Yes, I had the same impression. I'm sympathetic to the author's perspective but I can't muster even the minimal optimism they've shown here. The "process engineers" as described would themselves quickly be replaced by an automated system. The "statistical engineers", I think, would never be able to keep up with the rate of change of the AI models, which would likely have different statistical behavior and biases in each language/context/etc with each update, and so it's unlikely anyone would pay them to develop that required deep expertise in the first place. More likely, that work would be done at an AI foundation model company -- but it would be done just once, and then incorporated into the training process.
From my reading, it's really quite brilliant. He just says that he's about to start the tasks that, in reality, he's just finishing up. Then he delays reporting that the task is done. His estimates are then always made with perfect hindsight.
This only works though, if one is working in isolation and no one else checks out the work before the time one commits to the work. That would mean being some lone expert in that part of the product, which is not a good sign for the business, because bus factors of 1.
Round cities are even better for rail. You run a line in a circle, like the Yamanote line in Tokyo, and now it has the advantage of periodic boundary conditions. Everywhere is central!
For new rail at least, whoever wants to build them gets to own them, right?
I think what it comes down to is that if automobile companies had to build and maintain the roads, we certainly wouldn't have so many cars. But railway companies need to build the train lines, while competing with taxpayer funded automobile infrastructure. It's not impossible (see Japan) but also not easy.
It has a big geographic disadvantage too: the entire country is a mountain range. Japan's railway network relies on countless tunnels and viaducts, adding greatly to the cost, especially for high speed lines which require larger clearance and therefore larger tunnel diameters, and larger turning radii.
Geography is no excuse for the US not having better passenger rail service, especially when geography was no obstacle to the US having fantastic rail service in the 1920s.
I was part of CEO recruitment process (sadly not FAANG-like, so maybe it wasn't "so much"-level yet).
Amount of people who are both seriously willing to take the job(considering pressure) and have necessary skills is not very high. Tbh same is truth for any management job - a lot of competent people prefer calmer life.
Obviously for the very top compensation is bonkers and there's fair share of frauds that ended in the position for various reasons, but if you want someone reasonable pool shrinks quite fast.
I may not be _that_ competent, but the calmer life is worth sacrificing a fair few dollars for. For each job I've changed, I've gone to a lower paying position, such that I'm currently still on $10k (around 8%) less than I was earning... 5 years ago, two jobs ago, which was probably about the same as the job I left 6 years before that. All previous jobs were worth leaving.
Parenting, maintaining a long-term relationship, playing (two, kinda) sport(s), home-labbing, keeping up with the state of the world, all take time and I enjoy all of them. It allows me to enjoy the work I do too, to not resent it for all the other things I could have been doing.
I'm in a position of privilege to say any of this, but I've also been careful and relatively well planned with my finances in order to reach this point. I'd be kinda f'd if I was out of work for longer than 6 months, but I'm sure I could re-plan and re-organise priorities and spending to minimise the damage (but we'd definitely be f'd if we both were out of work...).
Have you considered being paid $20 million for one year, and then staying home with family for the next ten years? Even after taxes, you'd still have much more of both time and money than a career in, say, air traffic control.
Following Boeing example you given below - it's not like the guy was given offer to become a CEO out of the blue & had to endure year to be set for life.
He was slowly climbing through the ranks of huge organization over the span of almost 30 years. Given later revelations I certainly wouldn't call it easy or calm - likely even morally challenging sometimes (not admiring anything here - simply any position of power comes with this kind of issues - no matter you're playing for good or bad guys).
Taste of Boeing shareholders for execs is whole other discussion, but I really don't think there's huge crowd of people both willing and capable of filling those shoes.
Maybe, if that was even offered as an option. I doubt I'm that competent, however, and just the fact I have that doubt probably excludes me from the possibility.
Hell, I would gladly take any job where "failing upward" was the rule. Do it for a few years, fail, and then move on to the next higher level opportunity.
Most job levels fail downwards, but once you get to a certain level, for whatever reason, nearly everyone fails upwards. I think "director at a FAANG" and "VP at a medium sized company" are about the level where roles start defaulting to failing upward.
>Amount of people who are both seriously willing to take the job(considering pressure) and have necessary skills is not very high.
I don't believe that. The average CEO is going to have perhaps ten direct reports. I'm willing to bet nearly every one of those is capable and up for the job.
In my past I got high enough to barely touch the ceiling of a multi-billion dollar public company. I routinely had 1:1s with directs to the CEO. Not one of them would be willing or able to take that CEO job.
Just the amount of public facing interviews on CNBC would disqualify half and the other half wouldn't want to do it. They were already being paid very well.
If everyone actually came together and decided that they are not worth paying what they are paid now the compensation would likely normalize to a few multiples of other roles.
But lot of compensation is just magic money out of air that is stock valuations so the stock holder just gives slice of what they think they made. So as long as they financialised system exist so will extreme compensations.
Since most of it is magic money, I'm not sure how much sense is there in evaluating them on it. I.e. if I were suddenly worth $1B, but 90% of that was in a mix of assets I can't do anything with, and some bullshit Wall Street fake money that, at this scale, disappears the moment you look at it, then the actual money I could use would be much, much less.
If it's fundamentals of ML, I'm surprised to hear that.
If it's "how to use ChatGPT for creative writing" then I'm not surprised. Why would someone take a class from a teacher who has had only just as much experience with these tools as their students have?
I actually feel the opposite. I don't think people from outside CS will have that much interest into the very basics of AI because there is usually a huge gap between "this is how back propagation works" to any AI model that is remotely useful. And if you are interested in the fundamentals themselves you would probably be majoring CS anyway.
A course on how to use existing AI tools will be pointless, but if there is anything I know about college students is they love taking easy courses for easy credits.
Agree… OP said “not CS” so doesn’t seem surprising. If we’re going by anecdotes, AI classes in the CS dept have risen in popularity in the past few years.
Yeah, that's the main reason why copper is used, it's relatively cheap, safe, and very easy to work with. There are far better heat conductors out there but they're either extremely expensive or extremely toxic, e.g. beryllium. So this new stuff would have to be both very cheap and easy to make and easy to work with in order to succeed in the market, otherwise it'd be confined to niche applications like the military with beryllium.
reply