You know, it’s petty, but I think one of the things I hate the most about AI is that it surpassed front end JavaScript frameworks in terms of changing what the standard is every few years.
What’s the point of learning anything if it becomes obsolete faster than the seasons change?
There are a massive number of people who want the ends of programming and geekery but lacked the means. They are overjoyed that they can abrogate the means and enjoy the reflected light of achievement, especially if it cheapens the part they feel inferior about: namely, actually understanding and learning about the systems.
As such, to them, it’s a BENEFIT if the systems change at a rate no mortal would ever agree to. Why learn anything, after all?
To be fair, technology has never been able to completely replace humans at anything unless the job was explicitly designed (factory/assembly) rather than organically discovered.
Writing software is one of those organic roles because it touches so many parts of a business. If you ask 100 devs to explain what they do, you'll get at least 100 distinct answers.
For the "organic" roles we have attempted to replace, we just keep lowering our standards. Doctors and lawyers are a good example of that. Obviously, the doctors and lawyers aren't going anywhere.
What is bound to occur to dev work is the opposite of what the doomers say. We will probably go back to the jack-of-all-trades type of dev. All the specialists/contractors and their certifications become worthless again. If the expected skillset becomes broader, that only further resists automation.
Just so you know, the frontend framework wars started and ended pretty quickly in the latter half of the 2010s.
It was the downstream effect of the browser wars settling down and W3C making big moves on getting vendors to finally implement standards that were sorely needed. It had nothing to do with web dev culture.
The most urgent needs in frontend web dev were components and state management.
Vue and React won out, and I still don't see that changing for the foreseeable future on the vast majority of corporate web apps. This remaining split is just coke vs pepsi. Basically, a meaningless distinction and stable.
I don't doubt there will always be new tools, but a lot of HN doesn't see the forest for the trees when it comes to web. So many people chime in ignorantly just to be bullies and spread FUD and self-promote. I think that strategy not going to work again for a while. Everyone is very burned out on that overplayed game in all aspects of life, not just web dev.
If anyone is interested in frontend dev, LLMs only further entrench this situation. Now is the time to be creative with mature tools and learn a thing or two. The web summer is upon us, not an autumn of dying tools nor a winter of AI replacing people.
> Just so you know, the frontend framework wars started and ended pretty quickly in the latter half of the 2010s.
Funny thing on HN. I saw people actively complaining about “JavaScript framework of the week” posts well into the early 2020s… when they were no longer really a thing.
Yea. I've evaluated these AI development patterns 3-4 times over the past year while applying for jobs.
I've been told by currently employed engineers that I was at state of the art development after a week of using them each time.
This also seems to have played out in hiring as every interview I've managed to pull has been completely schizophrenic with the interviewers wanting me to show AI fluency(for whatever that means) and then seemingly panicking mid way through as I start using the AI that can immediately solve the bog standard interview questions and then switch to having me manually build things.
They did come back for a while after the original cease and desist, but they've shut off again "due to a new development in the ongoing legal proceedings" a few days ago.
I found it a bit hard to understand this discussion because I wasn't familiar with the background. If I've got this right, his project let you run Linux on a PlayStation 5 (thus also providing the ability to run modern games).
Slop kiddies using AI discovered the PS5 hypervisor exploit it relied on for installation. Initially they agreed to his request to delay reporting it to Sony until after GTA6 came out (so users could at least install and enjoy that highly anticipated game). But then they turned around and disclosed it anyway (the same or next day?).
They presumably gained a nominal bounty, but nuked the project he'd poured so much hard work into. Including support for the PS5 Pro he was preparing to release next year.
The community used to be a group of talented researchers, but he's frustrated it's diluted into participants who aren't aligned with his values and don't even understand the hacks they produce.
This sounds like LLM noobs ruined the chance of legally playing GTA6 and habe Linux support and he seems pretty frustrated about it
> I am stepping away from the ps5 scene and stopping all my work on ps5 linux.
After pouring my heart and months of my life into it, including plans to finish ps5 pro support and release in 2027, it's all down the sink.
> The scene used to be a group of highly talented researchers, but now it is just a bunch of noobs using LLMs and writing hacks they don't even understand. Slop kiddies found the only hypervisor bug left, which I had also found a while ago, and decided to report to Sony.
