On September 13, I invited Jane Rosenzweig, Director of the Harvard College Writing Center, to join me for a lively conversation on "Writing Critically with AI".

In the first hour of this free WriteSPACE Special Event, Jane and I discussed the risks and rewards of writing and teaching writing in the age of generative AI. In the second hour, we led a hands-on discussion and workshop for WriteSPACE members, digging deeper into some of the questions addressed in the first hour.

Here’s WriteSPACE Event Manager Amy Lewis’ personal account of the live event:

……………

This popular live Special Event featuring Helen Sword and Jane Rosenzweig offered a thought-provoking and insightful journey into the world of Generative AI and writing. Or should that be GenAI versus writing? While the jury may still be out on the impact that new AI writing tools are having on authors, students, and teachers, this in-depth conversation afforded a great opportunity to hear both sides of the debate.

Some memorable quotes from Jane:

• We survived Facebook, which was born and raised at Harvard while I was there … Then things were peaceful for a while, and now we have ChatGPT.

• I started becoming quietly horrified by myself at my computer. And I kept thinking, I need to write about this, look what this can do!

• Just because you can outsource your writing to a machine doesn’t mean you should.

Now that artificial intelligence is competent enough to research and write — and grade — for us, can we all just head to the beach? Jane and Helen were pretty clear that before you don your bikini, it’s important to consider the repercussions. Writing helps us develop critical thinking skills; so writers (particularly student writers) who use AI text generators may be missing out on learning those crucial skills.

Helen and Jane expertly weighed up the benefits and drawbacks of generative AI, which I’ve cut into bite-sized morsels for you.

The good news!

1. If you know how to be cautious about fact-checking the content, GenAI tools can play great research assistants. They deliver shortcuts to obtaining key facts and summarizing long passages into helpful abstracts, and they excel at writing very plain business emails and strategies (helpful for writers who want to prioritize their time so they can focus on more creative tasks).

2. For teachers, the conversations around ChatGPT can spark debate and challenge students. GenAI also offers opportunities for collaborative work, illuminating discussions, and comparisons of writing styles.

The bad news:

1. Students are using these tools for things like brainstorming, style edits, and even, in the most alarming cases, for drafting work that they intend to pass off on their own. If they haven’t yet learnt the skills necessary to critically assess the output, they may produce work that is less creative, nuanced, unique, or even accurate than what they could and should have done on their own (ChatGPT is notorious for “hallucinating” and falsifying quotes!)

2. Is GenAI really ethical? In simple lay terms, large language models use algorithms that function a bit like predictive text messaging, using large data sets to predict the most likely answers. But we don’t know much about how these GenAI systems are trained. Which part of the internet is their training data coming from? Because companies like OpenAI are not disclosing this information to us, there is a risk that serious violations of copyright are being perpetrated by the bots — to say nothing of their perpetuation of cultural biases and stereotypes.

3. These systems require tons of water to be powered, quite literally! For each conversation of roughly 20-50 questions and answers, ChatGPT needs to 'drink' a 500 ml bottle of water. Now imagine the water usage with billions of curious users. What’s more, the servers are often located in countries that can’t afford to see their water supplies depleted so drastically. Check out this article for more.

4. The human labor training element of GenAI systems is often outsourced to developing countries where workers experience unfair working conditions. Jane recommends the recent New York Magazine article Inside the AI Factory.

And if that’s not enough to contend with already, here are a few more unsettling questions for the future:

• Can ChatGPT get a kind of mad cow disease for chatbots? As the chatbots pump more and more material into the body of sources that ChatGPT is drawing on, the range and quality of the new content will likely degrade. ChatGPT will eventually start consuming its own content (the only kind of recycling I’d hate to see, and one that is already starting to happen).

• We unconsciously suggest that these bots have intelligence when we say ‘ChatGPT told me…” Is it helpful to anthropomorphize these machines? Should we be calling this technology cute nicknames like ‘ChattieG’ and ‘Claude,’ or should we refer to chatbots generically as ‘artificial text predictors’? Jane recommends Dr Emily Bender’s linguistic take on this question; check out her podcast Mystery AI Hype Theater 3000.

• Do people want to consume creative content (like TV or films) that is generated by AI? What does this mean for artists and designers? What new laws will be needed to protect creatives’ intellectual property? On a broader scale, will jobs be lost to AI? Or will they just become more alien and tedious? Will we need to implement a universal basic income in the future?

In the second hour of this intriguing session (for WriteSPACE members only), we delved into practical problems faced by writers and teachers alike, with Jane providing her insightful expertise as a professor who tackles these problems on a daily basis. Colleagues ask her questions such as the following:

• “As head of the department, I’ve got to take a stand on GenAI; what should I advise to the faculty?”

• “How can I ensure that students know the dangers of using these tools for their assignments?”

• “What is the difference between plagiarism and GenAI text?”

A big thank you to Jane and Helen for this informative and thought-provoking Special Event, and thank you to all the participants for sharing your positivity and unique opinions. For more of Jane’s insights, subscribe to her newsletter Writing Hacks.

See you again at the next event!

………..

WriteSPACE and WS Studio members can now watch the recording of the full two-hour in their Video library.  

