Not necessarily famous, but faces existing in training data or false positives making generalizations about faces based on similar characteristics to faces in training data. This becomes problematic for a number of reasons, e.g., this face looks dangerous or stupid or beautiful, etc.
At least as of 10-15 years ago US Cellular owned their rural markets and leased their urban markets. In NYC they are likely reselling Verizon’s network.
Research indicates that SUVs are indeed more dangerous to pedestrians compared to other vehicle types in the United States. A study by the Insurance Institute for Highway Safety (IIHS) highlighted that late-model SUVs are more likely to cause fatalities to pedestrians than cars. This is attributed to the higher front profile of SUVs, which tends to result in more severe injuries upon impact. The study found that at speeds greater than 19 mph, SUVs caused more serious injuries and were more likely to result in pedestrian fatalities compared to cars. Specifically, at speeds of 20-39 mph, 30% of crashes with SUVs resulted in pedestrian fatalities, compared to 23% for cars. At speeds of 40 mph and above, all crashes with SUVs resulted in pedestrian fatalities, compared to 54% with cars. This indicates a significant increase in the risk posed by SUVs at higher speeds[0].
Further research supports these findings, showing that trucks and SUVs with hood heights greater than 40 inches are about 45% more likely to cause fatalities in pedestrian crashes than shorter vehicles with sloped hoods. The study, also by the IIHS, used data from nearly 18,000 crashes and noted that tall, squared-up hoods, characteristic of many best-selling SUVs and trucks, contribute significantly to the risk. The number of pedestrian deaths has significantly increased, with pedestrian fatalities jumping 13% to 7,342 in 2021, marking the highest number since 1981. This rise in pedestrian deaths has outpaced the increase in overall U.S. traffic deaths, highlighting a growing crisis in road safety related to larger vehicles[1].
These findings underscore the need for vehicle design changes to improve pedestrian safety, particularly as the proportion of SUVs on U.S. roads continues to rise. Despite advancements in vehicle safety that have reduced overall motor vehicle crash fatalities, the increased lethality of SUVs to pedestrians poses a significant challenge that requires attention from both manufacturers and regulatory bodies.
Good summary. Something ChatGPT missed is that SUV's are taller, and that tends to increase speed, because _perceived_ speed is lower the farther you are from the ground
And they present very little risk to pedestrians as a result. You're sitting LOWER than the pedestrians. You can see a bottle-cap on the road. You don't feel superior to anyone at all.
It’s funny, I’m often tempted to fact check data or lookup jargon, etc. and comment to save someone else the trouble. I once did this on the seriouseats subreddit with copy paste from a relatively reliable source and met with an insane heated argument over what amounted to semantics and a flurry of downvotes. I wonder if attribution to ChatGPT increases civility towards the commenter or if HN is just generally more civilized.
Lots of folks taking this approach and feels like the wrong entry point, e.g., similar to asking LLMs to generate bytecode when compilers exist or 3d printing a machine vs. building a machine from 3d printed components.
1. Business users aren’t prepared to talk to their data in meaningful ways and this is an opportunity for LLMs to enhance the way users ask questions.
2. SQL modeling languages exist (although I’m not sure there are well maintained open source good ones and this is the biggest obstacle to what I’m working on now) and LLMs can extend these effectively by adding components such as dimensions, metrics, relationships, filters, etc. with less chance of hallucination
3. Deterministic SQL generation from a modeling repository is easier to troubleshoot and provide guarantees than end-to-end generation.
4. Existing SQL can be parsed to generate modeling components that can be committed to the model repository with LLM assistance
5. Much of the richness of going to data to answer questions is context, e.g., how does this dept compare to others, this month to same month last year, etc. Modeling languages are good at expressing these transformations, but business users and often analysts aren’t good at capturing all the permutations of these types of comparisons. Another area where LLMs can help apply tooling.
IMO, LLMs are more effective at using tools than generating complex outputs.
Quite a number of studies suggest that chatbots are an effective tool for mental health support. Doesn’t need to be either or, but one could imagine scenarios where it may be more effective than a human mental health professional, e.g., 24/7 availability.
I do think there’s some nuance to that research though - just the act of typing out thoughts alone to a neutral party seems like it would be helpful, regardless of what’s said in response.
