Been reading about (a) Elon’s plan to use AI to analyze the “what did you do last week” responses as well as (b) angry redditors vowing to use ChatGPT to generate said responses. At this point will ChatGPT just be training Grok? I wonder how OpenAI will feel about that. - ThreadSky

brianfung.me • 8 days ago

Been reading about (a) Elon’s plan to use AI to analyze the “what did you do last week” responses as well as (b) angry redditors vowing to use ChatGPT to generate said responses.

At this point will ChatGPT just be training Grok? I wonder how OpenAI will feel about that.

Comments

greenhombre.bsky.social•8 days ago

garbage in garbage out

laureanne.bsky.social•8 days ago

Jeezes that’s not the issue with what Musk is doing.

laureanne.bsky.social•8 days ago

Uhm really the whole he’s evaluating employees without any standing to do is the primary issue.

brianfung.me•8 days ago

That’s been pretty well established by this point.

laureanne.bsky.social•8 days ago

Has it? I’m not interested in oligarchs’ monopolistic drama. Only in oligarchal monopoly. And political capture. Sort of the issue of the day, no?

brianfung.me•8 days ago

Yes. A whole bunch of agencies (FBI, DHS, DOD, and more) have given their employees guidance not to respond to Musk’s email (because they don’t report to OPM, they report up their ordinary chain of command). We are several news cycles into this story already.

laureanne.bsky.social•8 days ago

I’m up on the cycle. Btw the Dept of Interior is overriding the OPM guidance.

The REAL story is how unprecedented Musk’s HR abuses are for employees. How damaging they are to people and to the economy… How it does not benefit US.

wherefore the 4th Estate? Journalism dies in click bait. sigh

brianfung.me•8 days ago

The stories are out there. Here are three I found with just a quick Google search.

It’s good to hold media accountable but this has been THE story of the past few weeks, everywhere.

https://www.reuters.com/world/us/fired-rehired-dizzying-confusion-trumps-government-overhaul-2025-02-21/

https://www.npr.org/2025/02/24/nx-s1-5305717/trump-layoffs-federal-workers-chaos

https://news.bloomberglaw.com/daily-labor-report/trumps-mass-firings-of-federal-workers-spread-chaos-nationwide

writearthur.bsky.social•7 days ago

It's a really good question. Though as I understand it, input data does not automatically train models. So unless xAI deliberately puts the OPM emails in a training dataset for the next Grok update, ChatGPT would not be training Grok in this instance.

writearthur.bsky.social•7 days ago

Of course, if the OPM emails end up leaking onto the open Web, that's a different story.

mjwhansen.com•7 days ago

This seems like a real risk overall, that LLMs end up being unwittingly cannibalistic the more AI content is out there.

Tons of AI content generated + LLMs unable to distinguish between human and AI content = LLM “source” data ends up being content LLMs themselves created

Comments

Posting Rules

Reply