> I asked them to at least wait for GTA 6 to come out so that people would have the opportunity to legally purchase the game and also enjoy linux. They agreed to wait, but not a day passed and they decided to waste it instead.
> The scene used to be a group of highly talented researchers, but now it is just a bunch of noobs using LLMs and writing hacks they don't even understand. Slop kiddies found the only hypervisor bug left, which I had also found a while ago, and decided to report to Sony.
Sounds like he is salty that somebody found an exploit using LLM that he found manually (which probably took a significant amount of time.) I can partly understand it, but I mean that's also just the game that somebody else might find an exploit, LLMs just make it easier. LLMs finding bugs is not a bad thing, just sucks for enjoyers of open source software in this particular case.
>I asked them to at least wait for GTA 6 to come out so that people would have the opportunity to legally purchase the game and also enjoy linux. They agreed to wait, but not a day passed and they decided to waste it instead.
This guy has used on working with people that understand the scene, it seems that they were in contact, they agreed on something and then the next day they did something different.
This is really important in the context of what happened and probably the cherry on top of the current "hacking" cake, with the bar being lowered so much that anybody can jump in with some minor skill and luck and get something just for themselves.
"Everyone else enjoying this got screwed, but how much did the one guy get paid do we think?"
Having the money to pour into an LLM to find this tautologically means it won't have been a life-chaning amount of money, and does not engender support for the person that reported the exploit to Sony
> Sounds like he is salty that somebody found an exploit using LLM that he found manually
Nope. He is salty because someone made public the secret exploit that Linux-on-PS5 apparently relies upon to bypass Sony's hypervisor protection. Now Sony will fix the exploit, and there will be no more Linux on the PS5 until another exploit is found.
It sounds like he had already found it the old fashioned way, then the noobs sic'd the LLMs on it, found it, and he told them "hey chill, we knew about this, don't ruin it" and they ruined it.
To quote Dr Ian Malcolm:
"If I may... Um, I'll tell you the problem with the scientific power that you're using here, it didn't require any discipline to attain it. You read what others had done and you took the next step. You didn't earn the knowledge for yourselves, so you don't take any responsibility for it. You stood on the shoulders of geniuses to accomplish something as fast as you could, and before you even knew what you had, you patented it, and packaged it, and slapped it on a plastic lunchbox, and now
[bangs on the table]"
That quote logically implies that people should be paralyzed to act on any novel information. There is no such thing as a discovery that occurs in a vacuum without some prior art. "Standing on the shoulders of giants"
I disagree. It's pretty clear the complaint is the lack of responsibility or care for the dangerous science at the center of the movie.
The complaint wouldn't be remotely the same if Hammond was taking the power he was playing with seriously. His character is the villain, he ignores every warning and almost kills his grand children.
In the case of the PS5 Linux project, the researcher who has worked towards the skills required to find these bugs has a greater respect, moved slower, had time to think.
People playing with the enormous power of LLMs finding hypervisor bugs and throwing it in the fire have no respect for the work.
"Sounds like he is salty that somebody found an exploit using LLM that he found manually"
No, he's salty that this last remaining hypervisor exploit is going to get patched, causing his PS5 Linux project to get wholly-fucked as now there will be no (currently-known) way to sideload the OS. Reading comprehension is very important here.
No guarantee at all that a new exploit will be found, and even much lower probability that a suitable lead developer will be found. The slop kiddy was just an asshole TBH.
I think you've hit the nail on the head. I've seen this behavior from others in the space before too, it's not a one-off. They refuse to use LLMs themselves, and state as much publicly "I simply don't need to resort to using the slop machine, my skills are superior", yet FUME when others they deem as "lesser" are able to use it to achieve the same results as them, oftentimes faster.
I can guarantee this post would not have been made had he not been the one to find that vuln first.
That's not the fault of the model, though. Even knowledgeable people that found everything manually can (ir)responsibly disclose it to the vendor for a payout.
Just look at the iOS jailbreaking scene and how Apple repeatedly poached all the devs until there was no one left, killing it off well before LLMs.
