There’s something brewing at OpenAI, and it’s not their often promised AGI. While Altman seduces investors with the speculative promises of Artificial General Intelligence and ever more powerful versions of GPT, a very real play is being made for much more concrete territory. Over the past few months, a number of product releases and updates all point in the same direction: OpenAI doesn’t see a place in its new world order for writers and teachers of writing.
ChatGPT and the End of Writing
Back in 2022, the release of ChatGPT and GPT-3.5 was heralded with a number of articles prophesying the end of high school English and the death of the college essay. While these reactionary articles haven’t quite played out with the existential clout they threatened, and while certain educators called bullshit on the potential of ChatGPT to bring an entire mode of communication to its knees, there are some reasons to be worried.
First of all, those articles were written back in 2022, when the nascent chatbot was still producing essays that could perhaps pass muster in middle school, but were hardly up to the challenge of senior secondary or tertiary level writing. GPT-4, the next cab off the rank, was a suitable step up from its November 2022 predecessor but it still didn’t offer much for actual writers. It’s stilted, hackneyed phrases and overused words (delving, navigating, endless dot point lists) have become something of a meme int he past 2 years.
But the technology just keeps on improving, and between the more capable language skills of a model like Claude 3 Opus (still better at writing that its own successor, 3.5 Sonnet) and the apparent lack difficulties the general public faces in discerning AI writing from human, we seem to have reached a tipping point.
You’d be forgiven for thinking that a company that has scraped every available piece of text and created an AI model that works predominately with language would be treading lightly on the threat of reducing the entire internet to slop. But instead of building tools to help writers, OpenAI seems set on replacing them. And it’s not an attractive proposition.
Tools for Writers
Take ‘Canvas’, for example. It’s a recent addition to the 4o family of models that OpenAI calls “a new way of working together—not just through conversation, but by creating and refining ideas side by side.” According to OpenAI, the tools a writer needs to collaborate with AI are:
- Suggest edits: ChatGPT offers inline suggestions and feedback.
- Adjust the length: Edits the document length to be shorter or longer.
- Change reading level: Adjusts the reading level, from Kindergarten to Graduate School.
- Add final polish: Checks for grammar, clarity, and consistency.
- Add emojis: Adds relevant emojis for emphasis and color.
Take a minute to think those through. Whether you’re a writer, a teacher of writing, or just someone who likes to occasionally put words down on paper or screen there should be a few things that seem a little off.
ChatGPT’s suite of tools to support writers don’t actually do anything to benefit the author. They’re reader-centric, and even worse position the reader as an inane, halfwitted moron who can only consume texts if they’re short, simple, and coated in emojis.
It’s like somebody took a few dozen TikTok videos on how to write social media copy, mangled the transcripts through ChatGPT, and spat out a listicle of the most useful tools for writers. In fact, I suspect that’s probably quite close to the truth.
If you compare the ‘uses ‘writer’s’ toolbar to the equivalent offered for programmers using Canvas’s coding feature, you’ll see what I mean:
Coding shortcuts include:
- Port to a language: Translates your code into JavaScript, TypeScript, Python, Java, C++, or PHP.
- Review code: ChatGPT provides inline suggestions to improve your code.
- Add logs: Inserts print statements to help you debug and understand your code.
- Add comments: Adds comments to the code to make it easier to understand.
- Fix bugs: Detects and rewrites problematic code to resolve errors.
So programmers get tools to review and refactor code, fix errors, and add logs: all useful parts of the programming process. Writers, on the other hand, get brevity, simplicity, and emojis.
A Student’s Guide to Raising the Dead
Not content with helping writers dumb down their work, OpenAI has finally turned its attention fully towards their biggest market segment: students. In their recently published Student’s Guide to Writing with ChatGPT they offer 12 suggestions for “ethically” using the chatbot to extrude words. Amongst the advice, my personal favourite is this:
I’m sure that Simone de Beauvoir, philosopher, author, feminist would love the idea of being resurrected through the Ouija board of OpenAI’s Large Language Model. The irony of suggesting a notoriously biased and sexist piece of software as a suitable vehicle for the digital necromancy of feminist de Beauvoir seems to have been lost on the creator of the guide. Presumably because the creator of the guide was, in fact, ChatGPT.
Others have already written critiques of the guide, and more broadly there has been a rise lately of writing instructors writing about resistance and refusal.
I think we can all say no to digital necromancy.
New and Improved!
And just this morning, as I sat down to prepare a session titled “writing against AI”, OpenAI offered us this gem:
No explanation. No evidence. No research. Not even a blog post. Just an announcement of improved creative writing skills (which went much more viral on X than LinkedIn, but I won’t be sharing any links to that platform since I deleted my account and moved to Bluesky).
ChatGPT now has “more natural, engaging, and tailored writing to improve relevance and readability”. Note again the focus on readability, as if people need this technology to simplify ideas for them. Let’s see:
I’m not impressed, and neither are the majority of the commenters on the post I shared earlier today with the same (admittedly brief and slightly odd) prompt:
In an earlier post about Canvas, when it was first released, I offered a few suggestions for what OpenAI could do to actually support writers. Here they are again:
- Check logic and structure
- Suggest tonal changes
- Improve clarity
- Change form (essay to blog to speech to…)
- Fact check
- Suggest counterarguments
- Improve cohesion and internal consistency
- Check narrative structure
- Anything other than add emojis ✌️
The problem is, OpenAI clearly doesn’t want to help writers. It wants to replace them. It wants to suck the joy out of writing and replace it with efficient, easy to understand microprose that goes down reeeeeal easy. Fast food for readers. Slop.
As a writer, a teacher of writing, and someone that actually gets a lot of use out of these technologies, it stings. I’ve written plenty of posts about how AI can be used as part of the writing process, teaching writing with and against AI, and how I use Generative AI myself as an author. But the way I use AI and the way I encourage students and teachers to think about AI is at odds with the message coming from the loudest AI developer of them all.
And I’m worried that the message will ultimately get lost in a sea of efficient, readable, emoji-riddled goo.
Want to learn more about GenAI professional development and advisory services, or just have questions or comments? Get in touch:
Leave a Reply