4 ms·

>retraining these fifty year old guys who have been truck drivers or coal miners While thats true, I think the premise is wrong. Techno-types love to imagine t

by _red 8y ago

>retraining these fifty year old guys who have been truck drivers or coal miners While thats true, I think the premise is wrong. Techno-types love to imagine that automation is just going to sweep through these industries. The devil of these things is in the details. Yes, perhaps the tech can get to approximate 90% of what a human can do. However, the real problem are outside of tech. Like liability insurance. Who gets sued when one of Elon's driverless trucks plows into a school bus? Surely not the "trucking company"? So the software company will have to assume that...Now, imagine there are 50M driverless vehicles on the roads, what does that monthly policy look like? What insurance company will underwrite that? What happens if a zero-day is found? etc. etc.

beat 8y ago
It's not just trucks, though. He gives a lot of examples, some of which are observable today. Like a lot of work done by paralegals and junior lawyers is just research - plowing through books and writing summaries and references. It's slow, tedious, and error-prone. A computer can do it much faster and better. How about reading x-rays? Train an AI by having it review the results from millions of other x-rays. That's a lot of radiologists whose work is replaceable. And that's really Yang's point. It's not just truck drivers. Any job that is basically rote work - even jobs requiring a high level of education and professional certification - is vulnerable to advances in AI and robotics. If your work isn't essentially creative, there's a good chance that it could be done cheaper, faster, and better by a robot at some point in the not-too-distant future.
- manquer 8y ago
  Unless you are a developer , while we built more abstractions and automation into our software engineering we also attempt more complex problems. I don't see dev being in lesser demand in the next 30 years
  - beat 8y ago
    Sure. Software is essentially creative work. I'm not worried about our industry. I'm worried about what our industry will do to other industries.
    - everdev 8y ago
      I think it will still affect software. Things can always be done faster. I think it'll be like farming where it won't go away but fewer people can do more work. I can imagine a day where voice recognition can allow marketing folks to just say what they want changed on a website and AI makes the update and asks for confirmation before pushing it live.
      - beat 8y ago
        I'm not worried. I've been in this industry for 25 years now, and seen multiple technical revolutions. (Back in my first serious job, we were hand-rolling streaming data protocols with lex and yacc on raw sockets! There was no ssh, no xml, no json, and http was a lab toy in Switzerland.) Requirements are hard. Understanding what needs to be done, and doing it all in the correct order, with buy-in from all concerned parties, is an essentially human, creative problem. It's not a simple matter of "marketing folks ask the AI to update the site", because they don't know how to ask it. Software is like all those tales of people getting their wishes granted by a genie or something, and regretting it because what they asked for wasn't actually what they wanted. And who turns someone's dreams into something the genie will do correctly? Programmers. We're the interface between the world of human needs and the world of technical realities. We're doing craft. Our tools may change, but our problem is constant.
    - gremlinsinc 8y ago
      Yeah, software devs will be the rich/elite in 20 years (we'll be the only ones employable).
      - krapp 8y ago
        Nope. Most software development jobs, along with many other "creative" jobs, will be automated as well.
ameister14 8y ago
>Who gets sued when one of Elon's driverless trucks plows into a school bus? Surely not the "trucking company"? Actually, yes, the trucking company gets sued. This is pretty well established. If your Toyota malfunctions and you hit someone, you get sued. You may sue Toyota afterwards, but you're going to get sued first. This has literally happened before. It's not a what if. Toyota recalled thousands of trucks about 3 months ago because of a braking issue. I fail to see the difference between this and car insurance of the future. Even if insurance were the issue, it's fairly easy to solve; in the contract you sign to use an automated vehicle, you'd assume liability.
- _red 8y ago
  >If your Toyota malfunctions and you hit someone, you get sued. You may sue Toyota afterwards, but you're going to get sued first. You extrapolating that future markets will work exactly like the past, because "its a car". But its doubtful it will. The reason why the auto industry evolved the way it did (driver gets sued), is because 90+% of the time it is operator error. Look at another example. Kraft distributes 10M packages of poisoned cheese, what happens: (a) All the local grocery stores get sued? or (b) Does Kraft get sued? Obviously "B" would be the case, since the local grocery stores would say: "We had no way to test the cheese for poison, it was pre-packaged from Kraft" This is very similar to what would happen in a 100% autonomous vehicle situation. The owner doesn't have access nor the capability to judge the engineering / code. >Toyota recalled thousands of trucks about 3 months Not the same. Toyota doing a recall to replace a $200 part doesn't come close to the potential liability that a autonomous manufacturer that had thousands of accidents...each potentially costing millions each. >Even if insurance were the issue, it's fairly easy to solve; in the contract you sign to use an automated vehicle, you'd assume liability. Yes, and the trucking companies will just say: No Their insurance rates would skyrocket so high, that it would be cheaper to just continue to use human drivers.
  - ameister14 8y ago
    No, people don't just sue the drivers because it's driver error, they sue because they have a chance to get money. People who get t-boned by a driver running a red light get sued by the insurance company of the driver that ran the light. It can have nothing to do with who is at fault, ultimately, just who is the easiest target. In your Kraft scenario, the store would likely be sued, as would the corporation running the store, Kraft, and potentially the packaging company. Here's a perfect example; the Romaine lettuce E. coli outbreak happens, right. A man eats a salad at a restaurant, gets sick. Who does he sue? Everybody. Has nothing to do with who is responsible, has a lot to do with who is going to pay his bills. Why is the liability for a $200 part lower? If the cars can't apply the brakes properly and people die in car accidents, are they any less responsible? Trucking companies won't say no, because it's extremely unlikely that all the cars will get into accidents all the time, and they don't have to pay more because the car didn't have a driver than they would if the car was being driven by their employee. Why would their insurance rates skyrocket? It appears there will be far less accidents with automated cars.

