Not by your definition maybe, but that misses the point made in the article:

> why did I conclude that 90% of my skills had become (economically) worthless?

The op has inserted a specific word here in braces because this isn't a question of what defines a writer, it's a question of who gets paid to write. By that definition, what defines a writer is not for you to decide, it's decided by the market. And the market, in my experience, is not particularly concerned with quality.

People who have money will always pay a premium for quality.

With regards to writing, if you can create quality information with AI faster than before, you will be more profitable.

If you are going wide, and throwing spaghetti at the wall with crappy writing, you may also be able to generate $1 from a million people vs. $10,000 from 100 people.

AI threatens the middle for sure, and will force all of us to change our strategies.

Google results and app stores are already filled with low-effort trash. If this gets worse, then the importance of "influencers"[1] as curators of the trash pile rise even more.

[1] Or YouTubers or whatever else you want to call them.

> People who have money will always pay a premium for quality.

The assumption here is that the Venn of "people who have money" and "people who can recognise / care about quality" is closer to a circle than a figure eight - that has not been my experience.

Well, yeah - I guess we're even calling into question the very definition of the word quality here. Your comment seems to imply that quality is a metric that correlates with quantitative metrics of mass appeal, rather than personal & individually diverse concepts of subjective quality, which would be more what I was referring to. But both can be equally valid given the context.

Nope. Writers write. Most writing is "economically worthless" — the number of novels that get written every year is astronomical versus the number that make money. If we were to let the market judge the quality of creative works, we'd be culturally fucked.

The writing skills that became worthless for this guy were the ones that weren't worth much to begin with.

Exactly. The market price for anything is determined by both demand and supply. The demand for good writing is relatively unchanged. The supply is also relatively unchanged. In the case prior to ChatGPT that demand and that supply were somewhat low, or as you say - not worth much to begin with.

Oh sure, there's always the Stephen King's and Tom Clancy's of the world, but guess what? They were always the exception anyway and they will continue to be. The same argument can be made for musicians.

Stephen King has written so much by now, his new books are basically remixed versions of his old books. Stephen King has become his own ChatGPT.

(No accident either - the guy is a writing machine - 30 pages a day, no breaks, that's over 50 200 page books a year...)

We're not disagreeing, you're just talking about something different than the topic of the OP.

An expert reader will be able to tell the difference. But assume that these people have read and understand the subject but have not put the millions of words together yet.

At some point, it will become like a calculator. I can multiply 20% off a dinner bill faster than getting the calculator out, but people won't spend the time learning the mental math and they will do just as good a job.

The barrier to entry to the role just got a lot lower, but the barrier to gaining enough experience to become a specialist just dramatically jumped up

Everyone else? They won't get the ability to establish their brand/style (assuming chatgpt gets faster at updating with recent information/sources).

I can see us relying far more on centralized institutions (again). Only reading from certain publishing houses, writers focusing more on trying to get into a certain house vs content of their work.

Yes. In an age of abundant junk, curation of the gems can be a successful strategy.

We can add this to a growing list, along with Google earth and second life.

I am a decent writer and communicator.

It takes me a little while to get from a bulleted outline to a crisp couple of paragraphs.

I now use ChatGPT to do initial composition from the bulleted list, and then a first edit. Then I do final edits myself. It saves about half the time, and it feels like it saves more than half of the energy.

ChatGPT's revision of my answer:

I'm proficient in writing and communication, but I require some time to transform an outline into a concise and polished piece of writing. To streamline the process, I use ChatGPT for the initial draft and first edit, and do the final editing myself. This approach cuts down the time and effort required significantly.

That's kind of irrelevant. People paying others for writing gigs just care about words on a page, covering a certain topic. Doesn't have to be artistic or to contain novel ideas.

So, regardless if some writers make it, hundreds of thousands of people currently making a living writing copy can (and will be/are) replaced by AI text. If it gets a little better at creatively rewriting news agency feeds, many journalists will too (that's like what a huge number of journalists do, and not original reporting).

At best they will be reduced, for less and more precarious pay, to glorified prompt writers and checkers.

>If you’re really a writer

Yes, and real true Scotsmen will be fine too. The rest, less real Scotsmen, not so much.

