Daniel BuzzoMar 15
Neil Brown

*Edit*: here at least, I am clearly not isolated!

Perhaps I am increasingly isolated in holding this position, but I have no interest in reading "AI"-generated slop.

I love reading.

I read people's blogs and toots and whatever *because people wrote them* and I want to read their own thoughts and opinions.

I buy books, and read numerous different authors. I like finding new authors, bringing new ideas, styles etc.

Same with "AI" images. I'd prefer no image at all.

Mar 15 at 6:52amWeb
Show threadAmin, minor deity of the legume realmMar 15

@neil

Heh, what a coincidence to see this right as one of my old posts saying the same (from eight months ago) has been getting boosted around again. ;)

https://polymaths.social/@amin/statuses/01K03C7KYJ50AE33CQAYRVTYHM

Amin, minor deity of the legume realm (@[email protected])

I will repeat this as many times as I need to: no matter how terrible you think your writing is, I would far rather read it than anything that came out of an LLM.

polymaths.social
Show threadKevin Karhan Mar 15

@neil I like that stance re: #AIslop:

https://social.treehouse.systems/@whitequark/116224927984042017

✧✦Catherine✦✧ (@[email protected])

the Glasgow Interface Explorer project's new LLM policy: > We do not accept contributions where code was generated partly or in whole using LLMs. These tools are designed to be extractive and place an unreasonable burden on OSS maintainers, both on the level of individual pull requests and for the ecosystem as a whole. If you didn’t bother writing it, we won’t bother reading. https://glasgow-embedded.org/latest/contribute.html

Treehouse Mastodon
Show threadRadomír ŽemličkaMar 15
@neil Don't worry, you are not alone.
Show threadLucMar 15
@neil +1. The only place where AI generated text is a useful read is generated software documentation. It works pretty good for that.
Show threadKim Crawley 😷 (she/her)Mar 15

@stroobl

No! 🖕🏻🖕🏻🖕🏻

Show threadLucMar 15
@kimcrawley thanks for sharing you kind opinion.
Show threadiweinMar 15

@stroobl
AI can write good docs is an unpopular opinion among people that know the value of documentation. Rightfully so because:

1. "AI" doesn't exist, and LLM's can't "know" meaning. They hide it well.
2. Writing documentation proves the author cares about the thing they document.
3. Documenting something surfaces it's subtlest bugs.

If devs use llm for documentation, the software is shit. If you use llm to describe others' software, you're taking a great risk of misunderstanding it.
@neil

Show threadLucMar 15
@iwein @neil you should be aware of the risks is valid advice for any tool you use.
Show threadiweinMar 15

@stroobl I'm aware, but why is that relevant?

My conclusion can be rephrased in that frame as: The risk return ratio of using an LLM to generate documentation is always less than 1.

The rephrasing of the premises is left as an exercise to the reader.

@neil

Show threadmirabilosMar 15
@stroobl @neil fuck no
Show threadDavidMar 15
@neil Definitely not alone on this.
Show threadDavidMar 15
@neil (and I'll take any author recommendations you might care to share)
Show threadTess, Joint Chief of StiffMar 15
@neil I'm still trying to find the right words for why Ai stuff feels so unappealing. I think it's that they have no provenance? Theres nothing to contextualise it.
Show threadqueer as in fuck your bordersMar 15

@diffractie I don't like the feeling that I might be the first person to look at/read/encounter something.

It just feels so inherently like it doesn't matter, it's like thinking you're on solid ground but when you check there's nothing holding you up, you've already gone over the edge of the cliff.

@neil

Show threadmirabilosMar 15
@diffractie @neil it also makes thr brain mushy
Show threadSignalElevenMar 15
@neil please. Isolated? Nobody wants to read slop. Not even people who like AI for some specific application.
What prompted this lukewarm take? :)
Show threadBlake C. StaceyMar 15
@neil I also continue to hold this position.
Show threadEm Mar 15
@neil Absolute same
Show threadBrynnbo the FrogMxstressMar 15

@neil What I found really depressingly interesting is that I find it really fatiguing, and I already suffer from a huge amount of fatigue. All the dodging AI, my brain-fogged head working harder to filter things out, the heart sink as I get tricked and only later realise something was off.

Like recipes! Following along on my phone whilst already overwhelmed by what I'm doing, and then finding out mid-way through following a recipe that, hang on a sec, this is AI nonsense.

Show threadBlake C. StaceyMar 15

@FrazzledBrynn @neil Ah, damn, that's gotta suck.