by _red 8y ago

beat 8y ago

It's not just trucks, though. He gives a lot of examples, some of which are observable today. Like a lot of work done by paralegals and junior lawyers is just research - plowing through books and writing summaries and references. It's slow, tedious, and error-prone. A computer can do it much faster and better. How about reading x-rays? Train an AI by having it review the results from millions of other x-rays. That's a lot of radiologists whose work is replaceable. And that's really Yang's point. It's not just truck drivers. Any job that is basically rote work - even jobs requiring a high level of education and professional certification - is vulnerable to advances in AI and robotics. If your work isn't essentially creative, there's a good chance that it could be done cheaper, faster, and better by a robot at some point in the not-too-distant future.

manquer 8y ago
Unless you are a developer , while we built more abstractions and automation into our software engineering we also attempt more complex problems. I don't see dev being in lesser demand in the next 30 years
- beat 8y ago
  Sure. Software is essentially creative work. I'm not worried about our industry. I'm worried about what our industry will do to other industries.
  - everdev 8y ago
    I think it will still affect software. Things can always be done faster. I think it'll be like farming where it won't go away but fewer people can do more work. I can imagine a day where voice recognition can allow marketing folks to just say what they want changed on a website and AI makes the update and asks for confirmation before pushing it live.
    - beat 8y ago
      I'm not worried. I've been in this industry for 25 years now, and seen multiple technical revolutions. (Back in my first serious job, we were hand-rolling streaming data protocols with lex and yacc on raw sockets! There was no ssh, no xml, no json, and http was a lab toy in Switzerland.) Requirements are hard. Understanding what needs to be done, and doing it all in the correct order, with buy-in from all concerned parties, is an essentially human, creative problem. It's not a simple matter of "marketing folks ask the AI to update the site", because they don't know how to ask it. Software is like all those tales of people getting their wishes granted by a genie or something, and regretting it because what they asked for wasn't actually what they wanted. And who turns someone's dreams into something the genie will do correctly? Programmers. We're the interface between the world of human needs and the world of technical realities. We're doing craft. Our tools may change, but our problem is constant.
  - gremlinsinc 8y ago
    Yeah, software devs will be the rich/elite in 20 years (we'll be the only ones employable).
    - krapp 8y ago
      Nope. Most software development jobs, along with many other "creative" jobs, will be automated as well.

ameister14 8y ago

>Who gets sued when one of Elon's driverless trucks plows into a school bus? Surely not the "trucking company"? Actually, yes, the trucking company gets sued. This is pretty well established. If your Toyota malfunctions and you hit someone, you get sued. You may sue Toyota afterwards, but you're going to get sued first. This has literally happened before. It's not a what if. Toyota recalled thousands of trucks about 3 months ago because of a braking issue. I fail to see the difference between this and car insurance of the future. Even if insurance were the issue, it's fairly easy to solve; in the contract you sign to use an automated vehicle, you'd assume liability.

_red 8y ago
>If your Toyota malfunctions and you hit someone, you get sued. You may sue Toyota afterwards, but you're going to get sued first. You extrapolating that future markets will work exactly like the past, because "its a car". But its doubtful it will. The reason why the auto industry evolved the way it did (driver gets sued), is because 90+% of the time it is operator error. Look at another example. Kraft distributes 10M packages of poisoned cheese, what happens: (a) All the local grocery stores get sued? or (b) Does Kraft get sued? Obviously "B" would be the case, since the local grocery stores would say: "We had no way to test the cheese for poison, it was pre-packaged from Kraft" This is very similar to what would happen in a 100% autonomous vehicle situation. The owner doesn't have access nor the capability to judge the engineering / code. >Toyota recalled thousands of trucks about 3 months Not the same. Toyota doing a recall to replace a $200 part doesn't come close to the potential liability that a autonomous manufacturer that had thousands of accidents...each potentially costing millions each. >Even if insurance were the issue, it's fairly easy to solve; in the contract you sign to use an automated vehicle, you'd assume liability. Yes, and the trucking companies will just say: No Their insurance rates would skyrocket so high, that it would be cheaper to just continue to use human drivers.
- ameister14 8y ago
  No, people don't just sue the drivers because it's driver error, they sue because they have a chance to get money. People who get t-boned by a driver running a red light get sued by the insurance company of the driver that ran the light. It can have nothing to do with who is at fault, ultimately, just who is the easiest target. In your Kraft scenario, the store would likely be sued, as would the corporation running the store, Kraft, and potentially the packaging company. Here's a perfect example; the Romaine lettuce E. coli outbreak happens, right. A man eats a salad at a restaurant, gets sick. Who does he sue? Everybody. Has nothing to do with who is responsible, has a lot to do with who is going to pay his bills. Why is the liability for a $200 part lower? If the cars can't apply the brakes properly and people die in car accidents, are they any less responsible? Trucking companies won't say no, because it's extremely unlikely that all the cars will get into accidents all the time, and they don't have to pay more because the car didn't have a driver than they would if the car was being driven by their employee. Why would their insurance rates skyrocket? It appears there will be far less accidents with automated cars.