He Said AI? Sharpen The Pitchforks!
The reactions proved my point better than the post did.
The reactions proved my point better than the post did.
Bloom proved in 1984 that one-on-one tutoring beats everything else in education, by miles. We just couldn't afford it. AI finally changes the economics, but only if teachers are trusted to use it.
LAUSD's AI policy is not quite as bad as NYC's in one critical area.
New York City is calling its new AI policy a balance, but for the 600,000 students in Pre-K through 8th grade, it's actually a ban, just a smarter one than 2023's.
From personalized instruction to removing the blank page for kids who've never had a way to show what's in their heads, this is where AI actually earns its place in the classroom, and why that matters most for the kids with the least.
Cheating, plagiarism, screen time, equity, privacy: the fears about AI in schools are real, but most of them aren't new. AI didn't break education, it just shone a spotlight on problems that were already there.
A room full of educators went quiet when I suggested assignments shouldn't have a word count. That reaction is the real subject of this post: the "well, this is how we've always done it" mindset that's slowing AI adoption in the classroom, and why it's worth pushing past, carefully.
A while back I wrote a post titled "Should We Still Teach Kids To Code?". The answer was yes. A lot has happened since then, and "vibe coding" is now all the rage. So, does that change my recommendation?
Adobe's Mala Sharma has posted thoughts in AI in education and the White House’s Pledge to America’s Youth.
I hear this question almost daily now: "With AI here, should we teach kids to code?" I was reminded of the following from our book, Captain Code , that teaches coding to kids. The intro asks kids "Why learn to code?".