A high schooler writes about AI tools in the classroom

LLMs can be amazing [^0] as an assistive technology, but using them as a "do it for me" button is just way too easy, so that's how they are de facto used.

I believe it will take about 5-10 years for us to fully comprehend how damaging unplanned remote classrooms and unchecked LLM use in the classroom was. Like heroin, it will be extremely hard to undo our dependence on them by that point. I'm pretty scared for how our students will fare on the global scale in the coming years.

[^0] I strongly believe that 60% of the value of LLMs can be realized by learning how to use a search engine properly. Probably more. Nonetheless, I've fully embraced my accidentally-acquired curmudgeon identity and know that I'm in the minority about this.

[^1] You won't believe how many people leave their laptops unlocked and their screen's contents visible for everyone to see. Committing identity theft has to be easier than ever these days. This basic infosec principle seems to be something we've lost since the great WFH migration.

From what I can tell, this is mostly a parent-led thing, well supported by overworked teachers who are more than willing to avoid even more work grading out-of-school assignments.

One school has been abdicating homework for more in-classroom practice, as homework adds more grading and scheduling load on the teacher for little overall benefit. The core idea behind this is that motivated students will always practice at home, even if they aren't explicitly asked to. Unmotivated students --- usually the majority in a typical classroom --- won't or will do a poor job of it.

Another school of thought is the "flipped" classroom. This approach doubles-down on homework by having teachers prepare a pre-recorded lesson for students to watch while they're home and using the classroom as a space for practice and retention. This increases the student's accountability for their own learning while decreasing the teacher's workload over time if they are teaching the same material for a long time (very high initially, of course).

Generative AI is new. Pedagogical research involving them is even newer. Teachers are rarely given resources to meaningfully explore new methods. Expecting teachers to stumble through updated processes to enable students to incorporate generative AI is a mess.

Students are also children. They'll take the path of least resistance if it is available to them. Expecting students to meaningfully incorporate generative AI into their learning process rather than just reaching for "ugh this essay is dumb - chatgpt give me an essay on the use of time skipping in To the Lighthouse."

The situation is a total mess.