I like to cook, but I don't even try looking up recipes online any more, unless they're verifiably from the pre-slop era. Fortunately, we have a shelf of cookbooks and I've gotten pretty good at winging things for a weeknight dinner.

Show threadMacCruiskeenMar 15
@bstacey @FrazzledBrynn @neil I have a bookcase dedicated to cookbooks, and I've worked as a recipe editor in a cooking school. There's no way I'm giving those up for slop. Even before AI slop most online recipes were SEO garbage anyway.
Show threadJaimie's EroticaMar 15
@FrazzledBrynn @neil
Yes! The constant gatekeeping is exhausting. In my profession, it's increasingly an issue where people (lay persons and professionals alike) have used AI to draft a document which looks legit but on closer inspection turns out to be a pile of shit.
Show threadwobweger Mar 15
@neil 🧡
you are not alone
you're not isolated
your are in fact well connected to the human condition, having resisted years long propaganda on almost all channels
#salami slop is no good
🖖
Show threadahnlakMar 15

@neil I find AI images a really strong flag that I'm not going to want to read an article.

I suspect at least part of the problem is that "experts" will tell you that a web page *has* to have an image to represent it (because that's what algorithms want), and people are panicked into finding *something* to fill that void.

Show threadRuby JonesMar 15
@ahnlak @neil The thing is... Pixabay exists! I used it all the time when I worked in Marketing. I think you can spend a lot more time messing around trying to get your 'prompt' to give you what you want than you would just going to Pixabay and finding a public domain image.
Show threadMarianneMar 15
@rubyjones @ahnlak @neil and Unsplash. And Wikipedia commons. And more besides I'm sure. There's just no need to use slop images, really.
Show threadRuby JonesMar 15

@noodlemaz @ahnlak @neil Oh, yes, others exist, although I still find Pixabay the best.

Perhaps a good time to plug my article, 'Create your own book cover - without generative 'AI' which is full of free or cheap resources 😉

Show threadHaelwenn /элвэн/ Mar 15
@ahnlak @neil Which is kind of funny because even before AI I found it so tailored towards some sort of ranking that it can end up looking a bit shady like sites that look borderline SEO-farming.
Show threadRoger BW 😷Mar 15
@ahnlak @neil people who believe that seem well correlated with people I'm not interested in reading.
Show threadDuke of Germany 💫Mar 15

Same for me, also with YouTube videos: The moment something AI-generated comes up in a video, I stop watching.

@ahnlak @neil

Show threadMaverynthia🌱Mar 15

@ahnlak @neil

But there's already so much royalty free, clip art and other free images that already exist.

Show threadAlison ChaikenMar 15
@ahnlak @neil Anyone else feeling nostalgia for Bettmann Archive? Their images were dull but at least on-topic and suitable.
Show threadRuby JonesMar 15
@neil Most writers and artists I know think the same, but the gap between those people and those who consume art and fiction - especially among programmers - is really depressing.
Show threadRobert KingettMar 15
And honestly? The best thing one can do is to actually read fiction. *Good fiction.* *Bad fiction.* *Just for shits and giggles fiction.* that, ironically, keeps you from getting half way through synthetic text and realizing, whoops, it's synthetic! Of course there is LLM fiction but the best part is, you can just stop reading at any time and *not* post about the shitty thing you just read. These people want attention, including negative attention. Picking up another bit of fiction is the best way to say fuck you to these generators. Even go back into public domain works such as https://standardebooks.org/ebooks/claude-mckay/home-to-harlem @rubyjones @neil
Show threadWurzelmannMar 15
@neil I'm with you on that hill. ✊
Show threadRokosunMar 15
@neil That is not an unpopular opinion, not at all.
Show threadAndré LampeMar 15
@neil yes, you are not alone. 100%.
Show threadMorten Juhl-JohansenMar 15
@neil In principle I mostly agree, in practice entirely.
I would say with the top 92% I would read, I share the sentiment. *In principle*, I would be okay with looking up LLM-gens for the last 8% which would be checklists or how-to on changing a setting on something or listing a number of events. But it is mostly, or at least unpredictably, bullshit. If these models are built holding everything that is smart and good, why is there so much bullshit?
It does not reassure me for humanity that the machine that extracts the average of everything generally comes up with visually well-ordered feces.
So I arrive at your conclusion mostly, but not entirely, following the same route.
Show threadGabriel Viso CarreraMar 15
@neil you're not, at least from my end. I am in the exact same page.
Show threadcoldclimateMar 15
@neil strong agree
Show threadArjan OostingMar 15
@neil
I agree!
Show threadHeddersMar 15

@neil Quite.

