Get a better job my guy. Unrelated to AI, why are you living like that it sounds horrible and you can just get another job. I'm happy to help you fix up your resume if you would like.
4,222 karma · joined June 2, 2009
Water is a renewable resource. I know you folks in California can't comprehend this, but in the parts of the country where it actually makes sense for humans to live water just falls from the sky multiple times per week. We get so much water the problem is making sure we get rid of it safely, we don't have to fight over who has the most senior claim to it or decide whether we want to have endangered species or almonds more. Our streams don't run dry 4/5 of the year. We don't have to check about water restrictions when we water our lawns because there are never water restrictions and we never have to water our lawns.
There are entire areas miles across in this country where if you drive through it you might throw up from the smell. The ponds full of animal feces make the air un-breathable. Where is the protest over that? You realize that this literally does ruin ground water and poisons surface water? Where are the people coming out to say they have to live 4 miles from oceans of pig feces and when the wind blows their way they can't go outside? Yet the media is able to find the 4 people on earth that want to say a building full of computers is 'loud' and that computers use lots of water? When was the last time you filled the water tank on your computer? Yet people are very quick to believe that somehow a warehouse with a bunch of computers in it somehow destroys water?
When did datacenters start using up all the water? Nobody seemed to worry about this until the US and China were fighting for AI dominance, then suddenly datacenters use water and are so loud people go insane from them. What is the first time someone mentioned datacenters using water and causing pollution?
It's absolute group psychosis. Are you actually so dumb that you don't notice that a thing that has been around for 50 years suddenly is a threat to our survival? Were you born yesterday so you don't remember 2 years ago when nobody had ever realized the existential threat of building datacenters in America? Why didn't anyone notice that datacenters were pumping the wells dry in 2021? Maybe they weren't and still arent...
What is going to happen is a complete revaluation of things like "finding a counter example to a famous problem". Even if someone finds a solution to a problem like this with pencil and paper, nobody will believe it, and they will assume that there was an AI involved.
Further, sitting and doing math with a pencil and paper will no longer be a reasonable strategy to build a reputation or career, beyond the benefit a mathematician gains to their own intuition and skill. People who work hard to build intuition and also use AI effectively will dominate the field.
In a world where everyone is using AI, the open problems that remain will be the ones that are AI resistant. This is no different that how things work now, mathematicians wait until they are fairly confident someone won't rapidly solve their problem before they start talking about it. They will do the same thing in the future, except in the future AI will be part of the toolset they use decide if they are ready to share yet or not.
Edit: Ok I believe I was generally right here, but I just read the details of what OpenAI did. They didn't solve a longstanding problem, they got tipped off to an approach a mathematician was using and would likely result in the solution very soon and they finished it first. If this turns out to be true I think my take above is not correct, in the short term people will have to stop sharing updates because otherwise openai will dishonestly race to finish their work.
Also, I don't think a PHD of Industrial Sociology is really a job anyone needs to have done, so it's unlikely anyone will bother training an AI to do it. That doesn't mean I think studying Industrial Sociology is worthless, I just think it's unlikely anyone will fund training an AI to do it. Besides, is that even how AI's are trained? By some professor somewhere teaching them how to do his/her job before being replaced? That's like how it worked in Player Piano by Vonnegut, but it's not how they actually work.
For example, when you paste the first 30 lines of a famous speech, you don't want it to finish the speech, you want it to give you the identity and some analysis of what you just pasted. From what I understand, that is the reinforcement part.
Now that large AI vendors have a massive corpus of user interactions however, the lines have likely become more blurred.
Just look at git status before you commit :eyeroll:.
The deep realization is that if you can predict the next token well enough, you can do things like this:
<paste the first 10 chapters of a mystery novel>. And it turned out the killer was
And if it's really good at predicting the next token, it has to understand the novel and the clues, which means understanding the context and the language and human norms and innuendo and story telling, and tropes, and red herrings, and predict who the killer was.
I think you want it to be something more complicated. It's literally not. It just turns out predicting the next token is equivalent to a universal compression algorithm, which is a form of general intelligence. And we have almost unlimited 'labeled' data to train autocomplete.
The fact that this made it to the main page is either some kind of coordinated effort or bots.
Where you might win by owning your own hardware: - Hardware costs go up, and thus api costs go up. You've locked in your pricing. - Chinese/Open models become illegal/hard to access the way we do now. OpenAI and Anthropic are trying very hard to build a regulatory capture scheme to do this. I think they will be unsuccessful because China just won't participate.
The assumption previously used was that you can run a Sol level model on an M6 or whatever hardware $20k gives you. That is not true, it was an assumption made to show that even giving your own hardware every reasonable advantage it still loses.
Lets compare buying tokens of the best model you might run on your own hardware (still being unrealistic in favor of your own hardware) vs that same class of model on the market. I think one of the best models you might be able to run is GLM 5.4, but lets just look at chinese models generally:
$20k workstation, best case: $15k M5 Ultra 512GB, 36-month amortization, ~$440/mo. Runs a GLM-5.3-class model at ~30 tok/s. Saturated 24/7 it produces roughly 58M output tokens/month.
