4 ms·
This logic is fundamentally faulty. Suppose, for example, that you're on "the front side" of a bell curve. As you go up the bell curve, until you near the peak,
by hn_throwaway_99 4y ago
This logic is fundamentally faulty. Suppose, for example, that you're on "the front side" of a bell curve. As you go up the bell curve, until you near the peak, you can look behind you and think "Look at this, we've been going up for 100 years, and we're still going up!" But, if/when you do near the peak, it actually can turn quite drastically, and quickly.
The primary issue for me is the rate at which technological progress is moving is making it ever more difficult for humans to adjust. I mean, suppose that in 10 years we have fully capable self driving cars. I'm not necessarily saying that's going to happen (in fact I actually think it won't), but I do think it will happen at some point. Right now, truck driver is the most common job in that vast majority of US states. When self driving cars do eventually come, all of those jobs will disappear pretty quickly. Do you honestly think someone who has been driving a truck for 20 years will be able to "retrain" quickly?
And it's certainly not just truck drivers. Count me as one of the people who are astonished by ChatGPT's programming abilities. While I don't believe it's "there yet", I was pretty amazed that I was able to ask ChatGPT to program something for me, and it did quite well but the code had a subtle bug, and then when I pointed out the bug ChatGPT was able to fix it itself. And this wasn't even on the GPT-4 version. This stuff only gets better with time.
- meh8881 4y agoThat’s not really what happens though. Truck drivers don’t disappear. They just become “truck driver attendants” who serve only to attend to occasional human requiring edge cases. They don’t require a commercial driver license and they pay minimum wage. And thus the lower class grows.
- hn_throwaway_99 4y agoBut do you really think we'll see the same number of truck driver attendants as truck drivers? Heck, one of the things that already exists is autonomous truck caravans, where one driver can "drive" a fleet, of say, 5-10 cars.
- meh8881 4y agoI think we will see more attendants than truck drivers now, because without truck drivers, trucks are cheaper, and we will want more trucks. Probably smaller trucks too. I don’t have strong opinions on the ratio of attendants to trucks. You may be correct.
- jjk166 4y ago> one of the things that already exists is autonomous truck caravans, where one driver can "drive" a fleet, of say, 5-10 cars. Trains long predate the truck, and allow a single driver to drive hundreds of cars. But the major advantage of trucks is last-mile logistics - going from the depot where large volumes of cargo pass through to the specific place the cargo is needed. The number of truck drivers, and of future truck attendants, is proportional to the number of stops, not the amount of material being delivered. If anything, removing the bottleneck of requiring a person with a CDL to make a pickup or delivery will cause the number of truck attendants to be much larger than the current number of truck drivers (as already seen for example with Amazon delivery services, UPS trucks, etc).
- nradov 4y agoProgress on autonomous vehicles appears to have mostly stalled. And even if it can eventually be made to work on some roads, truck drivers do a lot more than just drive the truck. There is no guarantee that the overall rate of technological progress will continue to increase. The opposite could well happen. Right now we are seeing the effects of multiple S-curves stacked on top of each other but progress is already slowing in many fields because the low-hanging fruit has been picked. For example, civil aviation has only eked out a few percentage points of improvement in efficiency and safety over the past couple decades; the latest airliners are actually slower than their predecessors. Or in medicine we seem to be running out of new small-molecule drugs to find and thus on average with all the failures it now takes more work than ever before to bring a new drug to market.
- valine 4y agoI expect there is lots of low hanging fruit to be picked from LLMs. Imagine an LLM running locally on your car, refined to pick your lane based on the current traffic conditions and upcoming map data. Many of the problems with self driving come for a lack of basic reasoning ability, an ability LLMs seem to emulate quite well. For example you can lay out driving scenarios to ChatGPT and get reasonable responses back. It’s not as simple as running a Llama instance on your Tesla, but I expect language models will have had a big part to play when we finally get self driving to work.
- nradov 4y agoLarge language models are totally unsuitable to autonomous vehicles. No one is going to write code to describe a dynamic driving situation as static text which could be used as an LLM prompt. Some of the underlying core technology could theoretically be used in that way by taking input from multiple videocameras and other sensors. But that approach is a dead end for consumer product development until someone solves the testing and explainability problem. Current limited AI systems are too unpredictable for safety critical applications. Just because AI software seems to work in testing doesn't mean you can have any confidence that it won't do something wildly dangerous in a novel real world situation. Vehicle manufacturers and their insurers will not take on the high potential liability even if autonomous systems are on average safer than human drivers.