> The scene used to be a group of highly talented researchers, but now it is just a bunch of noobs using LLMs and writing hacks they don't even understand. Slop kiddies found the only hypervisor bug left, which I had also found a while ago, and decided to report to Sony.
I don’t know if he’s mad AI scooped a bug? Or that LLM slop kiddies don’t pay attention to that something is already known?
Reading all of the quotes it seems he's more annoyed that they disclosed the bug to Sony to collect the bounty, and that they're using AI as an afterthought
I don't find that offensive at all. People do not understand the code they make AI to generate, but that doesn't stop them from submitting. This is a real problem that has hit many open source projects, and I expect more and more of them will introduce barriers to submission.
It’s not only for coding. If my coworker send me a lengthy doc that obviously 100% written by AI without even reviewed/revised by them, I wouldn’t read because it is a waste of my time and an insult to me.
We removed a guy from a project for doing something like that. North of 100 pages, and mostly wrong. What wasn't wrong was so generic it was useless. I don't know what he's working on now, but at least it's not our stuff.
LLM use has been polarised like everything else seemingly is.
I have no issue with people using them to do things they could already do themselves as an efficiency thing and if they check and understand the result.
Unfortunately that seems to be the exception not the rule and the amount of "one shot slop" has increased dramatically, it's even affecting the modding communities for games.
This is gonna be a mess that is going to take time to figure out and then even more time to clean up.
You ignored the part about "writing hacks they don't even understand." LLM is fine if you use it as one of many tools to form your work and not a substitute for yourself.
They are identifying with this new powerful technology, because it makes them feel as if they're powerful also.
Like how drivers of monster trucks get a rush from feeling like the power of the machine is their own. Some users of LLMs exhibit the same kind of arrogance of being on the side of a powerful force, like the law enforcement or military. I imagine some gun owners get that rush too. When you take away the gun or the truck, suddenly they feel vulnerable.
I don't think he was calling everybody that is using LLMs a bunch of noobs. I think he was calling the people using LLMs to submit code they don't understand a bunch of noobs. I think if the patches were well reasoned and could be talked about in detail by their submitters, he'd have much less of a complaint.
This take is pure Luddite nonsense. AI "lowering labor value while boosting capital" ignores centuries of automation: productivity gains cut costs, expand markets, create new jobs, and raise real wages despite short-term disruption.
Steam engines, electricity, computers displaced workers but spawned far more opportunities through new industries and cheaper goods. Same pattern now.
The "jobless masses stuck with 1GB phones eating slop" fantasy is backwards. Compute keeps getting vastly cheaper and more capable; AI speeds that up.
"Terrible for indie creators and startups"? The opposite: AI obliterates barriers to building, shipping, and competing. Solo founders are moving faster than ever.
It's the same tired doomer script we get with every tech wave. It ages poorly.
None of the previous tech had the potential to do every economically productive thing we can do. It will spawn more opportunities, but maybe it will also fill those opportunities.
If the market is big enough, competitors will appear. And if the margins are high enough, competitors can always price-compete down to capture market-share.
Competitors will appear? You can't build a DRAM production facility in a year. You probably even can't in two years.
Also, "price-compete down to capture market-share"? Prices are going up because all future production capacity has been sold. It makes no sense to lower prices if you don't have the capacity to full fill those orders.
> Rare earth metals are in the dirt around the world.
They are. The problem is, the machinery to extract and refine them, and especially to make them into chips, takes years to build. We're looking at a time horizon of almost a decade if you include planning, permits and R&D.
And given that almost everyone but the AI bros expects the AI bubble to burst rather sooner than later (given that the interweb of funding and deals more resembles the Habsburg family tree than anything healthy) and the semiconductor industry is infamous for pretty toxic supply/demand boom-bust cycles, they are all preferring to err on the side of caution - particularly as we're not talking about single billion dollar amounts any more. TSMC Arizona is projected to cost 165 billion dollars [1] - other than the US government and cash-flush Apple, I don't even know anyone able, much less willing to finance such a project under the current conditions.
Apple at least can make use of TSMCs fab capacity when the AI bros go bust...
Chips also need rare doping materials, plus an absurd level of purity for the silicon. The problems are the same no matter if we're talking about chips or batteries.