However, speaking personally - if I were suicidal, called an emergency line, and got connected to a computer instead of a person: that would feel like a brushoff and make things much, much worse.
My belief is that chatbots can be great for general mental health maintenance but are likely a massive liability for an acutely distressed population in the general case. I have zero faith in the US govt.’s implementation to respect that subtlety.
I feel like the people who complain about Uber/Lyft from a service perspective never used taxis extensively. Living in Chicago (Lakeview) for years while traveling for work made me absolutely hate taxis. When scheduling, they would no-show at 5 in the morning causing missed flights. Rides from the airport would require standing in line many times over an hour, especially on a Thursday night. They all absolutely reeked of body odor. The drivers would consistently scam “card machine broken, cash only” or “I forgot to turn on the meter” and unless you threatened to report them, they would take advantage of passengers. Drivers were sketchy and rarely matched the credentials on the taxi medallion.
I’ve also lived in areas where taxi service was essentially nonexistent. I wonder how many DUIs and related accidents have been prevented by ride share apps.
Traveling abroad in Europe the apps work simply regardless of my command of the local language to explain my destination and keeps the drivers honest so they aren’t taking “the long way”.
How is anyone supporting taxis as superior to this? There was absolutely no accountability.
From an business model perspective, I would wager that eventually you could get this to a point of sustainability that doesn’t require armies of engineers and various support staff, a la Twitter.
Living in Chicago (Lakeview) for years while traveling for work made me absolutely hate taxis.
Hello from up here in Edgewater.
I realize there are huge problems with the finances and liabilities of Uber…but taxis before ride shares were a pure nightmare, from a UX perspective.
Uber brought us…
* Deterministic pricing (for the most part)
* Flawless ubiquitous credit card transactions
* The ability to point to a random location on a map and just magically have a car take you there
I know Uber’s finances are shit…but you know what? I would gladly pay more, because you get a car right to your door that delivers you to an arbitrary location with nobody extorting you for extra physical cash.
I used taxis in LA plenty of times and in general it was nothing like what you're describing.
I only ever had one no show and that was because a lot of us ordered taxis from the same place around the same time so dispatch though it was some sort of prank.
Getting a taxi from LAX was as easy as grabbing my stuff from baggage claim and strolling out the door directly into the first taxi waiting at the curb. Usually there was a person there on the sidewalk directing the taxis too.
When I was new at this, I just gave the driver the address I wanted. One of the first taxi rides I took, the driver handed me a book of paper maps and asked me to find the place for him, lol. After that I learned to say "take me to (address), it's by (street) and (cross street). It's not far from (big street) and (other big street)" Worked 100% of the time.
On arrival I'd pay in my preferred method (cash) and tip the driver knowing he would get the entire tip.
Nowadays, I've had at least 5 lyf/ber cancel on me. Getting one out of SFO involves dragging my bags all the way out to the top floor of the parking structure where I get to wait around for 45 minutes in the rain while other people's rides arrive in random order. The rating system is useless because if I rate anything other than 5 stars it's the same as asking for the driver to be fired. Any tip I leave in the app is getting a sizable chunk shaved off the top by the company. At least my destination is automatically in the driver's GPS. Unless I'm going to Stanford which is notoriously hostile to cars and GPS often directs the driver to the wrong place...
Agreed. I take taxis now in many places because they're now better than uber, but pre-uber they were not. What I used to like best about taking uber was sticking it to the taxi industry. They were a textbook "disruptor". I'm glad there's a credible threat to taxis that means they had to improve and look over their shoulder. In the markets I travel frequently, uber sucks now though, and I've gone back to taxis, which also suck, just less, and are way better than the pre-uber days.
But while rideshares can be better for me than listening to some taxi driver go off on some misogynist rant, that doesn't mean that they're better once you start taking into account wage problems, increase in CO2 production, etc.
But seriously, taxis in Chicago can get fucked. They were so terrible.
I don’t buy wage problems because my estimate on Chicago taxis was most medallions were held by investors who were not drivers and most drivers were undocumented with less protection than Uber provides.
