Me and ChatGPT
I’ve been wondering about A.I. It seems like everyone is, doesn’t it? When the podcast Freakonomics did a 3-episode series on it, the hosts said the goal was to help people understand A.I., not just to speculate about all the horrible (or wonderful—but it seems like mostly we hear about horrible) things it might cause in the future, I thought that would be worth a listen. I recommend the series.
Then, another podcast I listen to, Nomad, did an episode where the host “interviewed” ChatGPT, especially about its intersection with faith. That was interesting, too, but I got tired of the responses from ChatGPT. They became rather repetitive and fake-sounding to me. I sympathized with what Joy Brooks, the co-host, said about it after the interview. Interesting, nonetheless. (Links below.)
Recently, I was having a meal with several people and one of them, a young man, talked about how he is using ChatGPT in his job. It was kind of remarkable to me how quickly it has already become just the way he works. He gave several examples of ways he uses it to spark ideas. He also talked about using ChatGPT to what you might call “professionalize” his work. He has to write reports as part of his job and the reports need to be in a certain professional format. I can’t remember the term for it but it was something to do with a professional therapist format. He said he inputs his report and then tells ChatGPT to put it into that professional format and it does that, no problem. Then he turns that in. Seems like a smart way to use the technology to me. And prompts lots of questions.
The other day, I had an idea for a blog. On that same podcast I mentioned earlier, Nomad, Joy said she thought her job as a therapist using “embodied coregulation” would be safe from being replaced by A.I.. “Embodied coreguation”? Have you heard of that? I had not! The other host teased her: “Embodied coregulation? I’m going to have to use that.” Joy admitted friends have to call her out for using those kinds of terms. I googled it and found that basically, it means the way people calm down when they are hugged or held by others. Their heartbeats and breath and so on slowly match the other person’s, slowing them and calming them down.
It reminded me of the way I can get babies to stop crying, sometimes even when their mothers can’t. It doesn’t always work, but often, if I take a crying child, stand and rock them gently, and focus on keeping my breath and my heart slow, the baby will calm down, and even fall asleep. I never knew that was called embodied coregulation, but there you go.
Anyway, I thought I’d try to write a blog about the words “embodied coregulation.” Then I thought, “Hey, I wonder what ChatGPT would write if I asked it to write the blog.” According to what I’ve learned, it is very important to give detailed prompts to ChatGPT. And I’ve heard people use “in the style of” in the prompt, to get ChatGPT to produce something that sounds like a real person. I opened an account on ChatGPT and prompted: “write a blog in the style of my blog, www.mavismoon.com, about "embodied coregulation." It responded with a fairly long entry in letter form (beginning, “Hello dear readers of MavisMoon.com,…” and ending, “With love and light, | [Your Name] | www.mavismoon.com”) that sounded kind of like an encyclopedia to me, complete with sub-headings.
Hmm. That was not what I had in mind, nor was it what I thought was my style. I thought my style was musing about things “on my heart and mind.” But I do link to lots of things written by others—poetry, articles, and so on. Maybe ChatGPT “thought” my style included the more informational style of the things I link to.
Maybe I needed to give more details in my prompt. So I tried again: “write a blog in the style of my own writing in my blog, www.mavismoon.com, about "embodied coregulation," musing about the use of such big, scientific words describing such a human, lovely action as hugging.” Now it produced another letter-styled response titled “"Embracing the Science of Connection: Musings on 'Embodied Coregulation.'" It still seemed quite encyclopedic-sounding to me, not what I had in mind. But I liked several of the sentences. You can see both promptings and responses using the link below, and also a link to the actual blog I wrote with no GPT assistance.
I noticed the free plan is GPT 3.5 and if I paid $20/month I could upgrade to GPT 4.0. From what I learned, the experts said that GPT improved significantly with each version. Maybe the responses would be closer to what I expected in the newer version, but I’m not ready to pay for the service—at least not yet.
I thought the whole experience was interesting and fun. I have all kinds of concerns and questions about A.I. that I won’t even begin to get into here, but I figured I’d share my first experience. What about you? Have you tried A.I.? Are you using it regularly already? What are your thoughts?
New Technologies Always Scare Us. Is A.I. Any Different?