I'm not opposed to machine learning; it's genuinely useful. I'm more ambivalent about LLMs, but that's more to do with the environmental impact, the way they were trained and the people involved in doing it than it is to do with the technology per se.

What profoundly offends me is the pretence that these things are capable of thought or creativity, of "generating" something that is actually going to be of interest to or provoke an emotional response in a human.

Show threadGary Mar 15

@hedders @neil I’ve been pondering this recently, too. I’m as much against the generative/LLM material being used in lieu of that created by humans as anyone else, but the knee jerk reaction of many people towards “AI” as whole now is grating on me.

AI is bloody brilliant, it’s doing incredible things with computers, from running my home central heating more efficiently, to modelling crowd behaviour at football matches, and predicting heart attacks.

Show threadGary Mar 15

@hedders @neil But Big Tech is selling us a dream of general AI that isn’t a reality just yet, and the consumers don’t know enough about the technology to understand that.

The negative sentiment towards AI in general is, imho, undeserved. And it’s all because the shite they’re selling at the moment is so dodgy. It’s a real shame.

Show threadMatZeMar 15
@neil You're not isolated, just normal.
Show threadpixl_xip 🪿Mar 15
@neil Yes! I read to hear other people‘s ideas, not for cobbled together fragments of ideas from an AI’s training data!
Show threadJustin CrozerMar 15
@neil I’ll never post AI generated content on my blog because why would you expect/want someone to take the time to read something you didn’t even create?
Show threadRevK Mar 15

@neil I should apologise for my current avatar, to be honest. Someone made it for me and it seemed amusing, and, to be frank, close enough to reality. But I agree, trying to read AI writing is hard work.

I have read a few auto translated books, not sure if “AI”, but the writing style stays as the original author I think, just gets the gender pronouns messed up occasionally.

Reading the occasional AI C code is similarly tortuous even if it is “right”.

[avatar now changed]

Show threadRevK Mar 15

@neil reviewing an AI commit is really hard work.

When a person writes the code the mistakes, which always happen, tend to be way more obvious.

When AI writes code the mistakes can be much harder to spot. AI code tends to “look correct”, which is the point. You have to get past that to see the errors.

Show threadRoger BW 😷Mar 15
@revk @neil specifically I can often work out what a human coder was thinking.
Show threadRevK Mar 15

@RogerBW @neil Indeed.

The clue being that "thinking" was actually involved...

Show threadRoger BW 😷Mar 15
@revk @neil Yes, exactly so. The process of debugging someone else's code involves getting to a state of "aha, they were trying to do _this_". This is literally impossible with LLM code because it was simply copying all the vaguely relevant code it could find.
Show threadChaotic UnicornMar 15
@neil same goes for music, code, 3d printable designs and much more.
Show threadSarah WMar 15

@neil
I quite agree. It's just not necessary.
AI has its uses in certain fields but generally I despise it
People need to remember that if AI doesn't know something, it makes it up, so we can't ever rely on it.

It's all part of a bigger picture of needing to be constantly entertained.
Yesterday I heard about kids' AI soft toys that talk to them. Some of the results are extremely disturbing .
AI is not a baby-sitter.

I honestly don't understand why people use AI instead of actual people.

Show threadmirabilosMar 15
@neil @Sarahw its use is only as a flag that the thing is not worth attention
Show threadDerick RethansMar 15

@neil I made that exact point in my closing keynote at #dpc last Friday. There was a lot of LLM generated images in many presentations, even when actual photos existed. My line was something akin to: I'd rather see badly drawn stick figures illustrating my points than fake generic "art", and I also made the point about the different ideas and writing styles of authors.

I'll turn it into a blog post soon.

Show threadSephsterMar 15
@neil best book you read last year? Looking for recommendations
Show threadJennifer Moore 😷Mar 15

@sephster

Book recommendations from anyone, or only from @neil?

Some of mine are here: @booktrail

#books

Show threadSephsterMar 15
@unchartedworlds @neil @booktrail welcome from anyone. Thank you for the heads up
Show threadElse, SomeoneMar 15

@sephster

"Against the Grain" by James C Scott was a paradigm shifter to me last year!

DEBT follows right after: https://mastodon.acm.org/@nobody/116232495598467927

@unchartedworlds @neil @booktrail

Else, Someone (@[email protected])

Attached: 1 image "As if time itself has become his enemy" @[email protected] thank you sooo much for bringing DEBT to my attention, it's full of gems

Mastodon