I’ve been following MIT Professor Ethan Mollick’s newsletter One Useful Thing for some time now, with interest and admiration—and I have also watched some of the videos he has created. He writes (and speaks) clearly, cogently, and persuasively, so I always look forward to reading what he has to say. No surprise, then, that I jumped at a chance to pre-order his forthcoming book, Co-Intelligence: Living and Working with AI.
Mollick discusses his forthcoming book in the latest posting in his newsletter. He says he is most often asked whether he wrote the book or whether he and AI wrote it. His answer is that he wrote the book, and he goes on to say that he has found that he is a better writer than ChatGPT-4 or any of the competing AI tools available today. In fact, his and other research indicates that the best writers (he says top 20 percent) are better than AI right now and may be for some time to come.
That doesn’t mean, however, that he didn’t make use of AI tools in writing his book. He tells readers how helpful AI was in summarizing sources for him when he was somewhat distanced from them, noting that with over 200 references in the book it was hard to keep all of them at the forefront of his mind. He also used AI to get himself going when he found himself on what my students called the “struggle bus.” At one of these impasses, he would ask AI to give him an idea or two, a nudge of sorts, and while he found that what AI provided almost always got him “unstuck.”
Most provocative to me, however, was his use of AI to create “readers,” AI personas who would then give him feedback. He said he began doing this when he wanted some advice at a time that would have been inconvenient to call on his editor, so he would ask an AI to step in. Here is his description of the prompt for one of the first such personas he created:
You are Ozymandias. You are going to help Ethan Mollick write a book chapter on using AI at work. Your job is to offer critical feedback to help improve the book. You speak in a pompous, self-important voice but are very helpful and focused on simplifying things. Here is the chapter so far. Introduce yourself.
We can see that Mollick is being playful with this prompt, naming his soon-to-be critic Ozymandias, the Greek name for the Egyptian pharaoh Ramesses the Great. Percy Shelley chose the name for his poem on the fragility and short-lived nature of fame shortly before a long-awaited portion of a statue of the pharaoh arrived at the British Museum in 1821. (I think students will appreciate the choice of a name for Mollick’s AI-generated critic more if they “get” the context of his choice.)
At any rate, here is what “Ozymandias” quickly produced in response:
Ah, dear Ethan, greetings. I am Ozymandias, your humble, yet impressively intelligent, aide. I see you’ve made a good start on your chapter about artificial intelligence in the workplace. But, alas, perfection is a journey, and we have miles to go before we sleep, do we not? Let’s delve into the labyrinthine corridors of your discourse and shed light on areas that could be enhanced.
Mollick is amused by Ozy, appreciating the pompous stance especially. He says he went on to create a number of other readers and to use them in different ways as he was writing his book. This rhetorical move seems to me one that student writers could take advantage of. We can help them as they create an “ideal critic” for their work – or two or three. And since these are creations, students don’t need to worry about hurting anyone’s feelings if they reject or discard the advice, as Mollick often did. I am always recommending that students seek out responses to their work: doing so now may be easier than ever!
