Hacker News Reader

carrolldunhamon Vice, Decayed Digital Colossus, Files for Bankruptcy

what on earth do people think neoliberalism means suddenly? it means free trade but out of nowhere I see zoomers treating it like it's some kind of bogeyman?

carrolldunhamon When you lose the ability to write, you also lose some of your ability to think

As a possible counterpoint to LLMs atrophying writing, in order to have fun on Character.AI, and tease out nuances with various GPTs, I've had to dust off my ability to write/think. I notice a lot more verbal reasoning going on in my mind after these weeks of back-and-forthing.

carrolldunhamon Are We Close to AGI?

It didn't fail logic, it failed to read your string. You didn't even notice that the terms being added in the 'show working' version don't match the given strings. And you are elsewhere talking about adding spaces to make the digits 'tokens'... you have no idea what you are talking about at all.

carrolldunhamon Are We Close to AGI?

the linked blog is yet another instance of making false inferences from the false assumption of the model working at a letter/symbol level. it cannot see letters. it cannot see the 1 or 0 in 11010 either. (note that a request to simply reverse a word would illustrate this usually, but some glue code has been added to do it)

carrolldunhamon ChatGPT has no inner monologue or meta-cognition

Anything to do with counting words or letters, manipulating letters, etc is just leading to false inferences about the power of the model, because it doesn't work with words and letters but tokens. It can't see words.

carrolldunhamon Derek Parfit: The Perfectionist at All Souls

I don't know what you're arguing here but that kind of reinforces anti-nasalism's point. It implies that 'natalism' persists only by meta factors and is therefore false/illegitimate in itself.

carrolldunhamon Show HN: Chrome extension to hide AI/GPT related submissions from HN

I'd like to hide the submissions about hiding GPT submissions

carrolldunhamon Show HN: AI Playground by Vercel Labs

It's not a mistake. It's just ambiguous - you and the OP (rather than "humans") share some contextual assumptions that to you make it imply one meaning. If someone asked me with no context to 'be william gibson pitching your next book' I might answer just the same as gpt did

carrolldunhamon Why don’t more people live close to friends?

I always get pushback for this but I think the idea that adults 'hang out with friends' is a meme from sitcoms. I think people grew up absorbing sitcom premises as representing life. I look at my parents and grandparents they don't have many friends - there's one or two people they see once or twice a year. That's normal. Having 3-5 people who you catch up with regularly through the week and hang out like teenagers, is a premise for a comedy with dynamics. It's not real life after 30.

carrolldunhamon Show HN: GPT-4-powered web searches for developers

Change the name of expert mode. You're having to tell each single user to turn it on, because probably like me they think it means "answer as if talking to an expert"

carrolldunhamon Defamed by ChatGPT

anyone using it is shown a page saying this bot makes things up

carrolldunhamon Windows 12 Plans Reveal a Bold Redesign, Tighter Security, AI-Powered Features

It's fine, because there's two groups of people - people who know when you install it to turn off Cortana and online stuff, and people who don't know that, who, it turns out also tend to think "ooh, a garlic festival?" "ooh, prince harry did what?!" instead of "what is this trash in my start"

carrolldunhamon Show HN: Document Q&A with GPT: web, .pdf, .docx, etc.

I mean, if you wanted to add no value to GPT, yes. The value of any 'read a pdf' app should be that it traverses the full document as needed for a query, copying into its internal context what may be relevant, re-iterating until the true answer is found regardless of whether it's scattered through the large document. Bing and the OP app both have as author says a "layer" that should do that but currently

- Bing isn't using the layer right - it can search the whole internet to orchestrate context for itself but doesn't seem set-up to so search one doc

- OP app seems to crash after a while

- the third service i forget the name of which has been shown on here (something to do with 'pdf') similarly to Bing doesn't 'search around' in long documents appropriately

carrolldunhamon Show HN: Document Q&A with GPT: web, .pdf, .docx, etc.

>would it also answer specific questions about the doc?

Yes.

carrolldunhamon Show HN: Document Q&A with GPT: web, .pdf, .docx, etc.

The 'creative tone' mode of Bing was upgraded to 'large context size', presumably 32K. By unstated (I've changed my post now) I meant that it doesn't report back "that's too long", it just says "sure here's a summary", which is not representative of the document. I saw another "pdf reading" service (not OP one) that suffered this too.

carrolldunhamon Show HN: Document Q&A with GPT: web, .pdf, .docx, etc.

Bing can do this (with GPT-4) but the problem is it has an unacknowledged limit to the amount it's able to read, and so it mishandles summarizing a large document, seeming to read the start and a few random pages. How does yours handle large sources?

