Writing without AI
What has always been natural for mankind is dwindling: Writing without AI.
In Science, there was a piece on it recently, maybe interesting to compare with our own experiences?
https://www.science.org/content/article/i-had-unplug-ai-rediscover-my-love-writing?utm_campaign=Science+Magazine&utm_medium=ownedSocial
Y'all have a wonderful week
Marcel
Howdy, Stranger!
Categories
- 3K All Categories
- 153 Research & Reading
- 698 The Zettelkasten Method
- 11 Knowledge Work
- 101 Writing
- 469 Software & Gadgets
- 155 Workflows
- 731 The Archive
- 15 Plug-In Showcase
- 88 Resolved Issues
- 225 Projects Logs and Journals
- 83 Project: Zettelkasten.de
- 53 Critique my Zettel
- 175 Random
- 375 Introduce Yourselves!
Comments
Was it ever natural? Writing without AI was the only option. But writing itself has never been natural for humans. Writing has been invented only a few thousand years ago. Writing has been only taught to all boys for maybe 200 years or so, to girls even later. Even today widespread literacy isn't a given in all parts of the world.
There's no need to learn writing when you can talk to people—or machines. :-)
Maybe this is less about AI and more about rediscovering the value of composing text in writing? When I was researching the origin of the phrase "writing is thinking" I found a manual for composition in secondary schools, published in 1913. It's neither the first nor the only occurance of the phrase. But I find the context interesting: schools. It has been known in pedagogy for a while that school kids needed some convincing, why they should care about text composition.
I think we're discussing the same question as the author did over 100 years ago: "why should I learn to write?"
What's new about AI is that you can produce apparently perfect text, but without thinking and without understanding.
Obviously, the problem for losing part of ourselves by AI (or better: The reckless neglect we engage in) is urgent and important.
I have two points of criticism:
(1) The author is mainly highlighting the hedonic damage done by the process. The verdict is not the result of a comparative and weighing but a unidimensional evaluation of a single aspect. Imagine a scientist who could vastly improve his productivity with AI which would give him a shot at curing cancer. But because he doesn't enjoy the process, he forgoes the opportunity. This would be a person who values his personal pleasure more than lives saved and life-shattering tragedies prevented. I omit the next step which would be the ethical judgment. The point is that this article offers a very limited quality of evaluation.
You'll find phrases like "AI was making me a less critical scientist". But at the same time, there seems to be little complaint from others who were impressed by both the quantity and the quality of the output.
(2) "(R)eviewers commended the quality of my writing. Except I did not author most of those texts. I came up with the idea, the experiments, the data, and the references. (...) I would feed the chatbot my results and ask it to write the story I wanted to tell, and it would always oblige, never questioning my ideas."
The author critiques that "the papers [AI] produced seemed good on the surface, once I started to actually read them in detail, I found they often said very little." However, it seems that both his supervisor and his peers were praising him for the quality of his articles.
Something doesn't add up.
To me, the first point of criticism is more important as it feeds into the second point, too. It just takes too much time, both individually and collectively, to arrive at an evaluation that is reasonably complex enough to meet the demands of the problems if you approach the problem with such methods.
My personal experience with AI usage for writing is overall very positive. I might be a special case because my natural English writing is pretty jagged and the smoothing effect of AI offers reasonable counter-balance. But I never use AI to produce other than in utilitarian situations like title suggestions for articles.
My negative experiences are the result of me thinking that AI is surprisingly incompetent when I think it should be more confident. Comprehensive criticisms are as much slop as its output is.
I make it a point to develop specific skills for my needs. I have, for example, a line editor skill that I stuffed with the specific rules and patterns of writing advice that makes sense to me. It basically guarantees that certain bad habits of mine are highlighted and shoved in my face, once I was made conscious of them. But I have to manually edit the text myself.
With the first question, you introduce the problem how to distinguish between what is natural and non-natural per se. The same criticism could be pointed towards language itself as we are lingual creatures for a comparatively period of time.
I think we can steelman a little bit by accepting that writing without AI is a well-established and deeply examined cultural technique that feels "natural" enough to individuals who practice this cultural technique. (Culture is also natural...
)
There is also no need to chew since you can grind everything to mash. However, your teeth, your jaw, your digestion, your mood, and a lot more won't be in good shape (which you only can experience, when you are aware of the alternative).
The concept of "need" is not the right one to apply here.
With the rest, I have to agree.
