's another topic), to me, is like PCs in the early 80s. Everyone can kinda sorta theoretically understand the value, but the technology has not advanced enough to have a meaningful impact on our day to day lives, outside of very well defined problems. The use cases that most of us can think of require massive leaps in technology. Once upon a time, people thought of computers as fancy calculators. Steve Jobs had to compare them to bicycles to get people to understand the benefit. Computers were so bad for so long that any movie that tried to feature a hacker subplot had to come up with all kinds of crazy visuals to help the audience relate to this strange new world. Same with AI. We can see what it's doing now, and be disappointed when we realize it's nowhere near as good as we've been led to believe. Or we can zoom into some distant future where AI is just as conscious as anyone else, but super intelligent. What we struggle with is the in-between, the journey to getting there. DALL-E is amazing. One day we'll look back and be able to see how crucial of an advance this is. But the PR story around DALL-E is all smokes and mirrors. It's the classic tale: Scientists make a discovery; the general public doesn't understand the significance of the discovery; so someone (usually in marketing or PR) has to paint a picture to show how incredible it is, otherwise funding will dry up. That picture, inevitably, is both useful and inaccurate. For another analogy: Just like a map of the world is inaccurate but useful, so are these fantastical stories. We just need to remember: the map is not the territory. It's just a tool to help us navigate the territory.

mejutoco 4y ago
> Same with AI. We can see what it's doing now, and be disappointed when we realize it's nowhere near as good as we've been led to believe. I won't be disappointed. I will be relieved! Tech fetishism can have dangerous consequences, as we are seeing with self-driving cars.
- nickelcitymario 4y ago
  Don't self-driving cars still have a much better safety record (per mile driven) than regular cars?
  - mejutoco 4y ago
    Since no fully autonomous cars are sold and widely deployed, we do not know. A lot of tech people are blind about fetishism of tech for its own sake. I am all in for new tech and modern inventions, but I am also surprised when people advocate for new things even when they do not work because of aesthetic reasons. Software engineers are not immune to this ("our scientists were so preoccupied with whether or not they could, they didn't stop to think if they should") Short rant about self-driving ahead. I think we will reinvent the train. Fully autonomous cars will work only in certified roads with special signage. This will offload the cost to the private car owner, as opposed to building public transport. Otherwise, it will remain a lane assist tech, which I am perfectly fine with. IMHO until I see fully self-driving trucks I won't believe it is around the corner, since the economics for trucks make much more sense. But, since the cars can be controlled remotely via 5g, I do not know if it would ever be worth the cost to have them be fully autonomous, instead of remotely controlled by humans. Maybe the solution is a remotely controlled car by a human. Interesting times.

mejutoco 4y ago

> Same with AI. We can see what it's doing now, and be disappointed when we realize it's nowhere near as good as we've been led to believe. I won't be disappointed. I will be relieved! Tech fetishism can have dangerous consequences, as we are seeing with self-driving cars.

nickelcitymario 4y ago
Don't self-driving cars still have a much better safety record (per mile driven) than regular cars?
- mejutoco 4y ago
  Since no fully autonomous cars are sold and widely deployed, we do not know. A lot of tech people are blind about fetishism of tech for its own sake. I am all in for new tech and modern inventions, but I am also surprised when people advocate for new things even when they do not work because of aesthetic reasons. Software engineers are not immune to this ("our scientists were so preoccupied with whether or not they could, they didn't stop to think if they should") Short rant about self-driving ahead. I think we will reinvent the train. Fully autonomous cars will work only in certified roads with special signage. This will offload the cost to the private car owner, as opposed to building public transport. Otherwise, it will remain a lane assist tech, which I am perfectly fine with. IMHO until I see fully self-driving trucks I won't believe it is around the corner, since the economics for trucks make much more sense. But, since the cars can be controlled remotely via 5g, I do not know if it would ever be worth the cost to have them be fully autonomous, instead of remotely controlled by humans. Maybe the solution is a remotely controlled car by a human. Interesting times.