Sunday, June 11, 2023

[tsondoku]

The plume has diffused and moved on to the south and west, and the fires are still raging in Canada. But here, the smoke had completely cleared by yesterday and given way to beautiful, mild June weather – the kind of “traditional” June weather we haven’t had in several years. It felt good to get outside and take my usual walk, enjoying the rich texture of all the different birdsongs. When I walked through the park, the Merlin app picked up a house sparrow, robin, song sparrow, house finch, crow, warbling vireo, European starling, and chimney swift. All at the same time! Then today, Nick and I took a walk in the Laurelwood Arboretum, which was still bursting with flowers amid the vibrant greenery.

The New York Times has been reporting extensively on the AI developments of the past months, and this week there was an article about the tutor bot created by Khan Academy. It’s already being deployed in schools, including Newark public schools, as a pilot program. The chatbot is capable of explaining concepts and engaging students in Socratic questioning, and based on my own experiments with ChatGPT, I’m sure it does at least a decent job. But I don’t think even a very sophisticated bot can ever do what a human tutor can. It can’t connect with students, build relationships with them, understand their individual needs, and provide the kind of genuine encouragement and support that we can. Of course, cash-strapped schools and parents might very well decide that what the bot does is enough. I worry that AI is coming for my job, and indeed, it’s probably coming for a lot of people’s jobs.  

I don’t think AI is inherently bad, though. I’ve been fascinated with ChatGPT since it was made public, and I’ve been using it myself as a self-improvement tool (not to mention having fun with it, like when Mom, Emmett, and I asked it to generate a mad lib for us). I have trouble with “logistical” conversations, so once it helped me write a script for calling my insurance company over a complicated issue. I’ve also role-played challenging conversations, which helped me prepare and approach them more confidently. Of course, that kind of role-playing would be much better to do with a therapist, but I actually did have a couple of major communication breakthroughs afterward. And I sometimes use an AI-driven app called Goblin Tools, which breaks down tasks into smaller steps and estimates the amount of time a task will take.

I think the technology itself isn’t inherently good or bad. Ultimately, it’s going to be a reflection of us. And ultimately, that’s what I’m afraid of.