Hacker News new | ask | show | jobs
by SilverBirch 994 days ago
I think this is eminently sensible as a solution to this problem. AI is a tool that skilled professionals can use for productivity. This idea that AI was just going to replace writers never really made much sense - the generation of the actual text is only one part of the job and not the important part of the creative process. This way the writers have the freedom to use this new tooling, the Studios don't get to dictate which tools writers use (just like they're not going to insist on you using a specific brand of computer), and if in aggregate this increases productivity... well that's just great, that's the free market. And importantly, the writer is the one using and experiencing the productivity benefit. In aggregate that might drive the total demand for writers down, but again, that's just the free market.
3 comments

I have used AI to produce an audio version and translations of my short story. I am not impressed. The best tools generate audio that is of high quality but missed a lot of the nuance of the written text, sometimes distort simple words. When it comes to translation, 80-90% of the sentences had to be fixed for grammar, spelling or other reasons (modifications to the meaning of the sentence, bad handling of gender in gendered languages. I came to the conclusion that writers, editors, proofreaders, and translators have nothing to fear from AI. I also noticed that AI is programmed to avoid getting its creators into legal/PR trouble. For example, if you ask AI to perform certain editing jobs on a piece of content that may be somehow connected to a political or a religious issue it will refuse to do so. Not much help if you want to produce a piece of writing that may be a commentary on such issues, be it an article or a screenplay.
> For example, if you ask AI to perform certain editing jobs on a piece of content that may be somehow connected to a political or a religious issue it will refuse to do so.

Some AI services (especially those run as services by megacorps with lively legal departments) are gimped this way (usually as a sort of "thought police" model running on top of the core model, as I understand), but once you get to self-hostable models not all have such limitations.

How good are the self hostable models though compared to those run by megacorps?
Well, "good" has a few dimensions to it:

1. Speed of output (not very fun to wait multiple seconds for each letter to be output)

2. Coherence of output (how far back does the model remember the context of the conversation?)

3. Variety of output (how's the diversity of the model's vocabulary? How about topics it can plausibly discuss?)

You can easily get comparable speed, so nothing of interest to really compare there.

I haven't done particularly strenuous coherence comparisons, but for my uses, at least, megacorp and self-hosted models are pretty comparable. Though you do need the better models to get the best coherence simply because they retain more tokens in memory.

Variety is, in my opinion, where the megacorp models still rule. Most of my dabbling has been with models designed to be writing assistants and they can certainly generate plausible strings of words and follow a general theme, but they barely "know" anything (generally when using them to write fiction, you would provide them a "factbook" that they can work from). ChatGPT by comparison can generate plausible responses to a surprising breadth of technical questions, although it definitely has a feeling of being generated from scraping certain online sources since it's decent at answering devops questions but bad at obscure grammar and physics questions, at least in my experience.

Doesn't say much if we don't know what AI you're talking about? The best LLM is far ahead of everything else. If it's GPT-4, translation, editing, proofreading is definitely something to fear.
It's one of the big three. I am deliberately not naming it, because I do not want to steer the conversation into "You clearly should have used LLM X". That's not the point here.
>That's not the point here.

Of course it is. If you say don't worry about x technology and you're not using state of the art then it's meaningless. The "Big 3" is meaningless. The best model is far ahead of the rest.

Well, don't be coy then, what's "the best model" and how "far ahead of the rest" is it?
I already said it's GPT-4 and that it's much better than the rest.
Fundamental problem is oversupply of content. Information Overload leading to Attention Theft games.

Its like the Chinese real estate market where more houses have been built than people available. Same story with content and eyeballs available.

Its an extremely unstable demand-supply equation. And it will break down until new ideas about the supply side emerge.

If using AI suddenly means that writers can instantly produce billions of acceptable scripts, the problem will just shift to selecting the good ones from the garbage. I'm not that confident in AI's ability to produce quality writing.
I think this is true, and nothing to do with AI. Netflix bought Big Tech money to hollywood and made it rain. Then all the traditionally media companies followed, creating a bidding war for content. It's probably not popular to say, but the last 5 years have probably been a golden age for hollywood, and some people like Shonda Rhymes have made out like bandits, some people like Megan Markle have literally robbed Netflix blind. But it's all going the other way now, and that's going to be massive downward pressure on the industry, at best AI is a good way of protecting Disney's stock price.
> This idea that AI was just going to replace writers never really made much sense

I disagree. Despite the first taste of the technology cannot completely replace humans, it will get there in an imho relatively short time from now.

When is AI going to replace the rentier class? That should be straightforward, no?
The discussed replacement concerns the production side. A replacement scenario of the pure consumption humans could be when engaging with a human customer will be a loss or pointless nuisance compared to an AI customer.