Writing is dead; long live writing!
My teammate Georgia Iacovou (a brilliant, un-automatable tech writer, producer and researcher) made this hand-drawn comic to critique this article.

Writing is dead; long live writing!

Writing is about to enter its digital photography era.

The process of writing, and the role of professional writers, is starting to shift dramatically, and this process is only going to accelerate. A few days ago (June 21, 2024), the Wall Street Journal reported that “Freelance jobs that require basic writing, coding or translation are disappearing across postings on job board Upwork, said Kelly Monahan, managing director of the company’s Research Institute.” This raises a lot of questions about writing and writers, and their future role in the economy and workforce. It’s tempting to tell a story in which AI replaces writers, and this might very well be the case for certain kinds of writing or writers. I think this is too simplistic. This story is contradicted by some previous technology introductions, such as digital photography. Digital photography didn’t end photographers. Quite the opposite, there are more photographers (professional and amateur) than ever before. I think evolution to digital photography from analog is an analogy for what is happening to writing now…and can provide an interesting template to think about the potential pitfalls and opportunities.

Before the advent of digital photography, capturing images was a laborious, expensive process. Film was costly, each shot precious. Fixing mistakes in post-production was time-consuming, if possible at all. Early digital cameras, while inferior in quality to their analog counterparts, offered one game-changing advantage: instant feedback. Photographers could immediately see if they got the shot, enabling faster iteration and experimentation. Suddenly, each individual frame mattered less. Shooters could afford to waste exposures chasing spontaneity and serendipity in ways unthinkable to all but the wealthiest artists in the film era.

As digital cameras evolved and photo editing software like Photoshop emerged, so too did the workflow. No longer did photographers carefully compose a handful of shots to be developed later. They'd rattle off hundreds or thousands of frames, then winnow the collection down to a choice batch. From there, they'd tweak and tune - adjusting lighting, removing blemishes, swapping backgrounds. Even before Photoshop, darkroom techniques allowed manipulating the final image. Rarely would a professional simply use an unedited photo straight from the camera.

The rise of digital photography upended the traditional economics of the industry. In the analog era, photographers could charge a premium for their expertise and the scarcity of their product. Developing and printing photographs was a time-consuming and expensive process, which meant that clients were often willing to pay handsomely for a professional's services.

As digital cameras became more ubiquitous and post-production tools more accessible, that scarcity diminished. Suddenly, anyone with a decent eye and a modicum of technical skill could produce high-quality images. This led to a proliferation of amateur and semi-professional photographers, many of whom were willing to work for a fraction of what established pros charged.

Writing is on the cusp of an analogous transformation. For most of history, crafting prose has been painstaking, the process of translating an idea to the page gradual and meticulous. While the cost of writing and printing has declined over time, it remains a slow, deliberate act. Professional writers know the first draft is never the last. Like photographers, they revise, rework, and edit, but the process is time and labor intensive. Clients commissioning written work are lucky to get two or three variations, let alone dozens or hundreds.

Advances in natural language processing and generation, embodied in large language models like GPT-3, are poised to upend the craft of writing in ways both profound and unpredictable. The ability to conjure up multiple drafts with a few keystrokes, each with a different style, tone, or perspective, will fundamentally alter the way writers work. Iterating on a piece will become more like refining a photograph in Lightroom or Photoshop, a process of progressive refinement rather than linear composition. Initial written outputs will be merely a starting point, a seed to be iterated upon. Writers will generate myriad versions, tweaking tone, swapping characters, and adjusting style as easily as applying an Instagram filter. Some may start with a single draft and branch out, others with many and narrow down. The "authenticity" of the process will matter as much as whether a photographer shot on film or digital. The analog look may retain its appeal as an artistic choice, but will be just one option among many.

I’ve been experimenting with interfaces for how this new form of writing might happen. How does a writer analyze 100s of different variations on text at a time? How does a reader explore variations on a idea in a way that is fluid and interpretable?

