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

AI is impressive because we've failed at personal computing

rakhim.exotext.com

Post reply on HN
Print view View original

171–180 of 190 posts

Re: AI is impressive because we've failed at personal computing

#171

Post by cestith »

jandrese wrote: post #121
whartung wrote: post #75

Earlier quoted context omitted.

Well, this runs straight into one of the massive, concrete pillars of computing: naming things. Because that’s what a lot of this falls into. Overwhelming amount of stuff with no names. No categories, no nothing. With extended file attributes we could hang all sorts of meta bits off of arbitrary files. But that’s very fragile. So we ask the systems to make up names for data based on their content, which turns out to…

No names is not the biggest problem. You just have to come up with a name. The problem is when things have multiple names, or even worse when people disagree on what names are appropriate for something. The world rarely allows you to neatly categorize large datasets. There are always outliers. For example, you have a set of balls and you want to sort them by color. Where does orange stop and red begin? What about str…

Not everything has to be named once and put into a hierarchy like a directory tree. Tags work well for data. A system like an LLM that understands synonyms and antonyms should be able to find and even update tags for concepts that don’t have a full set already - as long as there are a few appropriate tags on the concept to start.

Re: AI is impressive because we've failed at personal computing

#172

Post by jandrese »

cestith wrote: post #171
jandrese wrote: post #121

Earlier quoted context omitted.

No names is not the biggest problem. You just have to come up with a name. The problem is when things have multiple names, or even worse when people disagree on what names are appropriate for something. The world rarely allows you to neatly categorize large datasets. There are always outliers. For example, you have a set of balls and you want to sort them by color. Where does orange stop and red begin? What about str…

Not everything has to be named once and put into a hierarchy like a directory tree. Tags work well for data. A system like an LLM that understands synonyms and antonyms should be able to find and even update tags for concepts that don’t have a full set already - as long as there are a few appropriate tags on the concept to start.

In practice if you're making up tags on the fly it's not much better than untagged data. A LLM that can figure out what the tags mean can probably just infer it from the data anyway.

Re: AI is impressive because we've failed at personal computing

#174

Post by nuancebydefault »

t-3 wrote: post #168
nuancebydefault wrote: post #152

Earlier quoted context omitted.

People always insist that any tool must do things right. They as well insist that people do things right. Tools are not perfect, people are not perfect. Thinking that LLMs must do things right, that people find simple, is a common mistake, and it is common because we easily treat the machine as a person, while it only is acting like one.

People and tools that don't do things right aren't useful. They get replaced. Making do with a shitty tool might make sense economically but not in any other way.

If you follow that reasoning, no person is useful and no tool is useful.

The little box I'm filling now is, compared to a lot of other interfaces, a shitty interface. That doesn't mean it isn't useful. Probably it is getting replaced, only with a slightly better inferface.

The karma system is quite simplistic and far from perfect. I'm sure there are ways to go around it. The moderators make mistakes.

That doesn't mean the karma and moderation are not useful. I hope you get my point but it's fine if we disagree as well.

Re: AI is impressive because we've failed at personal computing

#175

Post by echelon »

armitron wrote: post #167
kbelder wrote: post #166

Earlier quoted context omitted.

He didn't say LLMs can outdo Pixar. That's ridiculous and they are nowhere near that level. He said that LLMs are at the point "where individuals can outdo Pixar." And that's very possible. The output of a talented individual with the assistance of AI is vastly better than the output of AI alone.

This is a very reductionist take that's to be expected from a software engineer but most definitely something that an artistic person would never utter. The creative process doesn't scale in the way that software engineers imagine. Coming up with genuine new ideas or magical moments of "synthesis" doesn't emerge from throwing lots of commodified tools together and calling it a day. So far we haven't seen a single iot…

I'm not talking about LLMs. I'm talking about image and video diffusion models.

Editors, VFX artists, and studios big and small are already using the tools.

I'm in this industry. They're widely deployed as we speak.

Re: AI is impressive because we've failed at personal computing

#176

Post by saati »

copularent wrote: post #49
xandrius wrote: post #28

Earlier quoted context omitted.

Is counting the number of B's vital? Also, I'm pretty sure you can get an LLM to parse text the way you want it, it just doesn't see your text as you do, so that simple operation is not straightforward. Similarly, are you worthless because you seem like you understand language but are incapable of counting the number of octects in "blueberry"?

As shown by the GPT-5 reaction, a majority of people just have nothing better to ask the models than how many times does the letter "s" appear in "stupid".

But the point is, why would you trust it for anything at all, when it can't do an incredibly simple thing reliably at all? (Yes, I understand the tokenizer makes this hard, but still, it's a quick demonstration that it's just bad technology.)

Re: AI is impressive because we've failed at personal computing

#179

Post by rini17 »

nuancebydefault wrote: post #152
rini17 wrote: post #107

Earlier quoted context omitted.

It is (maybe not directly but very insistently) advertised as taking many jobs soon. And counting stuff you have in front of yourself is basic skill required everywhere. Counting letters in a word is just a representative task for counting boxes with goods, or money, or kids in a group, or rows on a list on some document, it comes up in all kinds of situations. Of course people insist that AI must do this right. The…

People always insist that any tool must do things right. They as well insist that people do things right. Tools are not perfect, people are not perfect. Thinking that LLMs must do things right, that people find simple, is a common mistake, and it is common because we easily treat the machine as a person, while it only is acting like one.

> Thinking that LLMs must do things right, that people find simple, is a common mistake

Show me any publicly known visible figure that tries to rectify this. Everyone peddles hype, there's no more Babbage as in the "Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?" anecdote.

Re: AI is impressive because we've failed at personal computing

#180

Post by mikewarot »

The root cause of this is that HTML is not a language for the markup of hypertext.

Because everything has to be copied and/or compressed into a single layer document along with its markup, you almost never see actual markup in the web, just a lot of formatting and layout.

Because you can't have multiple markup of a given source material, adding multiple hierarchies doesn't happen. Any given information structure is necessarily oversimplified and thus only fit for limited use.

It's almost as if someone read Vannevar Bush's description of the memex and decided to actively prevent its reification.

He said the organization of knowledge was the largest challenge facing mankind post war. Clearly he was right, and we've failed miserably.

The thing LLMs provide is an impedance match between the data we do have, and the actual needs of mankind.

Post reply on HN