[HN Gopher] Turning the Tables on AI
___________________________________________________________________
Turning the Tables on AI
Author : zdw
Score : 92 points
Date : 2024-06-14 17:42 UTC (5 hours ago)
(HTM) web link (ia.net)
(TXT) w3m dump (ia.net)
| Evenjos wrote:
| ...Or just ignore AI and use your human creativity to write an
| amazing story. Something that is NOT a stale echo of another
| story. Something that adds to a global conversation/debate or
| examines a concept/premise a fresh new light, rather that
| parroting ideas that have already been expressed ad nauseam.
|
| And if you can't think of anything new? Maybe the creative fields
| aren't for you. The world doesn't need more writers. There are
| much easier ways to earn a living.
| d3m0t3p wrote:
| I don't agree, imo being asked questions still draws creativity
| from yourself. It also help to clarify your thoughts. It's like
| pair programming or rubberducking. The goal isn't to wait until
| someone else tells you what to write.
|
| But I agree that a lot of people are copy pasting from chatGPT
| until it works
| CuriouslyC wrote:
| You can use your human creativity to come up with an outline of
| a story, then have AI do first drafts of chapters according to
| your outline. It's tremendously faster to go back and rewrite
| the AI draft than it is to generate a first draft by hand and
| to be honest for most writers the quality will be better.
|
| That doesn't mean you don't painstakingly labor over
| characters, story flow/pacing, events, meaning, etc. It just
| means you're never stuck with writer's block - though you might
| get stuck with storyteller's block, which AI can also provide
| suggestions to help push past.
| tivert wrote:
| > You can use your human creativity to come up with an
| outline of a story, then have AI do first drafts of chapters
| according to your outline. It's tremendously faster to go
| back and rewrite the AI draft than it is to generate a first
| draft by hand and to be honest for most writers the quality
| will be better.
|
| It's also tremendously crappy and less creative. If all you
| want to do is an outline, just publish that instead of
| fattening it up with AI slop.
|
| Also, your idea is great if you always want to be a terrible
| writer and develop your writing skills poorly, and just care
| about volume of output. IIRC, real writers say the writing
| part isn't actually the hard part, but for some reason that's
| what you want to automate. The hard (and creative) part is
| the thinking you have to do when you're figuring out what to
| write.
|
| There are books I've read where I seriously wish the author
| just published an outline, with maybe a bit more exposition
| around world-building (yeah, they were sci-fi). And that was
| human written stuff, reading AI slop would an even bigger
| waste of time.
| CuriouslyC wrote:
| I've known literally dozens of published writers and I can
| tell you that the reason that pro writers say that putting
| the words down on paper is not the hard part is because
| they've trained themselves to mostly vomit on the page,
| then edit it to where it needs to be. Writers that try to
| write good prose out of the gate get blocked, almost
| always. It's a basic piece of tradecraft that people who
| write for a living all know.
|
| The process is outline, trash draft (who gives a shit who
| does this) then refine until you can't bear to read your
| work anymore and neither you or anyone you share the work
| with has critical points that don't have very solid counter
| arguments. Most wannabe writers have notebooks of
| worldbuilding and story arcs but can't even get the trash
| draft done.
| pizzathyme wrote:
| The first suggestion, ask the AI to ask you continual questions
| to help you clarify your thinking, is a great technique. I have
| seen this type of conversation partner very beneficial in my own
| writing and planning.
| sharkjacobs wrote:
| In spring of 2023 OpenAI was introducing custom system messages
| and one of the examples was a Socratic teacher who teaches by
| asking you questions. I'd messed around with ChatGPT a bit but
| that was the big moment that i realized these things were
| really interesting.
| appstorelottery wrote:
| Socratic questioning is quite amazing, I created my own
| prompting for this and found it to be invaluable in exploring
| my beliefs.
| jordemort wrote:
| I thought this article was going to be a pun on the domain it's
| hosted on.
| peddling-brink wrote:
| Isn't it?
| memothon wrote:
| I really like the point about getting AI to ask you questions.
|
| The focus in the AI tutor world is basically a chatbot to ask
| questions of. But if you're trying to learn something, it's
| really helpful to have targeted questions asked of you!
| walterbell wrote:
| Bonus answer data for AI training.
| denton-scratch wrote:
| I think this is very interesting.
|
| As an expert, you can ask an intelligent layman to interrogate
| you about some idea you have, or some essay you've drafted. To
| answer their questions, you have to be clear, explicit, and avoid
| jargon. That in turn forces you to think about what you mean.
|
| You can then ask your layman to play back to you what they think
| you said; they might play it back using a structure that is
| clearer than the one you started with. [Edit] So someone who
| doesn't know your subject-matter can still help improve both your
| phrasing and structure.
|
| FWIW, I suspect you'd get better results from an intelligent
| (human) layman, than from a chatbot.
|
| wrt. the examples: I didn't understand what I was seeing. Are
| those snippets of real ChatGPT exchanges? I think they were [sort
| of] promo material for iA Writer; which is not so much vapourware
| as an imaginative fiction, used to make a point. Is that right?
| 1oooqooq wrote:
| > wrt. the examples: I didn't understand what I was seeing.
|
| guess they should have hired an expert to write the article in
| a way everyone could understand? :)
| mattgreenrocks wrote:
| Integrating some sort of AI into the creative process seems
| like a way more useful approach than how we think of it today.
| You can bounce ideas off them, have them steelman opposing
| ideas, rip your ideas apart, or treat them as a muse.
|
| Best part: it works today. No hopes and prayers for future AI
| needed!
| faitswulff wrote:
| I was hoping this was about how LLMs can't read tables
| accurately, but alas
| krupan wrote:
| Money quote at the end
|
| "Enthusiasts present AI as a magic wand that can solve humanity's
| biggest problems. In the meantime, it uses an exponential amount
| of energy to make everything the same."
| ugjka wrote:
| It is a magic wand for CEO's who are looking for more reasons
| to lay off more people
| mattgreenrocks wrote:
| Same CEOs who hire only the best now seem to magically forget
| those standards when AI's involved.
| vouaobrasil wrote:
| The best way to turn the tables on AI is to take a hammer and
| smash the server that runs it.
___________________________________________________________________
(page generated 2024-06-14 23:02 UTC)