Bill didn’t say anything. Prices won’t go down because AI replaced humans. If anything they will rise, as they are.

Oh hey both of these guys can get a data center shoved up their asses sideways

Prices will definitely rise to cover operating costs and to recoup all the investment costs.

I honestly hope the whole thing crashes once people realize they can run their own LLM on their own hardware.

Data centers can hum very loudly, I wonder if they vibrate too

I have located the PLUG and I'm ready to PULL.

Just say the word.

I still can't decide if these people are delusional or I am.

Every single time I use an LLM it fucking "lies" to me or otherwise completely fails at the task. The people talking like this seem to me like they've never actually used it, or haven't actually vetted the accuracy (like most AI users).

Maybe I'm just not using the "good stuff". Or I'm not imaginative enough to foresee a near future where these problems are actually corrected and it becomes trustworthy.

I've never been so torn by a technological prediction.

And what hardware is that, exactly? Speaking as someone with what I'd consider to be a pretty decent homelab, I can't self host the kind of AI that I use for work, not without literally emptying my savings into the hardware and energy needs. AI is already expected and required in my job, so the only move forward is to give elon money for access to cursor, which my company is already doing. If the company were to self-host, they'd prolly still use one of those shiny new datacenters since there's real value in offloading hardware ownership to a third party, meaning those never go away and they still dictate the cost of using AI. I have a hard time believing this will happen in most companies simply because they already shelled out tons of money for SAAS software that has been self-hostable for over a decade, so why would you think they'd stop and think "hmm maybe we could cut costs by taking over the hosting and maintenance of the AI ourselves"?

I think we're at least a decade away from having user-accessible hardware for AI that doesn't break the bank. I don't think we have a decade to spare, however.

AI is already expected and required in my job

then you should be looking for a new job... unless you happen to be on a very specific niche, any company so committed to current AI that cannot survive without, will implode in short time

I'm curious what you are using. The free versions of chatgpt have been like that for me, but even Gemini flash with extended thinking, also free for a while longer, is giving me pretty reliable results as long as there training data out there to derive an answer from. The higher (paid) Claude models will one shot most coding tasks.

GATES: Go to San Francisco and ask people "would you rather ride in a Waymo or rather ride with a human driver?"

So, he's right about that—I've seen past polling that shows that people in the Bay Area do say that they prefer to not have to deal with a human driver and would prefer the robotic taxi over the human-driven one. But it's also true, unless things have changed recently, and I have not been monitoring the situation, that Waymo's rates were higher, and so it's also not presently an even comparison. You have to pay a premium for the Waymo.

That being said, my bet is that a lot of Waymo's costs are fixed, so those costs will drop on a per-ride basis as their fleet scales up. And they are expanding.

Understanding is a long time coming. It will probably take businesses realizing they are stealing all their trade secrets when using LLMs for any meaningful information to come out.

I've not yet fucked with Claude. I don't want to pay for it, and I really don't like the surveillance aspect of these centralized systems. Mostly I'm using Gemini, whatever DDG had in their search results, and local models I've been fiddling with (like Qwen3.8 right now).

All more or less garbage once I get into the details of anything on the edge of my expertise.

Give us an example of some of the prompts you're using and what LLMs. I'm curious if it's a use case difference or you're using the dollar store's customer service AI to try to help you with your coding homework.

I've been using OpenCode with whateverthefuck free models they have listed on there and they all seem to do fine with agentic tasks like building me scripts or executables to make my work tasks easier.

I used Gemini at the start with "Frontier Knowledge" and it seemed to do worse than the ones listed on OpenCode, but maybe that's because i could only do like three prompts a week since I refuse to pay into an AI.

end of the day, its just LLMs writing code for me, but I cannot see how this would be useful for a large scale code base, but also #NotAProgrammer.

I've been able to find a workflow where it's mostly correct and can handle most of my gamedev related coding without making too many mistakes. I still have to actually read through the code and pay some attention to what it's doing, and if it misses something and goes on a wild goose chase, it's unusable and I have to start over (so someone who didn't know what they are doing would be cooked), but whatever.

Sure, it does require a lot of looping adversarial reviews, and my average token cost is around 3000$ (4B tokens) a month (we have unlimited budgets and a pretty accurate tracking, and also probably cheaper than consumer prices per token with how large company it is), which is actually more than my monthly salary, but it's just a job, for a company and on a product I don't really care about, and I can 1) keep slacking in my job while doing my own coding stuff and projects, and keep seeing how absolutely unreasonable the prices are if you want to get at least semi-submitable results.

You could ask if it's necessary to spend so much tokens - but so far most adversary review rounds do find serious blocker bugs the first implementation had, most of them being some hidden stuff that would be difficult to reasonably find by hand unless you really understand everything around the code, which you should, but the point of AI is that you shouldn't have to.

Is it worth it? Lol, no. The whole team is loosing codebase knowledge, we're getting bottle-necked by pending PR reviews that are just stacking up and no one wants to do, so we're not even more effective, the cost is absolutely absurd and in no way near sustainable or worth it. And the longer it goes on, the less I can realize it's spewing bullshit and I should stop it and nudge it into a different direction in our codebase, because I'm slowly loosing touch with it, while for now I'm still relying on what I remember from before we went all-in on AI.

