Hacker Newsnew | past | comments | ask | show | jobs | submit | ryeights's commentslogin

Remember, that's the most expensive this capability will ever be.

If it's model is opened up and can run on commodity hardware. Otherwise price could go up as RAM and silicon prices climb.

Yes, but the problem with these models isn't a gradual shift, it's a step function. With a gradual shift, the world has time to react and adapt.

Sorry, this guy is a hack and this is cope. Most of the things he's saying re: Mythos are objectively false.

- Open source models found the same bugs? Sure, if you tell them "here is a file which may contain a vulnerability, look for a bug in how function XYZ handles ABC"

- It's all mostly false positives? According to Anth, each suspected vulnerability came with a bug report and working PoC...

- "Humans had to fix the things"? As in, he thinks models are incapable of writing the patch?


> Open source models found the same bugs? Sure, if you tell them "here is a for which may contain a vulnerability, look for a big in how function XYZ handles ABC"

In one of Anthropic's blog post, they describe that that's basically what they did too. They run the agent many times, each time specifying a different file to focus on. [1]

From my experience as a security researcher, manually finding a fishy file and sicking even sonnet 4.5 yields great results for most memory corruption bugs.

No comments otherwise. I don't have a clue as to who that guy is, and I haven't watched the video yet. You might be right overall.

[1] https://red.anthropic.com/2026/mythos-preview/

Edit: looked at the open source model claims - I agree that they suck. Basically all the details are given away in the prompt - not just the file.


Yes, but Anthropic didn’t already know the answers. In the OSS ‘reproductions’, they fed the model the one file that actually has a vuln and even told it which parts of the code to focus on. This is obviously a much easier task.

If OSS models are equally up to the task, why not find novel vulnerabilities?


Yeah, totally agree now that I've looked into it more.

> If OSS models are equally up to the task, why not find novel vulnerabilities?

To be fair, in the same blog post Anthropic mentioned costs in the tens of thousands of dollars per project looked at it. So it's a big ask to do an experiment that compares. Would love to see it though.


The original formulation of Chinese Room deals with ‘understanding’. One supposes that understanding implies some form of subjective experience. I take OP’s use of ‘think’ to refer to an equivalent concept

Is it surprising? It seems likely you could build a complete working model of the universe with no provision for consciousness at all. As far as modern science goes, it's an intractable problem

It doesn't seem likely to me that in, just a couple hundred years, humans have developed such a thorough understand of every natural process as all that.

You can do it with no provision for molecules too.

That wouldn’t be complete though. I mean that an outside observer could view a simulation of the universe by your model, down to the subatomic particle level, and find no differences with the real world. There is no measurable “consciousness field” as far as we know, so how can our science even begin to approach the topic?

What do you mean by "wouldn’t be complete"? If no differences with the real world are found, then the model is complete.

If the "new software engineering" is Jira, planning, and convincing PM's about viability all day, you can count me out!

Until recently I would have described myself as an AI skeptic. HN has been a great source for cope on the AI subject over the years. You can find nitpicks, caveats, all sorts of reasons to believe things aren’t as significant as they seem. For me Opus 4.5 was the inflection point where I started to think “maybe this isn’t a bubble.” The figures in this report, if accurate, are terrifying.

>I had no idea this story existed and woke up to claims that I was obviously* suppressing it.

To be fair, it seems you’re saying the submission was being suppressed, just not intentionally. Lots of props of course for transparency and reboosting the story


When people use the word "suppressed" they usually mean that we were personally intervening to do something suppressive. This being the internet, they say that with supreme confidence whether it's true or not.

For example, the comment I was referring to, which was the first one I saw, said "It is being suppressed by @dang" (https://news.ycombinator.com/item?id=47457010). You can't get more personal, definitive, or wrong than that.


Okay but my comment here said that it was being suppressed (intentionally?).

In my other comment, I actually did not mean to write “it is being suppressed by dang” but rather “it is being suppressed @dang”… Because my impression is that that alerts you somehow? I may be wrong about this.

Please give your long-time readers the benefit of the doubt. I was correct that it was being suppressed. I'm also very thankful for your moderation of the site. I know you do a lot of hard work on that front.


Those things all exist within our conscious realm. “Human brain cells in a vat used for computation” suggests horrors beyond understanding


>If we are serious about climate policy, we have to set up trade barriers proportional to greenhouse gases emissions to limit this effect.

Consumption economies can incentivize production economies to emit less.


We simply drop a giant tub of baking soda into the ocean every now and then.


Thus solving the problem once and for all

https://www.youtube.com/watch?v=VW66EX75jIY


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: