Should you use AI to write your LinkedIn posts?AI scepticism is at its peak, and Substackers debate the introduction of Pangram to check whether posts are 100% human. Part 1 of 2 about how AI can help you write for a general audience
It’s no joke. Last year, we still rolled our eyes at students’ AI-generated essays and reports. This year, outrage at “LinkedIn’s AI slop” and “Pangram Police” has reached new heights — not least here on Substack, where the introduction of Pangram as a feature that allows everyone to scan posts for its provenance has led to yet another great “gnashing of teeth (that could destroy whole planets)” — a great phrase that I joyfully attribute to the great Simon K Jones. So, go ahead, and scan this weekly missive with Pangram. I did write it myself, but frankly, by now, Claude and I are so good at fooling the Pangram detector (which is by no means as accurate as they claim) that you wouldn’t know if it was if I did not tell you. #SorryNotSorry. Catch-up service:
In this high noon of feelings on technology’s latest moral panic, let me pour some truly ice-cold water over the whole thing. Because I’m extra caustic right— I just had a big grant application rejected. No justification. No feedback. No resubmission. Just a short email for over three months of work... OK, let’s move on. Academics on social mediaWhen I write for social media, for example, Notes here on Substack, or for LinkedIn, I normally write myself, but at other times have my AI provide the summary. And you know what I’ve found? The AI-written stuff almost always performs better than what I write. Yes, people indeed like slop. Or do they? Let me be honest — I see a lot of people just post AI-generated text. I could not help myself the other day on Instagram, and wrote under a post “Hello, Claude!” because I could see the linguistic tics of the Opus family of models clear as day. But then I also extend invitations to academic colleagues to write for our Business Research Unpacked Substack when I see their interesting research packaged for a LinkedIn audience. Even when I look at it and think: that’s ChatGPT writing, not so-and-so. Why? Because they actually made an effort to make their research accessible. And that is hard, for a range of reasons we do not fully comprehend and cannot easily overcome. The curse of knowledgeEver heard of that one? Funnily enough, not a widely used phrased in among experts. We are more “knowledge is power” kind of people. But it was coined by a bunch of economists (which may explain why it is not popular): In 1989 three economists, Colin Camerer, George Loewenstein and Martin Weber, ran experiments in which better informed people were asked to predict how less informed people would behave. They could not do it. They kept assuming others knew what they knew, even when the mistake cost them money. They called it the curse of knowledge. (This was Claude, btw. You did not think I’d read that myself, did you? I’ve got better things to do.) Steven Pinker later argued this is the best single explanation for why clever people write badly. Bad academic prose is rarely vanity or laziness. It simply does not occur to the writer that the reader lacks the background, so the shorthand and the “linguistic gates” go unnoticed. The awkward part is that none of this is visible from the inside. You cannot feel yourself doing it, which is why “write more clearly” is such useless advice. Nobody can proofread for their own blind spot.
Of course, we can ask someone to give it a friendly read. Two problems with this: Is it another academic? Can you guess what the problem might be? Plus they won’t have much time, either. If it’s not another academic, are they friends & family? Well, by the very nature of their relationship with you, their primary reason to agree is to manage their relationship with you, and guess what that means for the critique of your writing. Sure, you can hire a professional copy-editor, because we have such vast research budgets and enormously generous employers (not). And what would a copy-editor rather do than help an academic understand the tropes of writing for social media? You could buy a book and learn how to do it, though the book was likely out of date already when it was published. Also, what about the pile of academic reading you were supposed to be doing for the next journal submission? So, this is the problem. Next week: what you need to know to better understand the difference between academic and general audience “internet” writing. In the meantime, check out Omeghie Okoyomoh’s TikTok that unpacks social science and entrepreneurship research for the YPs out there. For full subscribers, some reading suggestions for excellent science communications here on Substack, and how you can set this up (with or without your AI) as a toolkit for your own writing. Too many emails, not everything relevant to your interests?Manage which type of content you receive from History in Organizations:
There’s a shortcut, too: the Unsubscribe link at the bottom of every email will offer to unsubscribe you from individual sections rather than the whole newsletter. So if one post too many lands in your inbox, use that link to drop the section, not me (please)! ReferencesCamerer, C., Loewenstein, G., & Weber, M. (1989). The Curse of Knowledge in Economic Settings: An Experimental Analysis. Journal of Political Economy, 97(5), 1232–1254. https://doi.org/10.1086/261651 Pinker, S. (2014a). The Sense of Style: The Thinking Person’s Guide to Writing in the 21st Century. Viking. Pinker, S. (2014b, September 26). Why Academics Stink at Writing. The Chronicle of Higher Education. https://www.chronicle.com/article/why-academics-stink-at-writing/ You're currently a free subscriber to History in Organizations. For the full experience, upgrade your subscription.
|
No comments:
Post a Comment