And that's while the whole industry is in the "Uber pricing" phase, so it will get a lot worse. But yeah, if you can burn 100-200$ per a simple implementation task, then it can have a pretty usable results. And that 3000$ a month does not include our CI review bot, that does additional rounds of multi-agent council reviews. And I'm not even working on anything complex, mostly just various menu screens UI. I can imagine the bill getting a lot higher if you get into more involved systems.

Here's a very common response to anyone that suggests they had a bad time with LLMs. You just aren't using the right model. You didn't ask the right questions. You didn't give enough context in your prompt.

It's not the fault of this infallible AI, it's PEBKAC.

Nonsense.

Speaking as someone with what I’d consider to be a pretty decent homelab, I can’t self host the kind of AI that I use for work, not without literally emptying my savings into the hardware and energy needs.

I think we’re at least a decade away from having user-accessible hardware for AI that doesn’t break the bank.

So, for coding, which is what Gates was specifically talking about being doable now, maybe we could corral up the hardware. Like, maybe one could cover specific fields.

There are still going to be issues like power and cooling and hardware cost and whether people who make competitive models are even interested in providing it for home use (since it makes it harder for them to make a return on their model). But set that aside.

As I've said before on here, I would say pretty confidently that we will not have local models running to do all of the stuff that cloud compute is used for or is being built out to for at least something like four to five years, and that's if we started immediate, massive buildout of memory fabrication to a much greater degree than we have. You cannot build a new memory factory in less than that timeframe, and we will not have that capacity with existing factories. The majority of fabricated memory now is going to cloud AI use, and cloud AI hardware will have considerably higher capacity utilization than hardware at home. You'd have to have many times over as much memory being produced to have the same compute capacity at home.

I'm not opposed to doing LLMs or parallel compute at home at all. I have a 128GB Framework Desktop and an XT 7900 XTX that I got to do just that. I'm just saying that we are not going to realistically be able to move all of the stuff in the cloud to the home for at least something like half a decade, and very probably more, because humanity does not have the memory available and can't build enough memory fabrication capacity for it in that timeframe. It doesn't matter how much value is being provided by some home user of that hardware or what their willingness is to spend on it if we don't have the memory. Like, even if every person in the world could produce, to pull a number out of the air, a real $1M in value every year via use of a home AI rig, even if all that demand suddenly materialized out of thin air, all that would happen is that prices would rise sufficiently to make the hardware unaffordable even at those extreme levels. The constraint is on the supply end, not the demand end.

How do you verify that everything it tells you is correct?

Ooh I'm way ahead of you. I had my pitchfork gold plated, just for this occasion. I'm ready NOW!

Human drivers, fuel, car parts, and compute are all non-fixed.

Have fun finding a coding job that doesn't expect you to use AI. My entire circle of contacts on the field are already there, so right off the bat my main way of getting a job would be crippled.

Uh... I was just curious because sometimes it's fun to see how the LLMs screw up. e.g. rocks on pizza.

It's not the lack of evidence presented, it's the person that asked the question that's the problem.

No you weren't. You literally suggested I was using a "dollar store’s customer service AI to try to help [me] with [my] coding homework."

Claude can one shot tasks until you get a larger system then it completely shits itself. These models are nothing more than autocomplete, and they can’t hold large systems in their heads. Anthropic literally tried to rewrite all of bun using Claude, they said they did it and yet it still hasn’t released six months later.

You check the links/references it gives you.

Gemini does a pretty good job of this because it doesn't seem to have much built-in knowledge. Instead, it just searches the Internet on your behalf and returns summarized results with links to where it got that specific information.

I use it to search for scientific research all the time and the summaries often aren't detailed enough so I actually click on those links. I've yet to encounter a situation where it fucked that up (invented links that don't exist) but I have heard about it happening.

So far, the summaries have seemed to be pretty spot-on when it comes to biology papers 🤷

I use AI all the time to parse log files for errors. Or write up simple scripts to do things that are one-offs or test of concept. Most, if not all succeed. I made a parser that translates the config of one brand of switches to another one, worked perfectly.

In what way are you using an LLM? It sounds like you're asking it moral questions, to which it of course can't give you an answer in any sort of objective sense.

Using a dollar store's AI to help with coding homework would be hilarious! Just like rocks on pizza.

You seem to be placing me in the wrong bucket. I use AI professionally, yes, but it also pisses me off pretty regularly. I'm not some "AI Bro" or whatever TF you're thinking. Just some guy who finds AI mistakes to be funny and I want to reproduce them.

Yeah the IT department of my company had an all hands meeting today and I was a little taken aback by how much AI push there was, how much devs are expected to use it, and how much it’s costing us. They say it saves time, but there was a notable drop in app quality over the last few months, I can’t say it was caused by devs using AI for sure, but it is the most obvious answer considering there hasn’t been a big staff turnover in the same time period…

Rebuilding a collapsed society does require a lot of work.

Understandably you won't be running meta level Machine models, but running olama and downloading a open LLM model can get users started for simple things they might have already been doing.

