Lol, well apparently Atlanta, TX is in fact a real town with a popn of ~5000. Apparently Google Search is aware of this, but still wants to show pictures of Atlanta, GA anyway.

Lol, honestly a tracking pixel was my initial thought here.

Can you create a border around the perimeter and then crop it in another platform?

(Genuinely curious what the use case is for this)

This was from a slack thread with my team - some wildly speculative thoughts on what may be happening here.

Fast Follow: Tried the following with Gemini -

"Create an image of a map of the United States of America with all state outlines. Put a red dot on the coordinates of 33.7501° N, 84.3885° W and label the dot "Atlanta, GA"

That isn't quite Atlanta, but it's incredibly close. Very interesting!

Would love to wargame the influential touch interactions from your feed that pushed targeting criteria to include Santa farting bubbles.

Is that B... Backlit? 👀

It has come to my attention, there are already 3. 4 is the optimal number. Apologies for the confusion.

Asking for a friend.... How many Simon's does this list have room for as I think 3 sounds like an excellent number if you were in the market for another simon

@adtechgod.com to the rescue?

@aripap.bsky.social is an excellent source of knowledge here.

Amazon ads is a juggernaut and TNF drives big numbers. This is a big rivalry matchup - makes sense they wanna give a few hours of top shelf treatment to their premium ad selling vehicle!

It's not the turing test, but it will be an important milestone once LLMs overcome the full glass of wine problem.

While it's a little abstract, it does represent an important issue - and that is the limited degree of control we really have over LLMs, there is an ever present reversion to the known, thereby limiting utility.

It's not hard for a human to imagine, it doesn't break the laws of physics - yet, it's probably never been part of an LLM training set because who would pour a glass that full?