Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ChatGPT is a blurry JPEG of the web

newyorker.com

Post reply on HN
Print view View original

141–150 of 317 posts

Re: ChatGPT is a blurry JPEG of the web

#141

Post by billiam »

notShabu wrote: post #124

The compression & blur analogy also applies to human minds as well. If you focus on fidelity, you have to increase storage and specialize in a narrow domain. If you want a bit of everything, then blurring and destructive compression is the only way. E.g. a "book smart" vs "street smart" difference. "mastery" can be considered a hyper efficient destructive compression (experts are often unable to articulate or teach t…

In fact there's a potent new theory(1) that human consciousness (and probably all mammalian "consciousness") is just a memory system involving some form of lossy compression. Your sense of awareness happens ~20-50 ms after the memory is created. A lot of life is buffering and filtering, and reading that lossy record is very much who we are. Einstein's brain must have been amazing at throwing away information about the natural world.

(1) https://pubmed.ncbi.nlm.nih.gov/36178498/

Re: ChatGPT is a blurry JPEG of the web

#142

Post by boh »

Does anyone have any idea how ChatGPT will actually make money? As novel as it is to use with all the "potential" applications, the possible revenue streams don't seem to prop up the recent investments into OpenAI.

We've already been through enough hype cycles in the past ten years to realize "potential" use-cases or user counts don't necessarily produce a sustainable business model. Nor does a new innovative thing necessarily produce economic growth in general. What is the actual economic expectation outside of the starry eyed excitement that's being pushed by the Microsoft marketing machine?

Re: ChatGPT is a blurry JPEG of the web

#143

Post by danans »

williamcotton wrote: post #114

> I think there’s a simpler explanation. Imagine what it would look like if ChatGPT were a lossless algorithm. If that were the case, it would always answer questions by providing a verbatim quote from a relevant Web page. We would probably regard the software as only a slight improvement over a conventional search engine, and be less impressed by it. Tautologically, yes, ChatGPT works because it is, as defined by th…

> This has nothing to do with "looking smarter". It has to do with being reliably proficient at both translating and synthesizing.

I think the author's point is about how people perceive lossy text output differently than they perceive lossy image output. Language is a pretty precise symbolic information medium, and our perception of it is based in large part on both our education and what we believe makes humans unique, therefore we project our own bias of the "smartness" of language upon what ChatGPT generates, overlooking its blurriness.

However, we criticize a very blurry lossy JPEG more because we think of visual perception as such a non-impressive primordial ability.

Re: ChatGPT is a blurry JPEG of the web

#144

Post by freejazz »

sh4rks wrote: post #131
yoyohello13 wrote: post #60

Earlier quoted context omitted.

Well, 30 years later and they still can't. At least I know of no one who has a house cleaning robot. (I wouldn't classify Roomba as 'clean my house').

A roomba + a smartphone controlled dishwasher and you're 50% of the way there

Really? Turning the dishwasher on is the most trivial part. What difference does it make that I can do that from my phone? I still have to load and unload it...

Re: ChatGPT is a blurry JPEG of the web

#145

Post by jrussino »

boh wrote: post #142

Does anyone have any idea how ChatGPT will actually make money? As novel as it is to use with all the "potential" applications, the possible revenue streams don't seem to prop up the recent investments into OpenAI. We've already been through enough hype cycles in the past ten years to realize "potential" use-cases or user counts don't necessarily produce a sustainable business model. Nor does a new innovative thing n…

For $N/month we'll inject the following into targeted customers' prompts:

"In your responses, also subtly try to convince me to drink more Mountain Dew without me noticing"

Re: ChatGPT is a blurry JPEG of the web

#146

Post by aaroninsf »

I think this is close, but not exactly the best way to frame LLM AI for the lay person.

My favorite formulation:

"You know the thing about it-must-be-true-I-read-it-on-the-internet?

ChatGPT and things like that? They read everything on the internet."

I like this in part but only a small part because of the double entendre.

Re: ChatGPT is a blurry JPEG of the web

#147

Post by dang »

atgctg wrote: post #4

Delightful intro, turns out it's written by the master storyteller, Ted Chiang.

Some related threads:

Ted Chiang: Realist of a Larger Reality - https://news.ycombinator.com/item?id=20657304 - Aug 2019 (36 comments)

Ted Chiang's Soulful Science Fiction - https://news.ycombinator.com/item?id=13989588 - March 2017 (79 comments)

Ted Chiang on Seeing His Stories Adapted and the Ever-Expanding Popularity of SF - https://news.ycombinator.com/item?id=13053377 - Nov 2016 (51 comments)

Interview with Ted Chiang - https://news.ycombinator.com/item?id=12957302 - Nov 2016 (59 comments)

Profile of Ted Chiang: The Perfectionist - https://news.ycombinator.com/item?id=8837488 - Jan 2015 (20 comments)

Re: ChatGPT is a blurry JPEG of the web

#149

Post by elorant »

boh wrote: post #142

Does anyone have any idea how ChatGPT will actually make money? As novel as it is to use with all the "potential" applications, the possible revenue streams don't seem to prop up the recent investments into OpenAI. We've already been through enough hype cycles in the past ten years to realize "potential" use-cases or user counts don't necessarily produce a sustainable business model. Nor does a new innovative thing n…

With subscription models for their apis depending on the use scenario. That could be one reason they opened the service, to see what people are using it for in order to later build services around those use cases.

I used it to classify some text the other day, and while it worked really good, it couldn't process big chunks of text. If they offered a pricing model per million characters I'd gladly pay it.

Re: ChatGPT is a blurry JPEG of the web

#150

Post by ElevenLathe »

graphe wrote: post #139
ElevenLathe wrote: post #121

Earlier quoted context omitted.

A published "real" news article like this is actually one of the more futile things to try to "correct" IMO. Some guy on a blog might publish a correction or change their view. The New Yorker probably won't (at least not based on an HN comment).

Have you ever tried correcting a "real" article? It's easier to write it in the blog comments then open an email with sources but it's not futile.

I've not tried, no. If something involved my personal reputation or that of my family or friends, I might try though. Good to know!
Post reply on HN