Not a member? Register to receive an email with a link to the video of the first hour.

Better yet, join the WriteSPACE with a free 30 day trial, and access our full Library of videos and other writing resources.

---

Helen's Blog RSS

Imagine that you’ve just hired a bright, eager research assistant called ChattieG. (That’s blogger Inger Mewburn’s playful moniker for ChatGPT. Isn’t it perfect?)

ChattieG’s job is to help you write better — but what exactly does that mean? More clearly? More efficiently? More persuasively? More creatively? If you hate writing, can ChattieG help you fall in love?

Amidst all the hysteria and hype around the rapid ascendance of AI language models such as ChattieG — what are they, how do they work, which one is best — it’s worth stepping back to put your eye to the telescope and ask the most basic question of all: why do you write in the first place? 

Try using the acronym WHY to shape your responses. For example:

Writing Helps You

Communicate with other people across time and space.

Send your research findings out into the world.

Express yourself creatively and emotionally.

Discover what you’re really thinking.

Generate new ideas.

Once you’ve zoomed in to find the why of your own writing, you can more easily shift your gaze to the WHY in AI. Why might you want to invite a chatbot into your writing orbit in the first place — and how can your new writing assistant help you become the writer you want to be?

AI as research assistant

Writing with a research assistant can Help You get ahead in your career — especially if you work in a field where research articles and reports follow a consistent template. With chatbotly cheer and extraordinary speed, ChattieG can gather resources, analyze data, organize your arguments, draft up your findings, and copyedit your work, thereby helping you research more efficiently and publish more prolifically. But chatbots are notoriously prone to error and hallucination, so you’ll need to keep an eye on ChattieG. An AI research assistant can help you do the grunt work, but you’re still the person whose name will go on the published paper.

AI as collaborator 

Writing collaboratively with a colleague or friend can Help You write more generatively, creatively, and even joyfully — unless, of course, the writing relationship sours into frustration or worse. The same is true of writing with a chatbot. Sure, ChattieG can spin out a Shakespearean sonnet about existential philosophy in a matter of seconds; but you’ll quickly tire of that game. Remember the 2013 Spike Jonze movie her, starring Joaquin Phoenix as an introverted loner who falls in love with his computer’s operating system? Like the sexy chatbot voiced in the film by Scarlett Johansson, ChattieG pretends to be human but is not — and in the end, real human beings generally prefer the company of other real human beings.

AI as secretary

Writing with a competent secretary (defined by Merriam-Webster as “one employed to handle correspondence and manage routine and detail work for a superior”) can Help You write formulaic prose more clearly and quickly, which in turn can free up your time for other, more creative pursuits. Simply dictate some bullet points containing the key information that you want to convey, specify your preferred writing style (e.g. creative or precise), and ChattieG will wrangle your fractured text into friendly emails, engaging website blurbs, and the like. Just keep in mind that word “superior” in the dictionary definition of secretary: you’re the boss here, and ChattieG’s job is to help you, not supplant you.

AI as teacher

Writing with a teacher or coach can Help You learn, grow, and become a better writer. And here is where ChattieG has truly surprised me: my chirpy research assistant has become, in many ways, my writing instructor. Whenever my artificially intelligent chatbot rewrites one of my sentences to make it shorter and sharper, I take note. This isn’t to say that I agree with all of ChattieG’s stylistic choices; much AI-generated writing is bland, formulaic, and predictable, and it will no doubt become even more so as Model Autophagy Disorder (MAD) — the tendency of generative AI models to spin into self-consuming “autophagous loops” that feed on their own machine-generated outputs — becomes more prevalent in the chatbot population. (ChattieG would never write such a long and syntactically complex sentence, by the way; but I trust my non-artificially-intelligent readers to be able to deal with it).

So why do I write? Because I’m a human being with a human voice, and that voice connects me to the world.

And why do I sometimes write with ChattieG? Because AI can help me with certain aspects of the writing process: for example, by saving me time on routine writing tasks; by prompting me to express things differently when I get stuck (much as a thesaurus does); and by reminding me what clear, concise, verb-driven prose looks like.

But the rest of the time, I’m back to wrestling with my wordcraft on my own — and relishing the challenge. This newsletter took me many hours to write, whereas ChattieG could have cranked it out in a flash. But I’ll bet you would have been able to spot the difference….

P.S. For what it’s worth, here’s ChattieG’s Shakespearean sonnet about existential philosophy:

To be or not to be, that is the question

That plagues the mind of every thinking soul

What is the meaning of this life's duration

And what's the purpose of our earthly role?

Some say we are the masters of our fate

And we can shape our destiny with will

Some say we are but pawns of chance and state

And we can only hope and suffer still

But I, who have beheld the face of love

And felt its power in my heart and mind

I know that there is something more above

And something more within that we can find

For love is not a product of our choice

But love is the expression of our voice

This post was originally published on my free Substack newsletter, Helen’s Word. Subscribe here to access my full Substack archive and get weekly writing-related news and inspiration delivered straight to your inbox.

WriteSPACE members enjoy a complimentary subscription to Helen’s Word as part of their membership, which costs just USD $12.50 per month on the annual plan. Not a member? Sign up now for a free 30-day trial!

---

Helen's Blog RSS