I'd like to connect what you brought up with my criticism: You are pointing the attention towards the cognitive benefit. The students ask why they should engage in something unpleasurable when there is no clear upside for them. The counterpoint to them is a simple "because it makes you smarter which benefits you independent of writing".
A more comprehensive perspective, especially, when it is not questionable if writing is beneficial, is how to balance pleasure and outcome (given that you can improve the outcome by using AI).
My take is to go one step beyond and examine the purpose of pleasure itself. But this would be one step to far for this topic I think.
I'd add one caveat which makes this statement even more extreme: You can produce apparently perfect texts without thinking, understanding, and without immediate access to the quality of the text. If the produced artifact is beyond your competence, you can't even evaluate its quality.
I am a Zettler
Right now, I am deeply reflecting on my personal relationship with artificial intelligence in the realm of developing knowledge and ideas.
I use it for certain things, and there are significant impacts, both positive and negative.
I am taking the time to formulate a personal conclusion, which will inevitably be two-sided.
Excuse me for not expressing it properly. I meant: When writing (in a certain intellectual sense, beyond shopping lists and the like), then it has been without AI (naturally, because it was not around then).
I think it resonated with me because when reading that title, "Writing without zettelkasten" came to my mind. What huge gap this is! Verbose, semi-digested training material from AI, but concise, thoughtful input from my former self in the zettelkasten. And still, all material, also from AI, may inspire me. In those cases I turned to AI, it usually was for more examples for some teaching material. Mostly, it were inspirations rather than directly usable answers.
Bastien Le Guellec's' Science post opens with a remark about learning to write: "I felt like I was learning to write again". He describes the experience as intimidating and overwhelming. As a reader I immediately wondered why he makes the effort.
We learn about his intrinsic motiviation: "I had loved writing since I was a teenager." We learn about the pride he takes in his writing style. We also learn that he was frustrated by the supervisor's feedback.
Interesting hypothetical. Under what circumstances should individuals sacrifice personal pleasure for the greater good?
I don't think that Le Guellec is facing such a choice. AI-assistance not only harms his motivation, it also reduces the quality of his work (emphasis added): "And although the papers it produced seemed good on the surface, once I started to actually read them in detail, I found they often said very little. I realized writing is a process that not only transforms ideas into text, but also allows the writer to continuously question and refine those ideas."
The same could be said about keeping dogs or building houses. But I didn't say such thing. I'm emphasizing that nothing is "natural" about writing, it's cultural.
Yes, a cultural technique can feel natural to an individual. But that's a different claim than saying that this cultural technique has "always been natural for mankind".
Chewing isn't a cultural technique that people have to learn. It comes naturally. (And yes, we live in times where you can conveniently buy and process food that you don't need to chew anymore. And if you combine it with a sedentary life-style you ruin your body. But that's a different topic.)
No. It doesn't benefit them "independent" of writing. Writing itself has benefit.
Writing enables them to do things that they couldn't do without writing. Richard Feynman once told Charles Weiner in an interview that his writing doesn't record work done in the head, but that he works on paper. If you can't write, such work is out of your reach.
Yes! And that's why I consider AI not only a challenge to my text composition skills and subject-matter competence, but also to my reading comprehension skills.
Thanks for the clarification! I hope you don't mind that I went beyond what you intended. :-)
I enjoyed the article, thanks for sharing. I liked how the author describes both sides of writing, an intrinsic motivation and the struggles.
The relationship between AI, Zettelkasten and writing is an interesting topic. My instinct is to look at history. :-) Let's see where this threads goes. Thanks for the inspiring title!
Please reread the second point I brought up. Something doesn't add up.
You operationalized naturalness (or the lack of it) by time ("But writing itself has never been natural for humans. Writing has been invented only a few thousand years ago. Writing has been only taught to all boys for maybe 200 years or so, to girls even later."). Therefore, I chose a rather extreme example to show that what the concept of naturalness doesn't hold up to scrutiny, unless you want to say that being lingual at all is not natural to use people. Then I'd say that being people isn't natural to us, since we exist as homo sapiens sapiens only for a small fraction of our evolutionary history.
We don't need to nitpick words. "Always" as @Marcel_Liauw wrote is a typical casual hyperbolism to indicate that it is a long time. When I say "I always tied my shoes in this particular way.", you don't object and say: "Even when you were a baby...?". Hopefully at least.
I didn't make any claims about the naturalness or unnaturalness of chewing as you didn't refer to the naturalness when you claimed the lack of need of learn to write when you can talk to people and machines. The point is about utility being the framework to argue within opposed to need, not about naturalness.