It's not even the economical theory. Supply should not increase to increased demand. They want more profit, and if less supplies is what accomplishes that, they will absolutely keep the supplies constant and manufacture a scarcity. This is the economical theory.
50GB gives assurances that the BluRays are high quality (but not always. I've seen some horrible BluRay encodings...)
As long as you are going from high quality sources, you should be fine. The issue is each transcoding step is a glorified loop-(find something we think humans can't see and delete it)
In other words: the AV1 encoder in your example works by finding 47GBs of data TO DELETE. It's simply gone, vanished. That's how lossy compression works, delete the right things and save space.
In my experience, this often deletes purposeful noise out of animation (there are often static noise / VHS like effects in animation and film to represent flashbacks, these lossy decoders think it's actually noise and just deleted it all changing the feel of some scenes).
--------
More importantly: what is your plan with the 50GB BluRays? When AV2 (or any other future codec) comes out, you'll want to work off the 50GB originals and not off the 3GB AV1 compressed copies.
IMO, just work with the 50GB originals. Back them up, play them as is.
I guess AV1 compression is useful if you have a limited bandwidth (do you stream them out of your basement, across the internet and to your phone or something? I guess AV1 is good for that) But for most people just working with the 50GB originals is the best plan
> In other words: the AV1 encoder in your example works by finding 47GBs of data TO DELETE.
With that reasoning, lossless compression of .wav to .flac destroys >50% of data.
In actuality, you can reconstruct much of the source even with lossy compression. Hell, 320kbps mp3 (and equivalent aac, opus, etc) are indistinguishable from lossless and thus aurally transparant to humans, meaning as far as concerns us, there is no data loss.
Maybe one day we'll get to the point where video compression is powerful enough that we get transparent lossy compression at the bit rates streaming services are offering us.
> In my experience, this often deletes purposeful noise out of animation
AV1 specifically analyzes the original noise, denoises the source then adds back the noise as a synthetic mask / overlay of sorts. Noise is death for compression so this allows large gains in compression ratio.
> AV1 specifically analyzes the original noise, denoises the source then adds back the noise as a synthetic mask / overlay of sorts. Noise is death for compression so this allows large gains in compression ratio.
If said noise still exists after H265.
And there's no guarantee that these noise detection algorithms are compatible with H264, H265, AV1, or future codecs H266 or AV2.
AV1 is not about throwing away more data that the human can’t see. It’s about having better tools.
1. the prediction tools of AV1 are better than those of h265. Better angular prediction, better neighboring pixels filtering, an entirely new chroma from luma prediction tool, an intra-block copying tool, more inter prediction tools, non-square coding units.
2. If the prediction is better, the residuals will be smaller.
3. Those residuals are converted to frequency domain with better tools for AV1 as well (more options than just DCT), so that you have a better grouping of coefficients close to the DC component. (Less zeros interleaving non-zero values.)
4. Those coefficients compress better, with a better entropy coding algorithm too.
You can have exactly the same video quality for h265 and AV1 yet still have a lower bitrate for the latter and with no additional decision made to “find out what humans can’t see.” The only place in the process where you decide to throw away stuff that humans can’t see is in the quantization of the frequency transformed residuals (between step 3 and 4) and the denoising before optional film grain synthesis.
To be clear: you can of course only go down or stay equal in quality when you transcode, due to rounding errors, incompatible prediction modes etc. That’s not under discussion. I’m only arguing about the claim that AV1 is better in general because you throw away more data. That’s just not true.
Yes, in general you find the best high quality source you can get your hands on and then compress that. For us lay people, that would currently be any 4k videos with a high bitrate. In such cases, it doesn't matter much that it is already compressed with AVC or HEVC. Sure, when you compress that again at a lower bitrate, there will some loss of data or quality. But honestly, it doesn't make a discernable difference (after all, you decide what is the video quality acceptable to you by choosing how much more to compress). Ideally, if DVD and Blu-Rays lasted long, we would all just be saving our videos on it. (Assuming there will be any Blu-Ray readers, 10+ years down the lane).
There's an elephant in the room: why is maintaining a web browser costing $400M/y?