If someone is able to delegate a task to a $5 worker or a 50 cent API call, and produce meaningful output, fine. But we have no such examples of tasks being able to be magically delegated in such a way that creates a cohesive, meaningful, broad whole publishing company and product together.

Nobody has used Fiverr to replace the NYT editorial staff with a paper that's just as successful. And I don't think NYT would find value in delegating their article creation to ChatGPT either.

Context, collaboration, high level planning, fact checking, deep understanding of market mechanics and sentiment, raw talent, etc, matter for writing news. I'm not even a fan of NYT or most modern journalism (I hate yellow journalism), but I don't think you can automate away an institution with a couple of GPT bots. I know everyone is breathlessly assuming that's the case today, but nobody is producing anything compelling.

"Genius is one percent inspiration and ninety-nine percent perspiration" as the saying goes. But for some reason GPT heads think that equation has been flipped. I don't buy it.

Tons of writers and graphic designers and such are already replaced by Fiverr contractors and other similar cheapo gig workers.

>Nobody has used Fiverr to replace the NYT editorial staff with a paper that's just as successful.

That would be like the last holdout. A good 90% of websites don't have such quality consraints or reservations.

It's like someone say "AI will replace graphic designers" and someone replying that David Carson wont have a problem. Duh, but most jobs don't go to the handful of David Carson's of the trade (or even anybody 1/100th as famous)...

Really? NYT is employing this? What organization is doing this (outsourcing their core competency) and succeeding bigly?

> That would be like the last holdout. A good 90% of websites don't have such quality consraints or reservations.

A good 99.99% of websites don't even matter, are just private personal sites with no product or business attached to them, so I'm not even sure what the assertion is there. Of course there's a power law for quality, impact, profitability, etc among websites just as there is for anything. If someone uses GPT on their personal blog site, that 3 people a year actually read, I'm not sure that's proving anything.

> It's like someone say "AI will replace graphic designers" and someone replying that David Carson wont have a problem. Duh, but most jobs don't go to the handful of David Carson's of the trade (or even anybody 1/100th as famous)...

I don't think AI, even at present capabilities, is capable of replacing graphic designers at any business where graphic design is the core competency. A skilled graphic designer is just going to have their productivity enhanced by having access to AI, and they're going to still be able to perform at far higher and consistent quality than "a prompt engineer". Consequently, however, demand for custom imagery will increase as people realize they can do more than simply add a stock image to an ad campaign.

But just as "stock photo libraries" didn't kill the photography profession, nor did even putting a HD camera in everyone's pocket with amazing filter and stabilization effects...I predict that AI graphic tools will not replace graphic designers.

What's with NYT again? I already wrote NYT will be the last holdout, i.e. NOT using this, but that means nothing, because 99% of news and content websites are not the NYT or have the same constraints and reputation to uphold.

>A good 99.99% of websites don't even matter, are just private personal sites with no product or business attached to them, so I'm not even sure what the assertion is there.

I'm not talking about personal websites. This is in the context of news and content websites, not personal homepages. We were talking about websites currently employing journalists and writers, and the danger AI would be to their jobs, remember?

>I don't think AI, even at present capabilities, is capable of replacing graphic designers at any business where graphic design is the core competency.

Which is irrelevant. It can still replace them where graphic design is not the core competency, but still provides a livelihood for hundreds of thousands of people - small time web designers, logo makers, etc. working for small businesses, individuals, and so on.

Dismissingt this as "where graphic design is the core competency", is more or less again like saying "David Carson and big design brands will do fine".

>But just as "stock photo libraries" didn't kill the photography profession, nor did even putting a HD camera in everyone's pocket with amazing filter and stabilization effects...

You'd be very surprised. Photography as a profession is near ly dead. The vast majority of professional news photographers, whole teams, have been fired from all major and minor newspapers that used to have them, and the profession in general has depreciated greatly.

If you mean "there are people who still do it as a profession", sure, it wasn't killed. People still offer horse carriage rides too, at least photography fares considerably better than that.

Ok, show me one meaningful publication doing this, that's what I'm getting at. I used NYT as my original example to illustrate you can't just fire a news room and replace it with a GPT bot. What business can you do this with, where writing is the core competency?