I am running a fully local model for my smart home needs, since Google's killed of Assistant and is now pushing Gemini down my throat, I figured I would just do it my self.

For anyone interested take a look at NetworkChuck on YouTube. His videos are a little to sensational for me, but he does have some interesting topics covered from time to time.

https://www.youtube.com/watch?v=QQEgIo4Juxg

By writing instructions to insist that it double verifies every (non obvious) claim with a minimum of two independent sources. I also told mine to always assume that the initial prompt is missing crucial context, and to ask as many follow-up questions as necessary until it has enough information to provide the answer to the question I'm really asking. (For speed and efficiency you can even make it give you multiple choice options to click on.) Because sometimes the problem isn't with the LLM, but with the user asking the wrong questions.

Using those two instructions alone, I've encountered considerably fewer hallucinations, and when I'm still not certain, I can simply click on the sources linked next to every single claim the AI makes.

Who's saying it will create more jobs? I've literally never heard that claimed anywhere. Someones probably saying it, presumably, but that's not at all even once what I've heard.

Either AI will prove incapable of replacing us, and we keep the same amount of jobs as we currently do, or it will replace us. Horses never really found that much work after we developed the combustion engine...

( ͡° ͜ʖ ͡° )

It definitely has an element of garbage in, garbage out.

Prompting an AI is very much a garbage in, garbage out type thing. Just like with any tool, you need to know how to use it properly to get the results you want.

Hell, some of the things I've seen people ask AI would confuse a human too.

Your experience is pretty unique then.

Yes, LLMs make mistakes, but even small, self-hosted ones are pretty efficient today if you prompt them well. They're not mind reading software so you need to be able to describe the task and HOW you want it done, not just barf in some basic instructions like "write me a copy of Facebook but better".

I think most people are so disorganized in their thinking, they can't "prompt" well. There's a lot of unclarified assumptions and leaps in how many people communicate

The bar is not that AI needs to be right. It just needs to be more right than the employee willing to do the job for the price it costs.

People have been bad at their jobs for years. Now computers can be to.

Like back when we had to teach people how to google shit. Tedious.

I'm in college and for one of my classes we actually have a prompt we feed to ChatGPT so that ChatGPT creates an excel spreadsheet for us as a step in an assignment process. The whole spreadsheet.

The class is for healthcare and I won't get more specific, but yes, you can use AI today to make whole ass Excel spreadsheets.

(One of the steps in that particular assignment involves fact-checking the spreadsheet, btw, but about 95% of the time all the answers are word-for-word in the book, 4% they're taken from the web and have more up-to-date info than the book has, and 1% is blatant lies.)

Where I am forced to encounter advanced LLMs is every company rolled out replacements for my "dumb" assistants. Like Alexa on an echo dot or Gemini on Android auto.

What I ask them to do does not take significant processing power. I need you to set a 5min timer. Not talk to me like you're alive. Don't overthink things, just do simple things.

All that extra banter they added in eats up processing power. So now it's slower and works like shit because it's formulating some complex response to my request to beep in 5 minutes. Just set the timer, my hands are covered in raw chicken juice.

No amount of prompt "engineering" will help when they outright make shit up.

Eh idk I still have it make mistakes constantly.

Recent example: I asked AI to help me come up with a search/replace string for editing a file in vim. My prompt was something like “I am editing a text file in vim and I need to replace with . How can I do this with :s/ ?” And it spat out something that didn’t work. It was not valid syntax (for vim, I think it was valid standard regex, or close to it), and it didn’t quite follow the pattern I was trying to describe (although ofc that could be my fault to some extent). But it was useful in that it pointed me in the right direction to come up with a correct formula with some follow up googling and experimentation.

That has been very typical of my experiences with AI. Useful sometimes, but absolutely not “does everything to the point you don’t have to think about it” that seems to be a common opinion.

For context I’m using ~20GB local models, so not Claude, which people who pay for LLMs swear by.

He doesn't take about what is, he talks about what's coming.

Right, but that sounds like you implicitly trust the LLM and only verify statements when they seem off. So your intuition is the final arbiter if truth?

No matter how much I carefully structure a prompt, define specific behaviors in skills, and tweak the agent md files it will still just go do something I don't tell it to or not do something it's got really specific instructions for. We have to write all code by LLM now at work and I'm trying to do my due diligence to review code before putting it up for PR. 9 times out of 10 when I tell it to show me a diff before committing it silently runs git diff in the background and prints "that's the full diff". That's with some basic "here's what I want when I ask for a diff" in the base context.

That's a fair point. I just don't know if the future they see can become a reality.

I got a free year of perplexity.ai which gives me limited access to claude sonnet and yeah, its honestly pretty good.

Whenever I visit SF for work once a quarter it feels like there is a cult running the city. So many cringe ads telling you that meta glasses will let you be more in the moment and some XY product will make AI work for you without really telling you anything about the product. Yet they can't even keep their elevators at the transit stops working cuz fuck the poor and disabled. Point is, SF has a weird sense of priorities and I don't really trust the common wisdom coming from that place. I love Oakland though, some of the best food to be found there