One simple idea I came up with is a coordinate plane with different vectors of change. The original writing sample is in the middle, with the hundreds of minor variations radiating out from it. I can imagine this being an option or feature within a “Photoshop for writing” software in the near future: https://coordinate-plane.vercel.app/


As with photography, this shift will likely expand the pool of people who can produce compelling written content. Just as Instagram and VSCO have birthed legions of amateur photographers with sizable followings, the proliferation of AI-powered writing tools may spawn a new generation of "micro-authors" catering to niche audiences. The notion of a bespoke novel or personalized comic book may seem far-fetched today, but so too did the idea of a high-quality camera in every pocket a few decades ago. Bespoke novels could commemorate weddings, tailored comic books fete graduations. While readership for any given piece may shrink, the potential to create incredibly bespoke, long form, high quality content will explode.

But just as the rise of digital photography didn't spell the end of professional photographers, the emergence of AI writing tools doesn't necessarily mean the demise of professional writers. It does mean the task of professional writing is going to change. I don’t think this will be a good thing for all writers. Some will be able to adapt their process to new client and audience expectations, and some will not. I don’t want to understand the difficulty of this, especially for writers who have spent decades honing their craft. Many writers will be forced to adapt in ways that feel uncomfortable or even painful, simply to remain competitive in a changing market.

But there is also more to it than just economics – AI-enabled writing calls into question what it means to be a writer in the first place. The prospect of algorithms generating coherent and even compelling prose is understandably unsettling for those who have devoted their lives to the art of writing. It's not just a matter of learning new skills or adapting to new technologies; for some, the very idea of collaborating with AI or incorporating machine-generated text into their work may feel like a betrayal of their craft.

Looking farther into the future, writing might be less like word-crafting and more like world-crafting. Writers might be responsible not for creating individual pieces of writing, but creating spaces of knowledge and information that can each produce infinite individual pieces of writing. Playing around with this, I created a simple, wonky demo that produces science fiction stories, images for the stories, and HTML/CSS for the website. Instead of writing one science fiction story, I created a simple world that the reader can use to access infinite stories. The generative UI part is cool, but what I am actually most proud of is that I think I actually got the "AI smell" out of the writing:

If you create any of these that really resonate to you, capture them and send them to me!

I sent this article out to some people for comments and feedback before publishing, and one reviewer (@mara) put this in a way I thought really brought the point home:

”While everything you are saying about photography is true in terms of ease of access and cheap tools and photography becoming accessible to many, photography as a profession or an art form haven’t disappeared. We still hire professional photographers and we still have photography exhibits. Some photographers (the ones who work for National Geo) go to places most don’t go to and are able to capture something unique that takes years of training to do well. In fact, some of the photographers are quite celebrated (have you seen Stephen Wilkes’ work?). I think similar things will be happening with writing—yes, many more people will be able to have access to writing tools but just like photography, good writing requires unique personal experiences, going to places that are not accessible to others, seeing things in unique ways. Data does not equal human experience and it is this unique human experience that good writers, some of whom might be using the AI tools as a part of their process, just like trained photographers use the digital tools, will use to continue in their craft. So I am on team human :-).”

The digital photography revolution transformed the way we create, consume, and think about images. The coming wave of AI-powered writing tools promises to do the same for the written word. The question for writers today is not whether this shift will happen, but how they will respond when it does. The question for the rest of us is what kind of writing and writers we want in our lives. No matter what technology is used, how do we support excellent storytellers, and avoid a world of endless AI slop? How can we encourage technology development that supports “team human?”

Note: this article was inspired by a conversation I had with the brilliant Ben Hamamoto of Institute for the Future, so please credit any interesting parts to him and anything of poor quality to me.

Follow more from me and Handshake at handshakefyi.substack.com.

Check out our course on AI Agent Design at maven.com/handshake.

Robert Franklin, CSP

Founder & CEO, Journey To Agility | Breaking down Al strategies, trends and tools for all tech professionals and enterprises

8 个月

Jeremy - Great meeting up yesterday at AIQCON. What are you thinking about when you say it is just the beginning of the story?

要查看或添加评论,请登录

Jeremy Kirshbaum的更多文章

社区洞察

其他会员也浏览了