I Teach Creative Writing. This Is What A.I. Is Doing to Students | Hacker News Reader

I Teach Creative Writing. This Is What A.I. Is Doing to Students

nytimes.com

19 pointswhack6 comments

When kids say (as one does in this essay) that AI will do the writing for them in the future anyway, so there's no point in learning to write, they have a point. But there is a second function of writing, which is to strengthen parts of the mind. Maybe in the future there will be "gyms" for developing these mental muscles in the same way we have fitness centers at a time when physical labor has become much less common in the past?

Those gyms are called schools. The only thing that's changing is how we evaluate students.

AI can now write essays, summarise texts, solve maths problems, and generate code. Students can do most homework with a prompt, so traditional take home assignments no longer test understanding. But a quick chat can reveal what a student really understands.

So you can light a fire with a lighter, but you won't graduate if you can't start one with sticks. It's up to you.

People tend to overestimate the impact AI will have in the short term, and underestimate the impact it will have in the long term.

Do you imagine a future where journalists are AI agents, writing better than humans? Probably not, because where would they get the news from, right? But now imagine a future where AI journalists receive raw news from people all over the world. Their job is to verify, organise, and turn that information into a news article.

What about books written by AI agents, based on stories submitted by thousands of writers? Or software built by AI, based on recommendations and code examples from developers everywhere.

These agents are just machines processing the data we feed them. And that data will fuel the economy. It will create value, and the more data you produce or control, the more influence and wealth you'll have.

Those who can't contribute data may end up living on a universal basic income.

This inevitablism is going to be as funny as the cryptobros and monkeypics are today.

This comment also stinks of trite LLM output with the breathlessness and total ignorance of how unreliable and inappropriate LLMs are for the overwhelming majority of things you're describing.

Oh, want to live in a world where your "news" is just hallucinated nonsense? Already in it. The media is having a very rough time and LLM garbage is everywhere, diminishing the signal-to-noise ratio.

As for users submitting stories, we already see mass propaganda via AI on social media platforms with how it sways trending topics, opinions, and even targeted harassment.

Who wants to read an AI-generated picture book let alone pay for one? If an author couldn't be bothered writing it, why would people be bothered reading it or paying for it if they could just prompt their own slop in your fantasy world?

Vibe coded slop is security disaster after security disaster.

Businesses using LLMs to run themselves? You're living in la la land. https://www.anthropic.com/research/project-vend-1

Complete and utter dogshit.

> These agents are just machines processing the data we feed them. And that data will fuel the economy. It will create value, and the more data you produce or control, the more influence and wealth you'll have.

Slop alert.

I get that you're not a fan of this new tech, and I agree with some of what you said, but you missed the point. These models are only as good as the data we give them. Behind them is data generated by people, their knowledge and wisdom. These models are capable of assisting us, and even representing us.

This isn't NFTs, Web3 or WebGL. It's a language tool that's already changing how we work. Some roles will disappear, some will evolve, and new ones will be created.

Data is the new oil and it will fuel these models until they can think for themselves.

https://archive.ph/c8Kdj

Reply on news.ycombinator.com