Edit: I tried this one with an 800kb .txt and after digesting it, it got two answers wrong (but at least related to the text) and then started spitting out "I don't know". I asked "what is this document?" because I saw with my previous test that it can get blocked and be working with a 404 page ("this document is a page that says suspicious activity request denied") but this time it just said "I don't know."

carrolldunhamon Understanding ChatGPT

I think you need to consider conditional statistics. "What are high probability options for the next word, given that the text i'm working on starts with the words please rhyme, and that the text 10 words ago was 'sun' and the text 20 words ago was 'fun'?" How it knows which parts of the text are relevant to condition on is the attention mechanism which is like "what is the probability this word is important to how to finish this sentence?". Both of these can be extracted from large enough example data

carrolldunhamon Google just launched Bard, its answer to ChatGPT

ChatGPT cannot search the web. Bing and Bard search the web. If ChatGPT doesn't know about your topic it just lies. Bing will ground its answers in web results. Bard is supposed to do that but isn't smart enough for it to work apparently

carrolldunhamon Bard uses a Hacker News comment as source to say that Bard has shut down

>People expect searches for factual information to be inaccurate probably 1 in 10,000 times.

It's okay they've recently cooled this off with google's 'answers' or whatever they are called that are placed on top of the results which are almost never right

carrolldunhamon Show HN: Get a Professional Headshot in Minutes with AI

The examples look like cartoons, have artefacts and are quite obviously the output of AI to anyone who has ever seen it, which at this point is a lot of people. Even someone who never saw stable diffusion generations is going to be wondering about why you have weird blobs in your eyes.

carrolldunhamon On the left is GPT-3.5. On the right is GPT-4

I, from the start, visualised the coffee cup having a lid. The author seems to have a blind spot in his assumption it doesn't. To me "coffee cup" slightly implies a lid because arguably a ceramic mug is simply "a cup" or "a mug"

carrolldunhamon A skeptical take on ChatGPT: Ezra Klein interviews Gary Marcus

>"It doesn't 'really know'..."

You write this out as if this is an idea you formed yourself and not a tired/expired talking point - it's such a funny comment because it's pure regurgitation, but the argument in it is against regurgitation being "real thought". So are you a bot? Or are you intending to actually proving that there's no difference between ML and 'real' intelligence by showing you do what ML does? Mind spinning stuff

carrolldunhamon Remaking old computer graphics with AI image generation

The title excited me - maybe someone succeeded making new art that looks like the old pre-renders, maybe a convincing imitation of the scanline render look. Instead it was a vapid article about tossing pixel art into img-2-img and getting some tenuously related junk.

carrolldunhamon The cost of being poor: Why it costs so much to be poor in America

pouring water and olive oil into some self-raising wholemeal flour and throwing it on a hot pan to make naans (90 seconds to cook) twice a day, comes in under $1/day and 30min/day with washup. for one person

carrolldunhamon Archaeologists ask Netflix to reclassify Graham Hancock’s docuseries as fiction

>It felt like he was just purely on the defensive against 'big archeology' from the very get go,

This is specifically the play though, it's what triggers some kind of mind-blow response in people's minds - don't give any explanation but insinuate that "everything you know is false", because the entire structure that exists is some kind of front, and I'm the only one who will tell you. This is what Qanon and sabine hossenfelder play on

carrolldunhamon Americans are choosing to be alone, but we should reverse that

>it seems safe to assume that the decline of our social lives is a worrisome development.

Gee ok, could we get any basis for that at all?

>Because people said they like things with their friends

When did they say that? Presumably before we all changed our preference, resulting in changed behaviour.

carrolldunhamon The Nothingness of Personality (1922)

Isn't it just buddhist Anatta? What is the point of trying to re-invent the wheel from scratch?

carrolldunhamon YouTube confirms that it has removed the “sort by oldest/newest” option

Out of date time wasting clickbait. The actual news is this:

TeamYouTube: >we added new tabs for different content types & when adding these tabs, we ran into issues keeping all the sorting options you're used to (like oldest to newest). as an FYI, if you want to find videos around a certain date, you can search "before:" ex. "dog videos before:2017"

>So to confirm there is no longer a sorting option, like oldest to newest.

>we know sorting options are important & we’re still exploring how to bring these back

https://twitter.com/TeamYouTube/status/1590386502865129472

carrolldunhamon Monumental (if correct) advance in number theory posted to ArXiv by Yitang Zhang

Then you spend your whole PhD reinventing a wheel that has a different name in the 30 year old textbook from the next field over, and neither your peers or professor have any awareness of that

carrolldunhamon Ask HN: What's Going on with the HN Algorithm?

>the standard model for the underlying process is a normal distribution.

I don't think it is, though?