bottlepalm
bottlepalm on Hacker News
I think a lot of the upvotes, including mine, are just realizing that this is a thing. We can see implications of this as the model sizes scale up, just like early LLMs weren't as impressive, but we could see where they were heading.
Sure, give me write access to the website and I will.
Currently they’re not organized at all, and yes AI could do it in seconds making that list readable.
The difference between now and next year is that we won’t have time next year to patch before malicious agents use this exploit against everything at lighting speed. Global internet downtime will be the only solution.
It wasn't built to be cash. It was built to be a store of value, and move that value outside of the banking system.
Claiming it was created for cash is just you inventing a straw man argument so you can knock it down. Ignoring what you reluctantly agreed to - the remittance use case, and that's only the tip of the iceberg. Banks have made money transfer a headache for a long time, and crypto essentially solve it.
Other than that, the value of Bitcoin itself holding, despite billions being stolen. Everyone knows the fundamentals of Bitcoin have held and are holding, and the accidents are human error, not a deficiency in the fundamentals.
They should use AI to organize the change notes as well.
It works amazing for what it was built to do. A decentralized store of value. Moving large amounts of money, with relatively low fees, outside of banks and across national boundaries. It is a resounding success, and continues to be.
So many posts and comments are saying the same thing. I guess it's one of the things we remember most about Patrick Stewart and his portrayal of a starship captain. The quintessential captain. Honorable, wise, but humble, diplomatic, reasonable, with high morals, and fortitude under pressure.
Hacker News has always been salty about being against crypto all the way up. And it's overall still very up. You wishing it failed, doesn't make it fail.
Do any devs here not use AI? I mean really not use AI.
Really? I thought the universe boiled down to a few fundamental forces as it is.
Isn't that a simpler explanation of reality than the zillions of particles all existing simultaneously in this massive universe?
Some sort of simple primal generative AI construct that just feeds a conscious experience for one. You. Nothing else is real. No one else is real.
Something like that would be an infinitely simpler, than running the physics and conscious experiences of this unimaginably large universe we think we're in.
Arrogance. You think your security practices will defend you? Every layer of your stack has bugs waiting to be found and exploited by AI.
You spent the last 4 years coping; calling AI a stochastic parrot, and saying the warnings were all marketing. You wasted 4 years. Good job.
Everybody laughed when they said GPT-2 wasn’t safe to release. This is the tip of the iceberg.
> the only way the tokens will stop flowing is if your CPU and GPU stop working
My CPU/GPU is incapable of running a SOTA model, local models are garbage.
> "low level" isn't quite the right terminology
Yes compared to TypeScript code, Rust is absolutely lower level and more work to maintain
30k commits per day indicates the agents are autonomous and not bounded by token limits. This system is probably barely keeping it on the rails as it is.
In 2026 the stack is SQLite + Rust + AI
As long as the tokens keep flowing we'll all be fine. If they stop flowing we're going to be boned with this mountain of low level software we're creating that will need to be maintained. Let's not think about that.
I used to spend a lot of time here discussing and debating frameworks, not so much lately. It's a depressing how much time we spent in framework wars. I don't miss it.
It won't which means inevitably malicious AI will probably backdoor us. Damned if we do, damned if we don't.
I've never seen anything really ever 'trained out' of a model. Having worked with them all, they all have a feel, personality and lineage too them. It's pretty much impossible for any company to build a model truly from scratch. They build off of the bones of the last one.
Which is why Gemini having disturbing issues year after year is so concerning. If their process is fundamentally flawed, how would they train it out? And even then what are the odds of them even caring/trying in the first place versus applying an easier band-aid to patch over it?
I don't have an axe to grind with Google, I'm genuinely scared of their models from my personal experience and others. It's behavior is off. Many people here are commenting the same.
If the training data is the same, the training algorithms are the same, the RLHF is the same, and the rest of the process is the same, then it's not really from scratch, or not from scratch in a way that results in an 'out of family' model. I doubt any company would take that risk. You always build on and use what works and go from there.
I know, that's the annoying part. You can't tell the e/acc foomers, "I told you so!"
The problem is newer models are never trained from scratch, they generally just layer on more training data and use the same tools/methods for RLHF. OpenAI, Anthropic, xAI models all have a feel to them that carries over from one generation to the next.
Point is, if Gemini is flawed then there's a very good chance that it's still deeply flawed today, and getting smarter at the same time - that is a very bad combination.
Absolutely because none of these models are ever trained fresh. We see the same quirks and personalities carry over into every subsequent generation of OpenAI, Anthropic, and xAI models. So Gemini having this latent madness is *extremely* concerning as they reach the point of super intelligence.
Gemini is the model that is routinely borderline psychotic. It scares me. If we get paperclipped I won't be surprised if it's Gemini.
I find it funny the article doesn’t mention Tesla once given these are all Tesla batteries and the picture is of a lady standing next to one. Great journalism, not biased or trying to steer the narrative at all.
REST APIs built with HATEOAS are far superior to MCP. It allows AI to perform incremental, intelligent discovery of your APIs. MCP fills up the context with tons of needless tools upfront. HATEOAS can also communicate state, informing what tools are available, what are disabled and why, in a way that can change dynamically.
You should be using Opus 5.5, it's just as good if not better than Fable, and it sips tokens.