I don’t know which of you is right, but I think you’re saying different things. They’re saying positions that are actually open right now are paying less, and you’re saying salaries for existing employees are flat. They could be right if the positions which are open are only recruiting those with lower salary expectations.
No, I’m saying pay for new openings is flat. I just went through a job search, the numbers in the level.fyi data match my own job search experience looking at current openings.
- soda is up about 50% for name brand, and generic soda is up around 80%
- diesel is up, what 50%? how much is regular gas in california?
- housing "values" where I am at are up 25-40% in the last two years
- Potato chips are up at least 40%
- Apples are much more expensive
- Restaurants are up at least 25%, some 50-75%
Many of these products went up by these amounts yet also engaged in shrinkflation. The biggest example of this is Chipotle which almost doubled the cost of its burritos but also shrunk them in size by about a third.
Governments are motivated to reduce inflation in inflation statistics. Perhaps the cost of goods for the ultra-rich hasn't really changed much, but for the poor (or cheap) they have all gone up quite dramatically in two years.
oh god, how will SF SWEs survive with more expensive potatoe chips and soda. Maybe stop drinking poison and processed foods and you'll be fine. A whole rotisserie chicken is only $8.99 at whole foods
In the news articles I’ve read, it says the vehicle was kept in place at the request of the police. Can you explain why that was the wrong thing to do?
Deep learning-based recommendation models are very popular in industry. Not sure what “proof” you’re referring to, but it’s definitely not the case that random forests outperform DL based approaches in general.
I did a bit of deeper dive here, and the evidence remains mixed, but definitely generally in favor of random forests for recommendation.
My two cents: Most recommendation engines are far from perfect, and the customer is too often left scratching their heads. Given this, why bear the orders of magnitude more expense to go with with a deep learning solution?
Stricter licensing requirements get talked about a lot, but as far as I know the actual evidence that they improve driver safety is pretty mixed (with some exceptions, e.g. vision tests for people over 85.)
IMO it’s more likely things like street design and cultural attitudes towards risk are responsible for most of the difference.
My understanding is that German highways are similarly safer than American highways, so it's not just street design.
I agree that cultural attitudes play a large part; a somewhat easy way to change America's culture of unsafety is to gate the privilege of driving to those who are willing to change it. It's what we did for seatbelts, airbags, and just about everything else that's saved countless drivers' (and passengers') lives over the last few decades.
Maybe I’m not understanding the idea - if you’re saying we should punish people for demonstrated unsafe behavior (e.g seatbelt laws) then I agree, but that’s not really related to licensing requirements. Or is the idea that you’d just not issue licenses to people who are generally risk-seeking? If so, that doesn’t seem like something you could assess without a socially unconscionable false positive rate (and probably wouldn’t be very effective in changing cultural norms IMO.) Or something else?
Sorry if I'm not expressing myself well -- I was trying to draw a comparison between seatbelt laws (which, when introduced, were broadly griped about by drivers) and stronger licensing requirements (which, if introduced, will no doubt similarly be griped about).
I don't think there's a good (fair) way to devise a test for whether people will be risk seeking, in the same way that seatbelt laws can't stop scofflaws from not wearing their seatbelts. Instead, the purpose of these kinds of laws/regulations is to change the cultural "baseline" around safe behavior: wearing a seatbelt is the law, and most people do it by default now. Similarly, instituting a more intensive licensing regime (where people have to demonstrate not just driving ability but proficiency in safe driving) can change the cultural baseline around how drivers behave on our streets, our highways, etc.
In other words: let's keep licensing people, but make getting a drivers' license "intense" the way it is in much of Europe, rather than taking it for granted as a part of being an American adolescent. I think that can go a long way in terms of encouraging a more serious treatment of the responsibility that comes with driving, and which is currently lacking on American roads.
(And of course we should induce behavior away from driving to begin with, reconfigure our cities to favor pedestrians and cyclists, fund mass transit, etc.)
> instituting a more intensive licensing regime (where people have to demonstrate not just driving ability but proficiency in safe driving) can change the cultural baseline around how drivers behave on our streets, our highways, etc.
As I mentioned in my original post, the evidence that implementing stricter licensing requirements improves driver accident rates is mixed (really, mostly negative but not exclusively.) Personally, it seems much harder to impact broad cultural norms compared to a specific driver’s behavior, so I consider that to be pretty strong evidence against things working as stated. The parallel to seatbelt laws also seems dubious because a) it’s not really clear to me that seatbelt laws are the specific reason for the cultural shift and b) they were accompanied by enforcement, which tends to be a much stronger way to change behavior. A more analogues policy in my mind would be one that punishes drivers for a specific, dangerous, behavior - e.g. DUI laws, which have a lot of evidence of working. But that’s a completely different type of policy from what you’re suggesting.
That said, it’s not totally implausible that stricter licensing could improve things - part of the reason it gets brought up a lot is that people find it very intuitive. But in practice it doesn’t actually seem to work that way.
Worth stating that “something right” in this case is adopting life-saving technologies like seatbelts and airbags, exactly the kind of thing the original article is arguing against.
A cool concept, but probably decades away from breaking ground, if ever - not exactly scalable. Perhaps trying to solve traffic deaths by building a bunch of new cities is not a practical solution?
It would be an issue if they were lying about how often humans were needed, but as far as I can tell they’ve never claimed anything like the “100% success rate” stated in the OP’s post.
I do wonder why people automatically assume "AI" is "100% AI"?
Maybe we need to acclimatize. We all know when a company claims their soda is the tastiest or that their cars will make you cool, it's marketing fluff. Or that packaged food has to have some amount of preservatives. We should just assume that "AI" means "AI system", one that involves humans for software maintenance, taking care of edge-cases, and providing training inputs.
https://www.levels.fyi/2023/