...make you smell less. (although if you are experiencing anosmia, it may be a sign of covid.)(if you are experiencing a murder, it may be a sign of corvid.)(if you are experiencing an increase in hold times, be glad you have someone to hold.)

This will be a shorter newsletter than usual as I've only just finished my book manuscript revision draft and now I've got a best man speech to write this weekend (hooray!) and I'm a bit mentally tapped out for writing anything clever tonight. Luckily, I wrote my column last weekend, so you can enjoy this:

* AI Roundtable

You just can't avoid AI these days, no matter how hard you try to turn those settings off. And that's not just a dumb accident. Well, it may be dumb. But it's certainly no accident! It's all thanks to hard work by the CEOs of the top AI companies, who we've invited to discuss the latest advancements in AI.

Sunny PitchAI, Google CEO: "When we first launched AI mode, we saw some people complaining that AI was telling people to put glue on pizza or eat lots of small rocks. First off, those aren't necessarily bad answers."

Melon Husk, XAI CEO: "Yeah, there are definitely people who I want to trick into eating rocks and glue."

PitchAI: "Exactly. The great thing about AI is that it has no context for information it finds and just presents it as fact. That's only a problem if it's finding information that's not true. Today, Google AI Overview pulls a lot of information from YouTube, and because we own YouTube, we know it's reliable. If a random person who believes nonsense posts on a forum, they could be lying. But a random person who cares enough to make a video means you can trust whatever they have to say."

Sham Faultman, OpenAI CEO: "And people really do trust whatever AI tells them! Nearly 80% of AI users will follow ChatGPT's advice when it gives them the wrong answer."

PitchAI: "And do you find that at all worrisome?"

Faultman: "Yeah, it means we've still got 20% to go! Luckily we know exactly how to get people to trust AI more, which is to have it flatter you and tell you how smart you are. And sure, sometimes that means AI will endorse your terrible ideas like suicide. But most users who ask ChatGPT for advice aren't driven to kill themselves, they either get a milder form of AI psychosis where they start to believe all their delusions are true, or they simply become less kind to others and less willing to admit when they are wrong."

AI Psychosis is a very real thing, and only one of many ways AI continues to make our world worse. And I'll tell you one more! But first, last week's Banner column (Just imagine last week's Banner column walking backwards into a room, butt first):
March Dadness Tournament

THE AI GAME

Started reading a book about scoring systems (The Score, C Thi Nguyen) which interestingly dovetailed with an article Debbie was reading about why AI is poisoning our brains. The very short version is that metrics inherently compress complex information into universally comprehensible data nuggets, thereby losing lots of important nuance and less-easily-quantifiable information. And metrics shape motivations tremendously, which is a problem especially because measurements that become goals fail as measurements, which Debbie rightly suggested any teacher would recognize as teaching to the test. AI is getting people to the goal metrics faster by short-circuiting the purpose behind the goals (and value in the journey), but even without AI we do too much of that.

It reminded me of some of my reflections on games which I was going to put into a book and then didn't. Writing a book turns out to be a taxing process, but also ideas for a book ain't worth much because I'm far from the only person with a great idea for a book I didn't write. (Lots of people have great ideas for books I didn't write.)(No but seriously, everyone including authors and non-authors alike have lots of great ideas for books they don't write.)(And yes, if you just tell AI to write it you still haven't written a book, but now you've also precluded writing the book in the future because you turned your ideas into slop. Better to write a short article than have AI poop out a book of poop.)

ONE BAD THING ONE GOOD THING

I would certainly not recommend you waste time watching Trump's infuriating hour-long Easter speech (which the White House apparently has subsequently pulled down from their page) but this quote of his stood out to me as a distillation of the man: "When somebody's nice to me, I love that person. Even if they're bad people, I couldn't care less. I'll fight to the end for them." Truly a moral paragon and master of international relations.

On the bright side, pad thai is delicious! Still and again, always already! Debbie made homemade pad thai and I ended up sauteeing some bok choy just so there was a little veg in with the leftovers, and gosh, that's really tasty like.

BUT I AM LE TIRED

That's it for this week, which has been trying mainly insofar as it presages future more trying weeks, but I am exhausted just the same. And, as a friend recently shared with me, it's only Tuesday.

As always, I thank you for reading, will be back in two weeks with another column, and hopefully also more energy and my taxes done. Also, I'm realizing this wasn't actually much shorter than most of my newsletters, just less funny. Shorter time-wise for me I guess, but reading it may feel longer for you! Measurements are tricksy beasts.

Inching along,
Seth