I Don't Like "AI" | Hacker News Reader

I Don't Like "AI"

ian.mccowan.space

43 pointsmeristohm20 comments

Check on all points. Well said. The misanthropic quality was well worth saying early. I hadn't cogitated on that one but I am now.

It stands out to me that people like Hinton aren't boosting and aren't seeking stonks. I suspect he's aghast.

>> 4. I never had any doubt that this would work. GPT is trained on the internet, which is replete with nerds talking about their nerd shit. Nerds love nothing more than to explain things. As evidence, see: this whole-ass web site

::applause::

Yes, autocorrect is also AI. LLMs are more impressive though.

Sam Altman saying we are all stochastic parrots says more about him and his view on the world more than anything.

This post was top 3 in HN. Now shadowbanned. I‘m out.

Saying that humans are not stochastic parrots because it makes you feel bad isn’t really proof that we aren’t parrots. I’m not saying we are parrots, just that many of the arguments in the post are unfounded vitriol.

Unfounded? Especially the ones about Altmann are verifiable quotes...

If two entities--one an "AI", one a real, thinking person--happen to say the same thing, one is a stochastic parrot, and one isn't. The "real, thinking" part is what makes all the difference. We use language to communicate thoughts. These hucksters are taking advantage of the conflation of languages and the thoughts they are used to convey to pretend they have invented thinking machines.

It's not necessary to provide proof that humans are not machines which do nothing but guess the next likely word. But also feeling anything at all is proof of that.

That's not what the author is saying.

Calling humans "stochastic parrots" implies that there's no actual thought or meaning behind what they say, that their output is a probabilistic process with no "driver". The author points out that his own output in response to this idea is driven by his feeling of disgust towards it. A disproof by counterexample.

This is like an extension for https://xkcd.com/1289/ - it works for everything: internet, phones, electricity, tv, etc:

1) Because the thing you are talking about is just rebranded thing that existed 20 years before

2) Because boosters are egocentric drug addicts, pederasts, druglords with crypto stashes, raped their sister, or everything altogether

3) Because it failed to solve MY task

4) Because it can't be more important than a sunshine

5) Because it attacks my intellectual property

1) What. How do you extend that?

2) Kind of hard to argue that for electricity and phones. Internet, maybe because it was a bunch of academics. TV was boosted by big business.

3) It fails to solve any task, in particular LLMs. You're tricked into doing the work instead. Other kinds of machine learning are useful and do solve tasks. TV also didn't exist to solve a task, at least it was honest about it. Phones solve the task they were meant to, as did the internet until it got enshittened. (Still does, but worse.)

4) whut.

5) It actually steals information. All the other tech didn't.

A poem I composed from the section headings:

It's Not "AI" Its Boosters Are Misanthropic The Hype Sucks And It’s Encroachingly Ubiquitous Its Obsequity Is Annoying And Its Prose Is Vapid Its Boosters Don’t Give A Shit About Consent So What Am I Even Fucking Doing Here

The post can be summarised "I am an old grumpy man. I don't like change. I don't like hype. I don't like people with different values than I have. Let's opposite all change."

Sure, the stuff is bad. You can use the same arguments for any tech: Internet, mobile phones, social media. But for most of the people, the stuff is good. ChatGPT has hundreds of millions of active users. Those users are happy, because it creates enormous value for them.

> Those users are happy

AI affects non-users in a substantially different way to phones or social media. When I phone someone, it doesn't matter if it's a landline or mobile phone (though, in the UK we can tell by the leading digit). But when I read someone's pull request or press release, it does matter whether it's vibe coded, or made by the person. Looking at the person using the tool does not show the whole picture. You're welcome to have different values, but don't hide from me what I apply my values to.

Eternal September is already way past. We cannot have Internet only for people we like and we want to interact with.

it's tagged as "rant" and that what it is. And I think he's more concerned about the hype or unethical behaviors, but not actual AI. But I am just guessing...

What do these „different values“ look like? Is stealing and lying ok for you AI evangelist folks?

For example, but not limited to

- Believe in capitalism and that venture capital allows us to build things that are better for humankind

- Technological progress is good

- Automation and the increased productivity it brings is good

- Individualism: people should be allowed to use tools as they wish, and they not should not be limited to the use cases someone else finds acceptable. Are people allowed to do stuff we don't like?

- Freedom of speech: If it is AI generated slop, you are still allowed to say it even if it makes you to look like an idiot

Let's take vibe coding as an example. The user tobylane would prefer not to allow vibe coding to exist, because it causes him personal nuisance to see vibe-coded pull requests. But then, most people who do vibe coding do it because they like it and it brings them joy and productivity.

I dont see the author expressing any views opposing these except for the „freedom of speech“ which is actually about ownership. And he is right to do so because it brings us back to the stealing and lying. If you offered me a book i would like to know if it had been made off the backs of others without compensetion or consent OR if it is a work of your own. And i feel i have a right to know.

Reply on news.ycombinator.com