How about graphic design? Is there an example of a graphic design firm that has fired its designers and replaced them with Stable Diffusion et al and a single PM to achieve the same output? Or even reduced their head count because these AI tools have so dramatically improved productivity that one designer can now do the work of 2 or 3 or 10?

> It can still replace them where graphic design is not the core competency (small time web designers, logo makers, etc. working for small businesses, individuals, and so on.)

Let's work with a concrete example here. I'm going to work under the assumption that there is a self-employed web designer out there who, despite the existence of hundreds of template web app creators (such as Wordpress, Webflow, Wix, etc), no-code tools like IFTTT and Fiverr contractors on demand, has managed to carve out a full-time job for themselves designing websites.

In order to carve out that niche in the presence of already existing WYSIWYG tools and outsourcing services, they are already doing something different than just boiler plate web design. Before bots come for this web designer's business, these bots would first have to make these tools completely obsolete. And yet, I know of no one today suggesting that an ordinary small business bypass Webflow for ChatGPT + Stable Diffusion to create their web site.

If ChatGPT + Stable Diffusion is really the dangerous job-killing combination that everyone thinks it is, it first has to replace the boiler plate tooling and the Fiverr contractors. Then AI's capabilities need to expand to accommodate long-term client relationships, long term contextual awareness of client needs, realtime integration of marketplace demands, institutional expertise and deep product knowledge and awareness about many implicit things not explicitly documented for any GPT to consume (and not necessarily possible to document, either). It would need to have wisdom to understand when it shouldn't, in the context of often incredibly vague product requirements. It would need to be able to sometimes even forcefully push back against client demands and recommend an alternative path altogether.

These are the qualities of a good web designer (and indeed, any creative worker), and they're just a small sampling of the reason why the already existing boiler plate services haven't replaced the need for good people.

We have data on this, the Bureau of Labor Statistics provides it: https://www.bls.gov/ooh/Media-and-Communication/Photographer...

Number of jobs, 2021: 125,600 Job outlook 2021-2031: 9% growth (faster than average)

> The vast majority of professional news photographers, whole teams, have been fired from all major and minor newspapers that used to have them

That's because no one reads newspapers any more, not because of automation. The internet + social media killed the newspaper, but it didn't decrease overall demand for professional photography.

First, those are much depreciated jobs. It's as if programmers became minimum wage "AI prompt checkers" and instead of 1 million programmers we got 2 million of them (some covering what Excel jockies do now). There are more of them, but hardly what it was before.

Second, the growth must be compared with overall industry growth.

"Both the Bureau of Labor Statistics (2014) and IBIS World (2013) report that the photography industry is in decline. While the Bureau of Labor Statistics (2014) expects to see the number employed in the photography industry increase by 4% between 2012 and 2022, this is well below the average of 11% increase across all industries. IBIS World (2013) reports a decline in the UK between 2009 and 2014 of 4.1% but also comments that this is a long term decline. Clifford (2013) and Photo Counter (2012) both support the view that the photography industry has been declining for some time. Clifford (2013) suggests that photographers have no clearly defined career paths anymore and that more photographers entering the market find themselves doing a wide variety of photography assignments to make a living. The two metrics used to measure the growth and decline of an industry are generally the changes in revenue and the changes in the number of people employed in that industry. Yang (2010) explains that the distinction between amateur photographers and professional photographers is blurring and to an extent becomes irrelevant since many amateur photographers / hobbyist photographers are selling their work on micro stock photography websites. Amateur photographers selling their images on micro stock websites are not likely to be included in the metrics used for measuring the growth / decline of the photography industry".

https://pure.solent.ac.uk/ws/files/24372581/Are_professional...

For mass consumption, sure. But the best writers almost never wrote for mass consumption. Some of them did acquire a mass following, but it usually was not a goal; their goal was to communicate the ideas they developed to a potentially interested audience. Ads and dark patterns are their enemies, they distract readers from ideas the writer wants to convey.

The last few decades saw a meteoric rise of blogs and other forms of writing. A small portion created something new, but the vast majority were regurgitating old ideas in a (slightly) new light or peddling outrage. I will not cry if 90% of those writers stop writing. This will also increase chances that whoever is still writing does so because he has ideas to capture, not because he optimizes for money and is happy to push ads or ideas based on today's ROI. My 2c.

