It’s the "who you know" club all over again. We’ve got directors jumping into "AI Lead" roles who clearly haven't read a technical paper in years. They’re still pushing 2023-era prompting guides like they’re some kind of holy grail, even though that’s basically Stone Age tech for modern models.
Just sat through a meeting where a lead told us to "stay away from the newest models because they’re still learning." Tell me you don't understand how LLM training or inference works without telling me. It’s embarrassing. We’re throwing insane amounts of money away because the people making the calls have zero technical literacy. If you know anything about how this tech actually functions, you’ll see why this is destined to fail.
Note: Ran this through a scrubber to strip out my specific lingo/quotes so I don't get doxxed. You should too.