That's not the issue - insurance companies have no problem giving insurance to humans which have "individual learning capabilities."

The problem is that while insurance companies can extract compensation from humans - essentially they can coerce the humans to work "for them" to pay them back - they can't get compensation from the AI, and there isn't a good framework to get compensation from the AI developers.

The issue isn't that the AI learns - the issue is that our legal framework doesn't have a good way to get "compensation" from errant AI - because the AI itself doesn't care if it gets shut off. You can't hold the AI itself "liable" like you can a person. So until we change the liability framework, we'll be stuck. This isn't a technical problem, so much as a social one.

Part of self-driving cars being safer than people is that they don't drive "very close together" - they follow at the appropriate following distance and speed for the lines of sight. This alone accounts for a large portion of the improved safety performance of these systems.

Autonomous cars are actually very good at the driving task. Where they fail is when they encounter situations that are "abnormal" and I think the industry has a problem here; state of the art is either to train specifically for every situation (impossible in the limit) or to kick the decision to a human in a support center somewhere. Things like emergency vehicles, school buses, farmer markets, construction zones, etc.

The industry needs to figure out how to get their systems to "get confused, then reason their way out of a situation."

This long tail is where humans still do better - when there is some "new" scenario, like a particular combination of debris in the roadway, on a Tuesday, with gawkers, during a fireworks display. You will never be able to "train" your way out of that - you need the correct structural formulation for the problem space.

The big takeaway I have from their study is that their study areas - major metropolitan centers - is that cities have way, way worse than average incidents per vehicle mile travelled.

I mean the national average is only something like 1.25 fatalities per 100M miles, injuries only 75 per 100M miles, and "all" incidents like 140 per 100M miles. But the benchmark numbers on the linked websites range from 200 to almost 700 injury incidents per 100M miles in those cities.

Like, if you are using the average number, you can't even reduce the number of incidents by 841 over 271 million miles, because the expected value is only 203. You have to have a rate of at least 300 injury-incidents per 100 million miles to reduce by 841 at over 271 million miles. So really, their results are driven by San Francisco, Austin, and Atlanta.

Their confidence bands are also pretty low; for example they quote 2 serious or greater injuries in San Fran, which on 82 million miles there is a rate of "0.02 per million miles". But 2 injuries is not really enough to make a statistical inference.

Least of all because all the miles are with different versions of their software - you can't make any predictive claim.

Statistics never lie, but you can make them say whatever you want...

Hamming distance is the term you're looking for.

I was going to post a very similar comment: these people are not coders but they are project managers, and they are "employing" AI as their coding employees.

The thing is - there's "nobody" to take credit for the work, so the manager gets credit for something they didn't do. So it's definitely a skill and is work, but it isn't "coding" at all.

It's an interesting world - the AI is an extremely inexpensive employee and has enough skill to displace increasingly higher-skill tiers of actual software engineering and programming.

If I was running these hackathons, I would disallow AI or I would allow people to hire "code-as-a-service" people. Those seem functionally equivalent activities, just with AI being vastly easier to manage the logistics and you don't have to pay employment taxes or benefits to the AI.

It's no wonder there is so much tension about the many uses of AI - instead of hiring people to do work, it's another instance of paying to use a machine to do work at a price point lower than paying people.

The important distinction though is if this was a "preventable" failure that is due to something the engineering community already knows but was just omitted or done carelessly, or if the failure was indeed due to some new physics or unique application.

But just saying "hey we learned that this didn't work" is only useful if you learned a new thing that didn't work - if instead you had a structural failure because you didn't employ known best practices... that's wasteful.

I don't think we know enough at this point to know which case of learning this is. Hopefully it is truly new learning and not just "oh whoops we forgot to inspect those welds."

The unfortunate aspect of that philosophy is that our society now confuses "don't censor political speech I don't like" with "don't censor falsehoods which are tied to politically-charged topics."

We should absolutely encourage discussions about things we may not agree on - but we should also not give audience to things which are demonstrably incorrect.

How much of this, I wonder, is that Qualcomm has patents on things integral to the physics? So that inherently anyone else trying to make a modem has to use alternate means to make it work, which are basically poisoned by the standard so of course they won't work as well?

It can't really be that hard to make a radio from a physics standpoint, but I bet it can be difficult to work around patents. Especially if it's a "dumb" patent like "put a filter here" which should have invalidated the patent due to "anyone skilled in the art" of radio devices... but because of it competitors can't put a filter in that exact spot, so have to figure out some other place to put it which of course doesn't work as well because it isn't where you'd want it...or "we set this frequency so it can only be done using a component with this material's band gap, and we have the patent on this material" or something like that.

I always laugh / cry when people say something about "stop indoctrination!" because what they really mean is "we don't want that kind of indoctrination, we want this kind!"

What we really need as a society is critical thinking skills, the ability to draw conclusions from evidence instead of reporting only supporting evidence for a preconceived conclusion and suppressing other evidence. We need to have people that can determine if the evidence is complete and conclusive, not just matching what people want emotionally.

I think you would see a price increase. In most food markets what happens is food is set at a price; if it sits on the shelf long enough it gets discounted. If it still doesn't sell it is thrown out. So what will happen is the the shelf-time of non-discounted prices will be longer relative to discounted price, meaning that the price level increases.

The other option is that the stock of goods at a low price sells out first, leaving only the higher-priced alternatives remaining on the shelves, also pushing up the price level.

I have never seen an example ever where increased demand leads to decreased price except where supply increases. If you have an example otherwise... I'd be interested in seeing it.

I don't disagree that this might happen in the medium to long term, as people see that prices are up (because of more money to spend on things) so they might be willing to invest to create companies to take some of that new profit margin.

But this lags the initial influx of money, meaning the final price level is more likely higher than the initial price level, even after new business come online.

Also, if there is less money in banks for the reserve ratio, or there is less money available for "investors" to create new companies, who do you think (and, with what money?) is going to be creating these new companies?

it would just be redistributing the same amount of money around. This would not cause inflation since the same amount of dollars would be chasing the same amount of goods.

Only in aggregate. It's generally the case that the ultra-rich are not buying the goods and services that the lower classes are trying to buy. This means that if you suddenly shift a bunch of money (spending power) to the lower classes, there will very likely be an increase in prices for the goods those lower classes purchase because you can increase their money supply way faster than the supply of the things they want to buy.

In fact, the argument goes that it will be a double-whammy because the thing "the rich" buy are companies that generally make stuff. So if you shift spending to the masses, you will increase demand for goods and services while simultaneously decreasing the available supply of new businesses.

An approach that actually gives the desired outcome on all fronts is to tax the rich and directly create new businesses, therefore creating new supply of all goods and services, dropping prices, and likely employing people. It also helps protect against monopoly by ensuring a steady supply of competition. Sadly, this idea is "too left" for most people.

Do you have any evidence at all of any "green" policy "destroying the economy?" If you're going to fret about policies that have a negative impact on the economy, you'll have plenty of data from the present administration in the USA to examine in the near future.