I buy the CO2 aspect. Many cases where Uber is now more convenient than public transport vs. El is better than a taxi
Anecdotally, ive personally lived in areas (california central valley, rural new england) where lyft and uber were virtually nonexistent and local taxi companies were the only option. ive had horrible experiences "reserving" rides with uber and lyft only to have nobody show up at the designated time, leading to missed trains. Taxis have always been far more reliable for me in rural areas, provided I call in advance and within their hours of operation. when i'm in cities, I find public transit to be more sensible terms of cost and reliability than any other option. I don't mind paying cash, I prefer it.
I cant help but feel as though a lot of the people defending these 'rideshare' companies on HN are living in coastal city bubbles. It's simply not an option in many places, and the apps will outright lie to you that they're able to get you a ride.
I live in a rural area, where the taxis, Uber, and lyft compete. The taxi companies don't offer calls in advance, or reserved rides, not one or four companies that could cover my area. You would have to hire a limosouine driver. Lyft is by far cheaper on average, but consistently fewer drivers. Hilariously, Uber is the reliable service for reservations.
Not enough people, especially on this site realize this when they generalize about consultants being lazy, stupid, parasitic, etc. - the majority of the folks in SV could never cut it at an elite boutique consultancy. That said, consulting at an elite level doesn’t scale well, the equity multiples aren’t good and the lifestyle is a grind at best and mentally and physically unhealthy at worst.
Does it pay better than FAANG? I can't imagine anyone wanting to do that for less money than they would make working at the companies they consult for.
Likely a bit lower if you're looking like-for-like, but there are trade-offs that make it worthwhile.
Major consulting companies hire everywhere and have offices everywhere. Excepting the last couple of years, which are looking like an anomaly at this point, FAANG requires one to relocate to NYC/SF/Seattle. There are a lot of bright people who can't make that move, so consulting ends up being a good alternative. In non-HCOL markets, consulting pay is usually some of the best.
Unless you make partner, comp is going to be just base + bonus without equity. Even outside of HCOL, base can end up being higher than base at FAANG, which means when FAANG equity is down big like it is right now, the gap narrows.
Partner at a Big-4 or McKinsey/BCG/Bain will reliably pull $1m TC after a year or two. IMO making partner is easier than making FAANG director. PWC and EY both have 3-4,000 partners, for example. McKinsey has 2,700 partners and only 38,000 employees (a good chunk of which are back-office non-billable). Contrast that with the number of L8+ at FAANG which is usually 5-10x fewer, from what I can gather.
Ultimately if you imagine a 28 year old consultant making $170k in Kansas City working remotely with a FAANG team of 24 year olds making $200k in Mountain View, it's quite possible that the consultant is banking more than the FAANG team, and with a different potential trajectory comp-wise.
Consulting is in a weird spot right now. It used to pay well and give good projects that allowed to get experience and nice exit opportunities (aka skip few years of usual career grind).
Now lots of projects are uninteresting. There are still elite and specialized consultants who do complicated stuff but they are a minority.
Often the teams now are just 'staff augmentation' - headcount outside of headcount.
For programmers consulting never really made much sense anyway. Why sit at BIG4 company making slides when you can sit in FAANG coding?
Consulting was always for finance guys. But now top finance guys go to investment banking, machine learning or (as funny as it sounds) crypto.
There are still good projects with good exit opportunities in finance, but it is night and day when compared to 80s or 90s - when consultants were the true elite.. just because they could see how things are made in different companies. Now the companies blog how they do stuff.
This is the type of uninformed comment I mentioned earlier. The draw of specialized boutique consultancies is the ability to step into challenging situations, take charge, make a big impact and move on to the next engagement once the problem is solved.
You get lots of at bats to do “something big” as opposed to 9-5 keep the lights on work that most engineers do for years in stagnant, highly politicized cultures year after year waiting for their boss to quit to get a promotion. It’s also a good way to level up a stagnant career.
The downsides include always “living in someone else’s house”, having to adapt to the clients tech and culture, having to leave your work behind and start from scratch.
Agreed that these type of shops are in the minority and once they scale, they exit to the big guys who then kill the culture and drive away the talent.
Palantir (from the outside) seems like a good example of this dynamic scaling along with the advantages of maintaining their own stack. Could you imagine what it would be like to be an engineer employed by the customers they serve?