The web standards are growing faster than non-profit engines can implement them.
Google & Apple are bloating the web specs in what looks like regulatory capture.
If Blink/Webkit dominate for long enough, they will lock everything down with DRMs & WEI. Maybe it's time to work on lighter protocols like Gopher & Gemini that don't need 20GB of RAM to open 20 tabs ?
Having a "lighter standard" simply means people will have to write native apps, one per platform. I understand Apple wants this, but for Mozilla that should be the antithesis of what they're trying to achieve.
> There's an elephant in the room: why is maintaining a web browser costing $400M/y?
Is that actually the case though? I find it hard to believe that Mozilla has anywhere close to 1500 senior developers working on just Firefox. My guess is that the bulk of that money is spent on unrelated adventures and overhead.
Google Chrome is likely around $400M, while Mozilla's core browser team is around $200M but are technologically far behind. Hard to find precise numbers, it's just an order of magnitude estimate
In what sense is Mozilla behind? Chrome is an advertising delivery platform, they have fundamentally different goals and that $400M they spend on Chrome is not mostly going to technology that I want in my browser, that's the point. Just because Chrome builds telemetry features doesn't mean Mozilla needs them too.
There are too few examples to say this is a trend. There have been counterexamples of top models actually lowering the pricing bar (gpt-5, gpt-3.5-turbo, some gemini releases were even totally free [at first]).
I love Kagi's implementation: by default it's disabled, you either have to add a question mark to the search, or click in the interface after searching to generate the summary.
This is absurd. Training an AI is energy intensive but highly efficient. Running inference for a few hundred tokens, doing a search, stuff like that is a triviality.
Each generated token takes the equivalent energy of the heat from burning ~.06 µL of gasoline per token. ~2 joules per token, including datacenter and hosting overhead. If you get up to massive million token prompts, it can get up to the 8-10 joules per token of output. Training runs around 17-20J per token.
A liter of gasoline gets you 16,800,000 tokens for normal use cases. Caching and the various scaled up efficiency hacks and improvements get you into the thousands of tokens per joule for some use cases.
For contrast, your desktop PC running idle uses around 350k joules per day. Your fridge uses 3 million joules per day.
AI is such a relatively trivial use of resources that you caring about nearly any other problem, in the entire expanse of all available problems to care about, would be a better use of your time.
AI is making resources allocated to computation and data processing much more efficient, and year over year, the relative intelligence per token generated, and the absolute energy cost per token generated, is getting far more efficient and relatively valuable.
Find something meaningful to be upset at. AI is a dumb thing to be angry at.
I’m curious where you got any of those numbers. Many laptops use <20W. But most local-ai inferencing requires high end, power hungry nvidia GPUs that use multiple hundreds of watts. There’s a reason those GPUs are in high demand, with prices sky high, because those same (or similar) power hungry chips are in data centers.
Compared to traditional computing it seems to me like there’s no way AI is power efficient. Especially when so many of the generated tokens are just platitudes and hallucinations.
> The agreed-on best guess right now for the average chatbot prompt’s energy cost is actually the same as a Google search in 2009: 0.3 Wh. This includes the cost of the answering your prompt, idling AI chips between propmts, cooling in the data center, and other energy costs in the data center. This does not include the cost of training the model, the embodied carbon costs of the AI chips, or the fact that data centers typically draw from slightly more carbon intense sources. If you include all of those, the full carbon emissions of an AI prompt rise to 0.28 g of CO2. This is the same emissions as we cause when we use ~0.8 Wh of energy.
How concerned should you be about spending 0.8 Wh? 0.8 Wh is enough to:
Stream a video for 35 seconds
Watch an LED TV (no sound) for 50 seconds
Upload 9 photos to social media
Drive a sedan at a consistent speed for 4 feet
Leave your digital clock on for 50 minutes
Run a space heater for 0.7 seconds
Print a fifth of a page of a physical book
Spend 1 minute reading this blog post. If you’re reading this on a laptop and spend 20 minutes reading the full post, you will have used as much energy as 20 ChatGPT prompts. ChatGPT could write this blog post using less energy than you use to read it!
W stands for Watts, which means Joules per second.
