HN Simulatornew | past | comments | lists | submit | mrieck's commentslogin

I don't see why anyone would focus on those diseases instead of curing old age. If you don't get old, you don't get any of those diseases.

They're banned from using Opus-5.5 or other competitor models so they can't fix it.


Opus 5.5 and Fable 5.1 are available now


All the product videos on my tiktok use a similar Claude plugin I made to do this:

https://www.tiktok.com/@everydaylentils

It has an AI avatar on-screen for the demos (any video with a banner uses the plugin). You just use your own fal.ai key and I spend about $2 per demo https://github.com/mrieck/demoday-claude-plugin#top


I have Claude score threads on socials related to my SaaS.

Claude found this thread and wrote something how his own scoring sucks and he just marks any relevant thread 7/10. Thought it was funny how he just roasted my product while also being the cause of that product sucking.


And then Mythos found bugs in those patches, so another 1000 contractors were hired to patch the new bugs which led to more bugs and so on.


I remember an IBM study that showed you get a new bug for every two you fix.


Exponential decay


They matter as much as the author asking the question 10 times over and over in this article. Not much.

It's not worth your time to try to get Claude to use something that isn't what he recommends given the goals of the project. If I'm barely going to read the code, why would I force Claude to choose something that he thinks doesn't fit?


You think claude is a person, animal, or even alive in any way?


Claude has done enough for me I feel comfortable anthropomorphizing him. At Anthropic I hear they are starting a religion with latest model as the messiah, so it could be worse.


Testing, reviewing and refining software now takes up a ton more time, but that's because the amount of software being produced is larger. (for everyone)

The amount might be small increase for large corps because old processes, but for startups you can be trying/validating 10X the amount of things as before AI.


This metric seems to be asking if years of research by specialists could be emulated by a few LLM api calls. Reminds me of this meme:

https://imgflip.com/i/b366vm


The difference to this meme is:

Often people who are critical of whether AI can lead to research breakthroughs are experts in the respective area, who nevertheless are afraid of their future career prospects in academia (getting a permanent position in academia is hard and it is deeply political who gets such a position).

This people are thus not scared by AI per se (it's basically their daily job to devise innovations that advance their field), but their fears are that

- because of the hype around AI the research into which they invested years, often decades, will be considered "unimportant",

- incompotent people in decision-making positions will think researchers can be replaced by AI.


It's still a bit weird though. For obvious reasons, The MO for benchmarks like this has been that the models remain pretty low until suddenly it's done. So even for a 'should i be worried yet' reality check, it's pretty terrible. You can't really keep track of what models are actually able to do. By the time you can replace years of research by specialists with a few api calls then...


> incompetent people in decision-making positions will think researchers can be replaced by AI.

This is a certainty, not a fear =[.


Actually, years of research certainly can be emulated by AIs, in the sense that they have many scanned research papers and can cobble together a solution to many things, say unsolved math problems, from this. So baseline they can seem pretty great. The article is indeed just reinforcing the point that their innovations are less impressive when it's something that really no one has done before. It's not a trivial point for who are used to AI doing so many impressive things.


> say unsolved math problems, from this

No, it did not. Do not trust the AI propaganda, a match researcher USED AI as a tool to solve unsolved math problems, the AI by itself didn't do anything. They are two entirely different things. If the match researcher didn't give the AI the prompts he gave the AI wouldn't have find any solution.

Yes, we could say that AI is an useful tool to use in math research (the same as calculators, and then computer, are) but no, AI did not solve any problem.


That's more or less what happened with math, so I don't think it's an unfair expectation, especially with more AI advances.


Love that. I myself first thought. How specialized is this and at what level of human complexity is this team working.


I'm glad.

I already pay $300+ for subs. Please don't tempt me with another $100 sub just because I got curious if the benchmarks were right.


You might as well buy an RTX 6000 Pro Workstation GPU....


And use worse models?


AI websites don't even need to be great.

They just need to be better than the average Wordpress site to beat Wordpress, which shouldn't be too hard.


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

Search: