Have you tried Radiolab for Kids?

Agree!

And really loved the aviation reference!

Me too! One of my current projects is looking at complicating concept of “human-in-the-loop” in our current AI context. It often gets pitched as “add human & stir and we are A-OK!” (See eg most bar guidance) Ok as starting point perhaps but we need to be thinking about this differently/more nuanced.

Finally got a chance to sit down to read this - an excellent post! Much of what I heard post-Mata was that this would be a short lived problem that would burn out quickly once lawyers saw the headlines. As you note, it looks like the opposite is true! Its getting worse and we need to pay attention.

I love this rule of thumb and the examples. It syncs up, imo, with @jackwshepherd.bsky.social great column about how to think about accuracy (e.g. does it matter? how hard or easy is it to verify) jackwshepherd.medium.com/generative-a....

Your Bear updates always make me smile!

Just saw this, which is on point to your question: lsj.com.au/articles/the...

Also worried about all the subtle hallucinations that are probably sneaking through at least to some extent. Fake cases are easy for courts to spot. Less easy to catch swapped out words in quotes or slightly wrongly stated legal tests etc.

I think big challenge here is that tech is newly deceptive in that it looks like a search engine and so people engage with it like that but unlike a search engine it actually isn't looking up things. We are used to tools saying "I don't know/can't do" -- not making up things in very convincing ways.

I do tonnes of lawyer education on AI across Canada & can confirm that the message definitely hasn't gotten through anywhere close to universally. That said, not sure how much more educators & regulators can do to get the word out. Most law schools on it now but will take a while to trickle down/out

Congratulations!! Looking forward to going through it and will be on my recommendation list for lawyers who want to learn more about AI.

‼️