You'd be very, very, surprised.

But I feel like it should be more like 50% in his case so it's a fair interpretation.

https://news.ycombinator.com/item?id=34648167

All of that just got way easier to do. A lot of it can now be done by people who couldn't have, before, because a machine can fix their writing so it's sufficiently non-shit to be acceptable for many purposes. The rest, that still needs some decent writer or editor involved, is likely to see a huge increase in per-worker productivity, which will either result in similar levels of employment with far higher total output, or... in most of the people doing it now, having to switch careers.

My expectation is that we'll just turn editors into writers-who-also-edit—aided throughout by AI tools—and have non-writers take on writing as part of their job (see: what happened to secretaries—we're all secretaries, now) and most of the writers are going to be out of a job. I reckon this'll hit commercial writing (where most of the writing jobs are) and high-volume-low-value fiction writing (e.g. the romance novel scene, middle-grades shared-penname book series, that kind of thing) hardest. Ghostwriters, too—one who embraces AI tools and uses them well will be able to do the work of five or more who don't, so we'll need far fewer people doing that work.

If anything ChatGPT will basically filter out these sophists because it is a Chad-Sophist. The important thing now, is "depth" of knowledge. Of course, good generalists with wide array of mental models have their jiggle, but in the world of LLMs an individual's depth of knowledge trumps.

I don't understand the vitriol HN has for professional writers and anyone who might work in gasp marketing.

Replace "writers" with "programmers" and your statement is still true but you don't hear it being shouted from the rooftops on every post about programming on HN.

Let's say we'd be better off without anyone working in, gasp, marketing. Companies might sell less, but society would be much better of.

If I had a nickel for every website I opened up and then closed because after 5 minutes I had no idea what the company or product did or why I should care, I'd be rich. Might be something really clever. Dunno. I just know it's 5 minutes of my life I'm not getting back because some techie thought their genius would be obvious to all.

Only in a pedantic sense. We could live with non-commercial marketing, if you want to call it that, without an advertising budget, ads, sponsored articles, and other such BS - like, say, the author promoting their FOSS on their blog. Or crafting a nice looking page with an enticing description.

As for commercial companies not being able to sell their products in that case, the world's smallest violin plays a heart-felt tune for them...

Or is this one of those situations where you ignore the marketing you do like when painting with the broad brush of marketing=bad.

They wouldn't unless the product was needed and worth it enough to stand on its own via word of mouth and reputation. That's the main benefit.

A marketeer might want to stir up "word of mouth" or build a product's reputation artificially. But I refer to good ole organic word of mouth and reputation built without marketing.

You know, just by buyers recommending it to other buyers, and the product getting recommended because the see it's good by itself. The way those things are normally understood, just like how "love" normally is understood as two persons taking a liking to one another, and not e.g. as in some guy stopping his car in a shady end of the city and paying $50 dollars in exchange for sex.

You can keep moving the goalposts of what true marketing is though.

Oh, an ad hominem! Nice. I could also work at marketing and hate marketing, doesn't change the point. And of course, I like HN despite it being a "marketing tool", not because of it. The "marketing campaign for YC" part is one of the main things I dislike about it. And the question wasn't whether marketing could ever produce something cool, but If it was that, I could add that I like some funny TV ads too.

>You can keep moving the goalposts of what true marketing is though

Really? As if differentating because some FOSS team making a cool site and promoting their project on some forums, for example, and corporate marketing, with a budget, a marketing department, ads, astroturfing, and all that, is impossible, right? "It's all marketing"

I don’t think it’s useful to look at the world that way. And I say that as someone who absolutely despise marketing and advertising.

I grew up in a culture and with an idea that there is no bad book -- trust me, I wasted years chasing some bad ideas from shallow content. So, the Bullshit vendors earned my vitriol.

People in glass houses...

> trust me, I wasted years chasing some bad ideas from shallow content

Fool me once and all...

My favorite book genre is roughly “Academic writes popular book after 20 years of researching a topic”. That will be very hard to replace with ChatGPT.

Someone like Gladwell though … yeah I can see LLMs being scary for that genre. Not quite yet, but soon.

LLM writes the first draft, human makes it sing.

