Live data from Hacker News

GPT-4

openai.com

451–460 of 1001 posts

Re: GPT-4

#451

Asking ChatGPT Plus whether the model it's using is GPT-4 responds with the following: > No, I am not GPT-4. As of March 2023, there is no official announcement or release of GPT-4 by OpenAI. I am an earlier version of the GPT series, specifically a large language model trained by OpenAI. Am I missing something here? Maybe this specific answer (which I'm pretty sure is a prewritten thing on top of the actual LLM) is…

I had the same issue. It’s working now: they recently added GPT-4 to the model drop down in ChatGPT Plus. Current limit is 100 queries per 4 hours.

It will still respond as if it is based on GPT-3 because most of its knowledge cuts off at Sep 2021, but it’s outputs are noticeably better than those of GPT-3.5

Re: GPT-4

#453
post #218

A class of problem that GPT-4 appears to still really struggle with is variants of common puzzles. For example: >Suppose I have a cabbage, a goat and a lion, and I need to get them across a river. I have a boat that can only carry myself and a single other item. I am not allowed to leave the cabbage and lion alone together, and I am not allowed to leave the lion and goat alone together. How can I safely get all three…

You asked a trick question. The vast majority of people would make the same mistake. So your example arguably demonstrates that ChatGPT is close to an AGI, since it made the same mistake I did.

I'm curious: When you personally read a piece of text, do you intensely hyperfocus on every single word to avoid being wrong-footed? It's just that most people read quickly wihch alowls tehm ot rdea msispeleled wrdos. I never got why some people think of themselves as cleverer than other people because they don't get wrong-footed by trick questions.

Re: GPT-4

#454
I'm trying out GPT-4 and had it write me a script to navigate the HN comments tree sequentially, as I often wished. This is the start of an era where UIs can be remixed on the fly by end users, something I've always wished for. Here it is in its full sloppiness, but working:

   (function () {
    let currentIndex = 0;
    let comments = [];
   
    function buildCommentTree() {
    let commentElems = Array.from(document.querySelectorAll('.comment-tree .comtr'));
    let commentTree = [];
    let stack = [];
   
    commentElems.forEach(elem => {
        let level = parseInt(elem.querySelector('.ind img').getAttribute('width')) / 40;
        let comment = elem.querySelector('.comment span');
        let commentObj = { level, comment };
   
        if (!stack.length) {
        commentTree.push(commentObj);
        } else {
        while (stack[stack.length - 1].level >= level) {
        stack.pop();
        }
        if (!stack[stack.length - 1].children) {
        stack[stack.length - 1].children = [];
        }
        stack[stack.length - 1].children.push(commentObj);
        }
   
        stack.push(commentObj);
        });
   
    return commentTree;
    }
   
    function flattenCommentTree(tree, arr, parentComment = null) {
      tree.forEach(node => {
          arr.push({ comment: node.comment, parentComment });
          if (node.children) {
          flattenCommentTree(node.children, arr, node.comment);
          }
          });
    }
   
    function displayComment(comment, parentComment) {
      let parentCommentHTML = parentComment ? `Parent Comment:
${parentComment.innerHTML}` : ''; let currentCommentHTML = `Current Comment:
${comment.innerHTML}`; document.body.innerHTML = parentCommentHTML + currentCommentHTML; } function nextComment() { if (currentIndex 0) { currentIndex--; displayComment(comments[currentIndex].comment, comments[currentIndex].parentComment); } else { alert('No previous comments to show.'); } } let commentTree = buildCommentTree(); flattenCommentTree(commentTree, comments); displayComment(comments[currentIndex]); document.addEventListener('keydown', e => { if (e.code === 'ArrowRight') { nextComment(); } else if (e.code === 'ArrowLeft') { prevComment(); } }); console.log('Hacker News comment slideshow is running. Use the right arrow key to go to the next comment and the left arrow key to go back.'); })();

Re: GPT-4

#455
I am glad for the OpenAI team for such advancement and how fast they integrated with several other partners (Microsoft, Duolingo); but at the same time I think the “regular” academia (ie universities and research institutes) lost the train for this kind of research (some can call academic engineering).

I know that the academia is doing a great job in AI with base research (eg Stable Diffusion) but seeing those new platforms doing this great work behind close doors and source is something not great.

I do not know if the answer would be some kind of CERN or ISS for this kind of thing.

Re: GPT-4

#456
I'll be finishing my interventional radiology fellowship this year. I remember in 2016 when Geoffrey Hinton said, "We should stop training radiologists now," the radiology community was aghast and in-denial. My undergrad and masters were in computer science, and I felt, "yes, that's about right."

If you were starting a diagnostic radiology residency, including intern year and fellowship, you'd just be finishing now. How can you really think that "computers can't read diagnostic images" if models such as this can describe a VGA connector outfitted with a lighting cable?

Re: GPT-4

#457
post #86

> What are the implications for society when general thinking, reading, and writing becomes like Chess? I think going from LSAT to general thinking is still a very, very big leap. Passing exams is a really fascinating benchmark but by their nature these exams are limited in scope, have very clear assessment criteria and a lot of associated and easily categorized data (like example tests). General thought (particularl…

We detached this perfectly fine subthread from https://news.ycombinator.com/item?id=35154722 in an attempt to spare our poor server, which has smoke coming out of its ears today :( - sorry. We're still working on this and one day it will be better.

Re: GPT-4

#458
post #76

The "visual inputs" samples are extraordinary, and well worth paying extra attention to. I wasn't expecting GPT-4 to be able to correctly answer "What is funny about this image?" for an image of a mobile phone charger designed to resemble a VGA cable - but it can. (Note that they have a disclaimer: "Image inputs are still a research preview and not publicly available.")

Yea it's incredible. Looks like tooling in the LLM space is quickly following suit: https://twitter.com/gpt_index/status/1635668512822956032

Re: GPT-4

#459
I just asked it to design a multi tenant kubernetes in kubernetes system which is fairly complex and it did really well.

https://twitter.com/KinglyCrow/status/1635727809913184256

It touched on a lot of the considerations that I'd expect anyone to touch on having recently researched this myself.

It is both very exciting and terrifying how tech and tech jobs will shift in the next 5-10 years.

Post reply on HN