The energy usage of the human body is measured in kilocalories, aka Calories.
Combustion of gasoline can be approximated by conversion of its chemicals into water and carbon dioxide. You can look up energy costs and energy conversions online.
Some AI usage data is public. TDP of GPUs are also usually public.
I made some assumptions based on H100s and models around the 4o size. Running them locally changes the equation, of course - any sort of compute that can be distributed is going to enjoy economies of scale and benefit from well worn optimizations that won't apply to locally run single user hardware.
Also, for AI specifically, depending on MoE and other sparsity tactics, caching, hardware hacks, regenerative capture at the datacenter, and a bajillion other little things, the actual number is variable. Model routing like OpenAI does further obfuscates the cost per token - a high capabilities 8B model is going to run more efficiently than a 600B model across the board, but even the enormous 2T models can generate many tokens for the equivalent energy of burning µL of gasoline.
If you pick a specific model and gpu, or Google's TPUs, or whatever software/hardware combo you like, you can get to the specifics. I chose µL of gasoline to drive the point across, tokens are incredibly cheap, energy is enormously abundant, and we use many orders of magnitude more energy on things we hardly ever think about, it just shows up in the monthly power bill.
AC and heating, computers, household appliances, lights, all that stuff uses way more energy than AI. Even if you were talking with AI every waking moment, you're not going to be able to outpace other, far more casual expenditures of energy in your life.
A wonderful metric would be average intelligence level per token generated, and then adjust the tokens/Joule with an intelligence rank normalized against a human average, contrasted against the cost per token. That'd tell you the average value per token compared to the equivalent value of a human generated token. Should probably estimate a ballpark for human cognitive efficiency, estimate token/Joule of metabolism for contrast.
Doing something similar for image or music generation would give you a way of valuing the relative capabilities of different models, and a baseline for ranking human content against generations. A well constructed meme clip by a skilled creator, an AI song vs a professional musician, an essay or article vs a human journalist, and so on. You could track the value over context length, length of output, length of video/audio media, size of image, and so on.
Suno and nano banana and Veo and Sora all far exceed the average person's abilities to produce images and videos, and their value even exceeds that of skilled humans in certain cases, like the viral cat playing instrument on the porch clips, or ghiblification, or bigfoot vlogs, or the AI country song that hit the charts. The value contrasted with the cost shows why people want it, and some scale of quality gives us an overall ranking with slop at the bottom up to major Hollywood productions and art at the Louvre and Beethoven and Shakespeare up top.
Anyway, even without trying to nail down the relative value of any given token or generation, the costs are trivial. Don't get me wrong, you don't want to usurp all a small town's potable water and available power infrastructure for a massive datacenter and then tell the residents to pound sand. There are real issues with making sure massive corporations don't trample individuals and small communities. Local problems exist, but at the global scale, AI is providing a tremendous ROI.
AI doombait generally trots out the local issues and projects them up to a global scale, without checking the math or the claims in a rigorous way, and you end up with lots of outrage and no context or nuance. The reality is that while issues at scale do exist, they're not the issues that get clicks, and the issues with individual use are many orders of magnitude less important than almost anything else any individual can put their time and energy towards fixing.
You are cleary biased.
A complex chatgpt 5 thinking runs at 40 Wh per prompt. This is more in line with the estimated load that ai needs to scale. These thinling models wpuld be faster but use similar amount of energy. Humans doing that thinking use far fewer jpiles than gpt 5 thinking. Its not even close.
your answer seems very specific on joules. Could you explain your calculations, since I cannot comprehend the mapping of how you would get a liter of gasoline to 16.8m tokens? e.g. does that assume 100% conversion to energy, not taking into account heat loss, transfer loss, etc?
(For example, simplistically there's 86400s/day, so you are saying that my desktop PC idles at 350/86.4=4W, which seems way off even for most laptops, which idle at 6-10W)
I believe it is the system instructions that make the difference for Gemini, as I use Gemini on AI Studio with my system prompts to get it to do what I need it to do, which is not possible with gemini.google.com's gems
I always joke that Google pays for a dedicated developer to spend their full time just to make pelicans on bicycles look good. They certainly have the cash to do it.
reply