Coding with LLMs (Claude Code, OpenAI Codex) is often presented as the ‘killer app’ for Generative AI. But looking at data, it seems the one piece of the puzzle missing is actual cost. …
Many applications are suboptimal to say the least but what’s been done with alpha fold and recently in mathematics is very far from a scam. Not to bring up what’s also been accomplished in cyber security. These models are proving open problems that have been around for decades and finding serious vulnerabilities. The issue is consistency and efficiency. Of course the other issue in making them stronger is continual learning and long horizon planning. I think too much investment came in too quickly and what is provided to the masses currently isn’t consistent or efficient enough. That said as a math and comp sci grad and someone who works in the field it’s been absolutely mind blowing to watch what’s already been done. In 2010 the concept of an artificial mind solving something like the Erdős unit distance conjecture would have been seen as pure sci-fi, maybe something we would achieve closer to 2100 than 2026.
For reference, it took Uber about 17 years to become profitable and Spotify 18. They were hemorrhaging cash for over a decade and a half before finally hitting their stride. As for the current AI development it’s honestly from 2017 when the white paper on transformers came out where shit started getting serious, so it’s been about 9 years since investors were serious. Before that point it was all passion projects, absolute moon shots as they call them.
Both Uber and Spotify (and AWS too) had economics of scale going for them - the more users they have, the more the infrastructure could be leveraged. This does NOT work for LLMs. More users means using more compute, more advanced tasks (like coding) uses exponential amounts of compute. A single user running a complex task can make 8 Blackwell GPUs run full tilt, and you don’t even have any guarantee that the output will be useable.
There are a few narrow areas where LLMs might be successful, like scanning for security vulnerabilities or searching large amounts of documents. The massive amount of money invested will never be recouped with these usage scenarios.
I don’t think anyone is assuming it will stay at its current efficiency and there will be zero improvements. A lot of the everyday AI use cases will likely be pushed to someone’s personal device aka your phone. In the same way a lot of Uber and Spotify is handled by your personal device today. What we’ve seen for years now is the development of these gargantuan models that are then condensed down into much smaller models with 90%+ of the same effectiveness. Simultaneously we will see and are seeing devices sold with better NPU’s for edge compute for AI the same we’ve seen the push for more edge compute to manage other services such as Uber and Spotify.
Across this thread and others there’s like this implicit assumption AI will never progress beyond where it is right now in spite of the evidence of its almost exponential growth. It’s really interesting.
Although, most people aren’t talking about Alphafold when they’re talking about AI. They’re usually specifically referring to the generative transformer models that are currently all the rage.
I doubt anyone would care too much about a linear regression model, or multi-layer peceptron , for example.
I’m pretty sure a lot of them don’t know the difference or understand how mind breaking it is that some of these achievements are happening. Of course alpha fold is old news and the solutions to the Erods problems is something that should be raising eyebrows. These models are fundamentally just math and a model that’s better than humans at math can theoretically design a stronger model than us.
To clarify, my argument is that you don’t know what you’re talking about.
Erodos unit distance conjecture is a proposed solution to a Erodos unit distance problem. What the LLM model did was disproving Erodos unit distance conjecture, not solving it (you don’t solve a conjecture), nor solving the problem (that remains unsolved).
Again, you seem poisoned by following news media cycle without understanding what they talk about.
Multiple new vectors of attacks, automation of attack pipelines…
Many applications are suboptimal to say the least but what’s been done with alpha fold and recently in mathematics is very far from a scam. Not to bring up what’s also been accomplished in cyber security. These models are proving open problems that have been around for decades and finding serious vulnerabilities. The issue is consistency and efficiency. Of course the other issue in making them stronger is continual learning and long horizon planning. I think too much investment came in too quickly and what is provided to the masses currently isn’t consistent or efficient enough. That said as a math and comp sci grad and someone who works in the field it’s been absolutely mind blowing to watch what’s already been done. In 2010 the concept of an artificial mind solving something like the Erdős unit distance conjecture would have been seen as pure sci-fi, maybe something we would achieve closer to 2100 than 2026.
For reference, it took Uber about 17 years to become profitable and Spotify 18. They were hemorrhaging cash for over a decade and a half before finally hitting their stride. As for the current AI development it’s honestly from 2017 when the white paper on transformers came out where shit started getting serious, so it’s been about 9 years since investors were serious. Before that point it was all passion projects, absolute moon shots as they call them.
Both Uber and Spotify (and AWS too) had economics of scale going for them - the more users they have, the more the infrastructure could be leveraged. This does NOT work for LLMs. More users means using more compute, more advanced tasks (like coding) uses exponential amounts of compute. A single user running a complex task can make 8 Blackwell GPUs run full tilt, and you don’t even have any guarantee that the output will be useable.
There are a few narrow areas where LLMs might be successful, like scanning for security vulnerabilities or searching large amounts of documents. The massive amount of money invested will never be recouped with these usage scenarios.
I don’t think anyone is assuming it will stay at its current efficiency and there will be zero improvements. A lot of the everyday AI use cases will likely be pushed to someone’s personal device aka your phone. In the same way a lot of Uber and Spotify is handled by your personal device today. What we’ve seen for years now is the development of these gargantuan models that are then condensed down into much smaller models with 90%+ of the same effectiveness. Simultaneously we will see and are seeing devices sold with better NPU’s for edge compute for AI the same we’ve seen the push for more edge compute to manage other services such as Uber and Spotify.
Across this thread and others there’s like this implicit assumption AI will never progress beyond where it is right now in spite of the evidence of its almost exponential growth. It’s really interesting.
Although, most people aren’t talking about Alphafold when they’re talking about AI. They’re usually specifically referring to the generative transformer models that are currently all the rage.
I doubt anyone would care too much about a linear regression model, or multi-layer peceptron , for example.
I’m pretty sure a lot of them don’t know the difference or understand how mind breaking it is that some of these achievements are happening. Of course alpha fold is old news and the solutions to the Erods problems is something that should be raising eyebrows. These models are fundamentally just math and a model that’s better than humans at math can theoretically design a stronger model than us.
Tell me you listen to media news cycle without understanding what that actually mean without telling me that.
That’s not exactly what happened, isn’t it.
Multiple new vectors of attacks, automation of attack pipelines…
Got anything besides my quotes to give your argument credibility?
To clarify, my argument is that you don’t know what you’re talking about.
Erodos unit distance conjecture is a proposed solution to a Erodos unit distance problem. What the LLM model did was disproving Erodos unit distance conjecture, not solving it (you don’t solve a conjecture), nor solving the problem (that remains unsolved).
Again, you seem poisoned by following news media cycle without understanding what they talk about.
Like literally just put that into Google, it’s not some study that proves it, it’s the multiple ones, and every cybersecurity expert talking about it. But if you want a one source you want to argue about, then https://blog.checkpoint.com/research/global-cyber-attacks-rise-in-january-2026-amid-increasing-ransomware-activity-and-expanding-genai-risks/