Perhaps, I phrase my point differently: You tell the students that writing is not about becoming a better writer, but to improve as a thinker.
I am a Zettler
So much depends on your goals. If you have taken up painting and want to have a pleasant wall decoration, getting a chatbot to produce one may be much better than you painting one. If you want to get good at painting because you love to explore subtle patterns of light and dark, you need to develop that skill; the chatbot can't take its place.
If we all lose our cognitive skills then we won't be fit to pass judgement on the work of chatbots.
What doesn't add up is that the author's readers prefer the AI-assisted output. We don't know why. The author is critical of the AI-assisted work, but the readers apparently are not.
I do think that a few hundred or thousand years are rather short compared to the whole history of mankind. In my mind I was comparing the history of writing to the history of other human inventions like steam machine, horseback riding, farming, metallurgy.
I could have explained it better. I should have mentioned terms like "culture", "evolution" or "tool usage". Thanks for pointing out, what was missing.
I wouldn't have objected, because the "I" in the example implies that "always" refers to your life span. And because we're talking about an acquired skill, the common understanding is that you mean "always since you learned that skill".
I didn't get your point, because chewing isn't a learned skill.
Let me take the shoes as an example. I don't say that you should learn to tie shoes because it trains your fine motor skills, but because you will be able to tie your shoes.
If becoming a better thinker is the goal, one could argue that interacting with a well-designed AI tutor might be more effective than thinking in writing. The AI tutor adapts the challenges to your skill level, constantly tests your knowledge and gives you exact the kind of feedback you need to improve. This kind of learning is much more common in the history of humankind, considering that most of the time writing hadn't been invented yet or was limited to a small group.
Composing texts in writing is a skill that needs to be learned. So why invest the time and effort to learn the skill, now that we have AI?
I think we should listen to educators who encouraged students to write, like the one mentioned earlier. The old book suggests that thinking in writing increases the precision of your thinking.
But who says that he is still correct? Maybe an apprenticeship or regular debate with an AI is a more effective training tool? Maybe you write better papers when you don't chat with yourself but with a bot?
@harr said:
I am too sceptic to jump on AI as enthusiastically as others (often younger people) do. I found an example fascinating by a pupil who prepared his final exam (Abiprüfung). Here is a screenshot.
If I translate this to myself and my zettelkasten, maybe supported by some smart AI, it might be a fruitful story, following @harr's thoughts. Use of AI has been reported in this forum more than once. Maybe I try more than before after this discussion.
I have read the article you posted, thank you.
I am not particularly surprised by what I read.
It is something I foresaw a few years ago.
Our cognitive abilities need to be trained; otherwise, we lose the habit of using them and, consequently, their effectiveness.
The use of AI un-trains us in certain cognitive skills. There is a huge hidden cost behind the productivity benefits it offers.
I am always very afraid of this effect. I use AI for some things, but I have reflected deeply on how to try to avoid its negative effects. After I've saw this discussion, I've spent hours today to reorganize and consolidate my position about AI in these contexts.
Of one thing I am certain: I was very fortunate to discover the Zettelkasten method first, and only later ChatGPT. Had it been the other way around, my cognitive life would have turned into a disaster. The practice of Zettelkasten, learning and knowing it have, in a way, shielded me significantly against the effects discussed in the article.
P.S. I wrote this post without any AI, but I translated it into English using Gemini.
Fully aware that while this practice will preserve my ability to reason and put my thoughts on paper, I will end up losing my ability to translate :-)
Update
In my reflection today, I built a framework crossing four different motivations for using AI I've identified in cognitive contexts (producing something based on ideas and knowledge, developing ideas and knowledge, learning, and art) with my three distinct personal motivations for using Zettelkasten (producing something, developing knowledge and skills, and deriving gratification from what I do). Within this framework, I tried to map out the risks, opportunities, and even the necessity of using AI.
It is something very personal, not based on any research that could support my ideas, but purely on my own feelings and reasoning. Still, it remains a bit more nuanced than simply saying "AI is good" or "AI is evil." I've realized that there is no single path in one unique direction; in every context of use, the weight of risks and opportunities shifts.
I arrived at a particularly deep conviction, for example: a under 18 student should not have access to AI. Studying is not a productivity exercise.
I set myself a challenge: try to do all this work without any help from AI, using the Zettelkasten. I succeeded—it took some time, but perhaps I am still a thinking being.