GradientDescent

@shredder7579
0 Followers
38 Following
44 Posts

This is ridiculous. https://news.ycombinator.com/item?id=46755999

New FAA rule that makes it so drone pilots can unknowingly violate airspace around mobile ICE operations.

I want to believe they'd set up some sort of public API that updates every minute or so with the geo coordinates for a no-fly zone (thereby undermining their "surprise" efforts to abduct people), but what's more likely is the armed thugs being emboldened to just start shooting things out of the sky whenever they feel like it.

FAA institutes nationwide drone no-fly zones around ICE operations | Hacker News

The "1 engineer, 1 month, 1 million lines of code" take by Galen Hunt of Microsoft is damaging and represents a total disconnect of where the company really should go.

Their OS products have suffered from endless bloat over the last few generations (both server and desktop).

They should be PAIRING BACK to increase quality. Not doubling down on quantity.

It also seems to have caused a great divide in the data science world, and more generally in programming. Even worse, the hype around transformer models drives a narrative that "if you don't embrace it, you're going to be left behind".

I just dont want these models using my email as inputs and then returning them to me as search results in an entirely different product (looking at you, Microsoft CoPilot and Sharepoint).

Fellow data scientists out there... How much of your day to day has become fighting the hype of LLMs?

I'm not against LLMs in totality. But I find the responsible uses limited (mostly just summarization where the "answer" is somewhere already in the prompt, avoiding the chance the LLM generates bad outputs).

The number of contexts where non-practitioners are applying this opaque tech seems to increase every day.