Buying those tokens:
DeepSeek V4 Pro @ $0.87/M $50
Kimi K2.6 @ $4.00/M $232
GLM-5.3 @ $4.40/M $255
Kimi K3 @ $15.00/M $870 (does not fit on the box)
And keep in mind I am giving 'your own hardware' no overhead or maintenance cost, despite your condition that it's in a large corporate environment. In reality corporate IT would make it almost impossible to set up and your would need huge lead times to buy the hardware and get it installed.
Even using multiple windows in parallel for as many as 5-10 hours per day, I find that I am not fully using my claude max (20x) and chatgpt pro (20x) accounts. I can for sure use up the claude max account, but chatgpt either gives me a free reset before I run out of tokens or I just fail to use the full quota. The quota for Sol seems like 10x that of Claude Opus at the same level, and forget Fable, you can use a 5 hour quota in 20 minutes.
But lets do the math:
Lets say a 20k workstation can run 1 inference at a time at the same speed you get with Sol hosted by openai (big assumption) and run an equally capable model (big assumption).
Each month this gives you about 100-170 inference hours on a Sol 20x Pro account, and 720 hours (if you utilize 24/7) on the workstation.
Assuming a 36 month amortization before the workstation has to be replaced due to no longer being able to run frontier models or is too inefficient due to electrical costs or what have you:
The monthly workstation cost is about $550 capex and $150 electricity -> $700/month
You would need about 6 Pro accounts to reach that capacity, which would cost you $1200 a month.
But this fails because:
- You most likely can't utilize the workstation 24/7. Your work hours will be concentrated into 6-10 hours per day.
- During work hours you are capable of utilizing more than 1 concurrent session. 6 Sol accounts would support as many as 20-30 during working hours, not all the time but if you could burst to that many (don't forget sub-agents and agent directed parallel agent workloads).
- In 1 year the cost of Sol level models is likely to cost a fraction of what it does now.
this leads to:
Workstation 1 Sol Pro 2 Sol Pro
Monthly cost $700 $200 $400
Raw capacity (hrs) 720 120 240
Usable capacity (hrs) 100-130 120 240
Concurrent sessions 1 3-5 6-10
$ per usable hour ~$6.00 $1.67 $1.67
Usable hours per $700 ~115 ~420 ~420Yes, this is the bet being made by everyone, together. No, nobody is actually making it.
The cost and benefit of a pipeline to train programmers is an externality. You do not benefit from training programmers as you do not employ them long enough to get a return on that investment. However you will hire some programmer hired by someone else and get the benefit of their investment. It kind of ended up as a wash, and there was enough 'shit shoveling' work to make it economical to keep some junior engineers around to keep senior engineers happy, and some of those junior engineers would learn and grow and end up as the next generation of actually productive engineers.
This entire system has already collapsed. Junior engineers are not low value and thus low wage anymore, they are highly negative in value, as before they would puzzle on something for days and finally come back with a small PR that kind-of worked but needed revision. Now they can puke out 50,000 lines of code that they don't understand in 3 hours. You'll never be able to make them sit and just do stuff by hand, and they can't supervise an AI. Having them around means you'll have to basically ignore them and not look at whatever they are trying to do, and never use any of the code, or else you will waste orders of magnitude more senior engineer time as they are forced to de tangle the mess to even understand what it does. Or you just go with it and in 3 months you have 25 million lines of code and no person or AI can make a change anymore without breaking 15 things.
There is no benefit to any company, at least in countries where you can't lock someone into a very long term contract, in hiring junior engineers anymore, so they will just stop doing this. If they did hire juniors it wouldn't help them, as they will still demand much more pay and leave if you don't give it to them the moment they can pass interviews at a more senior level. So they are just going to stop, the pipeline will dry up, and they'll pay a lot more for senior people as the supply gets more and more scarce. This is self reinforcing, the more you pay them the earlier they will retire.
By the time governments and the industry are feeling enough pain to do something about this salaries will be out of control.
If your boss doesn't care about your software being slow he's incompetent.
https://services.google.com/fh/files/blogs/google_delayexp.p...
https://business.google.com/ca-en/think/marketing-strategies...
https://medium.com/ft-product-technology/a-faster-ft-com-10e...
https://www.pingdom.com/blog/page-load-time-really-affect-bo...
I think I've worked with close to 200 software engineers close enough to get a good understanding of their abilities in my 20+ years in the industry. I think the current version of claude and chatgpt are more effective at writing code than about 85%, maybe up to 95% of them. That is to say, if you let me hire any of them for the cost of my Claude Max + Chatgpt subscriptions ($400 a month), I would probably hire less than 5% of them at that cost. I get more done, and with higher quality using Claude by myself than I would with a team of 50 average software engineers, and it takes less of my time to do it.
The industry is going to get rocked.
I think a company that right now has 10-20 full time software engineers will have 2-3 in 5 years, and realistically 2 of those 3 aren't necessary to get the coding done, they are there so you have coverage over vacations and redundancy when someone leaves.
The only reason I think offshore is going to be completely eliminated is it's too high friction to work with them. They are far away, you have to hire a firm to manage them which means hiring a manager on site as overhead, and they are awake when you are asleep (or pretending to be awake when you are awake).