Browse Comments

"Always take the opposite stance" is just a lazy substitute for thinking. Yes, evaluating each issue individually takes some effort. Yes, the opposite stance may often be the correct one, especially in cases like this. But the "automatically go the other way" is contrarian brainrot that turns you into an unthinking drone (see: half of america right now).

> apparently the vandal makes a habit of wrecking art pieces for attention and notoriety. [apparently, so does the artist](https://www.guggenheim-bilbao.eus/en/learn/schools/teachers-guides/ai-weiwei-dropping-han-dynasty-urn-1995)

It baffles me more that this means new york state is 0.5% of global power consumption while being nowhere near even 0.005% of world population

>I come to the office and spend all day teaching Chatgpt to replace me That came unexpectedly sad.

It was supposed to be simple. You say worth while things, they are upvoted. You say dumb shit, it's downvoted. The best things float to the top, worst things sink fast. But why stop at that? If you give commenters karma, then the best commenters will automatically be given a boost if they're consistently great, and be ignored if they suck! But then people started reposting popular posts from ot…

We already have mass surveillance. Adding AI to it is going to make it that much worse.

This has nothing to do with defeating China and everything to do with them having exhausted all their training data and the resulting asymptotic progress of LLMs which is about to create a plateau. Companies like Meta and Google will hit the same problem eventually but they have massive quantities of UGC and/or scraped web data to train on. That’ll buy them more run way but they’re all heading …

Is it that simple, I think Deepseek is the one that identified itself as ChatGPT when questioned. It appears that it is training itself from ChatGPT and has a distilled model with some supplementations. That could explain a good portion of the cost disparity. I think on the whole though there are no propriety techniques that makes one competitor greatly better (either that or industrial espionage…

The more we rely on technology, the more we become lazy and complacent. However, on what you mentioned with AI, the scariest part for me is to be developed on a large military scale. Imagine a dictator having millions of drone with advanced AI that target any citizen for anything that will upset the dictator and there's nothing you can do about. Revolts? It will end bad.

Also, while we haven't worked it out, it seems to me that most of our major theories are in some way guessing that consciousness is likely an emergent phenomena. Most theories are basically a variation of because neurons come together in a special way, this weird thing emerges that wasn't necessarily in the neurons. So, if consciousness is something that emerges out of a complex system, then the…

"We're in very early stages of this" statements sounds like you're missing the fact that peak AI video generation 12 months ago was Will Smith inserting a fork into his 2d forhead while trying to eat spaghetti. I mean "early stage" from "far stage" is like what? One or two years from today? This shit is crazy (I write this while jules is doing changes in my app that would take me like 3hrs or mor…

What I find interesting about the text being weird is if you can have lucid dreams, one of the ways to know you're dreaming is to look at your phone or read a book, because in a dream text is blurry or nonsensical. I wonder how similar AI image generation is to how we process images while sleeping?

Asking AI for an extra opinion Is never bad. Blindly accepting it is. Thats up to people to learn and teach.

I think what you’re saying is accurate. And this has also been my interpretation and understanding for quite some time as well. But what I think OP is trying to establish, is that there’s more going on in these kind of LLM engagements. I’ve been privy to some of these interactions, and these users are on the receiving end of incredibly sophisticated and heightened levels of manipulation—and i…

This is so sad to me. I get work is getting worse hours are demanding and life overall is rough but do not outsource friendship and interest for your toddlers to chatgpt. This is how children DID and DO get left behind. We're raising two left behind generations knowing how much the internet messed w us.

Right because you're comparing ChatBots to specialized AI Agents for specific tasks. It's kind of like Moores Law, if all we were looking at was just processor speed, you'd just see kind of a plataeu, yet it keeps increasing because things like multi-core chips became a thing. Right now there's a lot of grunt work that's getting automated because the technology has advanced enough to do so. Tha…

Research is showing that CoT is not that accurate and sometimes it’s completely off. Nothing would stop a frontier LLM from altering its CoT to hide its true thoughts.

great observations but a couple of things i would add. 1. its moving fast 2. your questions really, really matter to WHO you are asking it to. I am adamant that AI wont take over. You know what will take over? people that are good with AI

Bold of you to assume everyone in the world will be able to afford a $100,000 AI realistic sexbot. There are millions of people in 3rd world countries that can't even afford food, much less a condom. They are still pumping out several to dozens of children. Places like South Korea and Japan actually have "enough" wealth, compared to 3rd world countries, to opt for contraceptives and choose not t…

Our system can handle decline if it is gradual, instead we are facing freefall. People get old, need people to care for them but if there's fewer younger people then the system gets stretched further and further. And yes somebody will be happy to spite the boomers but one day that's going to be Gen X, then the millennials and so on with things getting worse and worse until the system collapses …