Maybe in a year ChatGPT will be retrained on a million AI thinkpieces and it will be able to construct an essay with some interesting points. But humans need to come up with and communicate those interesting points first.

For a historical comparison, the same could be said of photography. You have to know instinctively what makes a great photograph if you're going to have a hope of making them on a regular basis, or for pay. The gear really doesn't matter much, it's just table stakes.

Having GPT spit out some new combinations of words can help you get around writers block, or expand your vocabulary, but I strongly agree it won't automatically make you a writer.

The wave of non figurative "modern art", can be seen as painters trying to find new markets.

Both modern art and abstract art trace their origins to the mid-1800s, which is contemporaneous with the development of photography techniques, but before the popularization of photography in the 1880s.

I wonder if the knowledge of photography could have helped spur the development of abstract art, rather than its wide adoption and economic effect.

One other thing to consider is that the starving artist predates photography - the "Bohemian artist" archetype was in full effect in Europe ahead of photography's popularization.

People who want to take pictures no longer need to be photographers.

I think it’s interesting to look at, but I think the parallels break down pretty quickly. The decline of legacy photographic equipment and the reduction of skill needed to take photos is all about a rather technical skillset becoming less and less necessary. The smartphone also made many other legacy things irrelevant for most people.

Language and writing are closer to the core of humanity in a rather existential way. It’s less clear that reducing the skillset required to participate in writing is a net benefit, even if it enables the generation of more text by people who might not have done so before.

Smartphones make a photographer’s job easier and a non-photographer’s goals possible. I worry that the same can’t be said for language models and the end result is likely to be closer to a new form of illiteracy than it is to the democratization of a process that formerly required greater degrees of technical skill.

When painting broad strokes, I see the similarity. Drilling down even a little widens the conceptual gap quite a bit.

It's easy to imagine the same thing happening with text. In fact, the speed that lawmakers and judges adopted the idea of generated text is a very big concern to me.

I sometimes hear people say „there is so much interesting stuff to watch on youtube/pinterest/whatnot“ and i always wonder how the vastly the meaning of „interesting“ can vary.

I know what they did to cameras, but photography?

You still need the person who takes the picture, and the process of taking a picture hasn't really changed that much.

I don't know why LLMs would do that with "writing" though.

The same might potentially happen with writing, where if anyone can generate a decent original story, poem, advertising copy, product UI writing, etc. in less than five minutes, the lowest and middle tiers of paid writing work will be eliminated.

It’s possible that because these writing-specific jobs were already low value / largely eliminated by the Internet at large, that won’t happen. But it could happen to almost any role that basically involves synthesis and summary of words and data, since LLMs are so general purpose.

But digital cameras for the everyday person existed long before the smartphone. And before that, point and shoot film cameras and Polaroid.

The marginal cost of taking a photo is now $0. It's a hobby that anyone with a smartphone can now take up.

It’s was already a near universal hobby.

20 years ago nearly everyone took pictures at some point. Every house had a camera or two, and disposal cameras were cheap and very widely available.

It was widely regarded as a universal hobby back then. My mom was a wedding photographer, and people would regulars make the same arguments back then.

If you look at historic BLS data the number of photographers has actually grown faster than the population has since 2020, and the wages in 2021 were about the same adjusted for inflation.

Clearly smartphones haven’t had a huge impact on job numbers or salaries. My hypothesis is because photographic skill was already as devalued as it was likely to get, and that actual technical skill is a very small part of the overall job.

Smartphone cameras are great and easy to use for a very narrow set of conditions. Outside those conditions they perform terribly. The kicker is that you don’t really know unless you already know… the “trained eye” can immediately notice.

The same is true for GPT text/code.

Even then, yes, it will still result shifts like we saw years ago with newspapers letting go of their photo staff and telling journalists to use their phones.

We already see this with book after book churned out with no real story attached. There are already way more people who write well and have little to say.

But vast number of people don't get paid, for writing that communicates but rather for spewing words that vaguely cover a topic, writing that's a rewrite of already written policies in a different formats (FAQ versus Mission Statement versus etc) or marketing newsletters that want high-value adjectives put together in a plausible way. ChatGPT can apparently do all this. I wouldn't count on less people being employed versus more pap being produced but we'll see.

It's funny that each time people were criticizing each new such technology that it's so easy that people will produce so much garbage content that humanity will drown in it.

In reality, while software developers probably enjoy the act of coding, it’s rarely the interesting or difficult part. It’s just the rendering you do to bring all the work you did to life.

I guess it was more common in the card data days where you could write comments by hand on the card.

We also had language models that could generate code out of very precise specifications back in the 90's. They were called offshored programmers. Send an email in the evening and get a .zip file in the morning with code. Sometimes it would even compile!

Engineering is much more than that. It's about being able to formulate, understand and analyze (conflicting) requirements as well as the ability to gain domain expertize as needed. The best example of that is Carmak and binary space partitioning. [0]

To me, what GPT is going to achieve with it's current performance is akin to what compiled and interpreted languages did: You no longer really care about word size, endianness or alignment issues. The compiler takes care of generating the correct assembly for you. Same thing with snippets generated with GPT.

[0] https://fabiensanglard.net/quakeSource/quakeSourceRendition....

You might end up writing the right solution, but because you didn't communicate "what", "how", and "why" you did clearly enough, you will setup other developers for failure.

The first hard problem is to understand what to build. The second hard problem is to build it in a way that is clear and maintainable.

I have already used LLMs to publish my first book. Specifically, I used sudowrite + chatGPT. LLMs are a true equalizer in writing. Correcting grammar, improving and rephrasing wording, and even helping to brainstorm and organize.

Basically writers now have tool that is as powerful as Google was to researchers.

So I completely agree the author is mistaken here.

To compare, as a C programmer a spent a lot of time creating custom collections: arrays, linked lists, ... Then in the 1990's, C++ with templates came along, and had it all built in. I feared this, initally: Some technical fears,e.g. would the compiler produce something reasonable? Deeper ones were psychological: Would I still have interesting work to do?

Looking back on these worries makes me look silly. Of course there was interesting programming work left after I stopped manual collection creation. In fact, it forced me to level up, made me more productive. Today, I consider a language without built-in collections as inferior.

Technology did this to all of us again and again. Banks had rooms full of human computers, calculating interests on paper. They were obsoleted by mainframes calculating a whole bank's worth of interests in less than a week. Then the spreadsheet came along, where people could simulate a week of finance batch runs in minutes.

I think writing is safe. We will still value people capable of communicate a book worth of ideas. Only now there is a tool capable of taking angry word salad and turning it into readable sentences. We all have to go trough an uncomfortable time leveling up, and in 5 years ChatGPT is the new normal and our worries will look silly.

Historically, a [sysadmin|developer|security professional|etc] who could "put words in a row" in a way that their manager, client, functional analyst, etc could even begin to comprehend, was worth their weight in gold. Majority of developers and sysadmins (and I speak as an insider of my own folk here:) could not communicate effectively or efficiently with people of different contexts & backgrounds; and in particular could not summarize key points that align with somebody else's priorities. Heck, in my current project, I have a team of dozen people who, yes amongst many other things, basically translate from the technical description of a problem or incident to a functional/business description of problem or incident (and yes, I've seen "Office Space":). It's just not necessarily an inherent skillset of an average techie. Nor of an average most professions.

So it's not that chatgpt will replace a unique top-notch professional writer.

But it can hugely augment the other 90+% of bell curve when it comes to putting words in a row.

That's an idealization. Most writers are communicating the client's ideas, or trying to come up with ideas that they think the client would like. Most writers are trying to appeal to a particular audience for money, and new ideas mostly annoy their target market, who want a new thing pretty much like the last thing.

Aside from that; the thing that differentiates writers is their skill in writing words. A hell of a lot more people have "ideas" than have writing skill. There's absolutely no reason to think writers have better ones; I'd even predict that they would have worse ones because they spent so much time studying how to write, so other people have more time for varied and exotic experiences and learning.

Writers are usually writing about other people; people unlike themselves. What happens when those people can write about themselves rather than being interpreted?

Someone suggesting that their writing skills are now "obsolete" due to ChatGPT might not have had appreciable skills in the first place. The troubling part is they now see the existence of this language model as a reason to entirely abandon any effort at improving.

The ideal here seems entirely backwards.