rllyniceguy.bsky.social
@rllyniceguy.bsky.social
i'm nice
so they're lying? that's your argument? i was having a conversation with someone last night about this exact thing. they are themselves traumatized from working this job.
Facebook inflicted ‘lifelong trauma’ on content moderators in Kenya, campaigners say, as more than 140 are diagnosed with PTSD | CNN Business
Campaigners have accused Facebook parent Meta of inflicting “potentially lifelong trauma” on hundreds of content moderators in Kenya, after more than 140 were diagnosed with PTSD and other mental heal...
www.cnn.com
July 5, 2025 at 8:15 PM
gonna open a polymarket position on it entering the DSM before 2040
June 22, 2025 at 10:20 PM
the tailscale specific upside is that i think it'll also allow anyone with the right level of ts network access to connect the mcp server to whatever chat session/provider they prefer
June 21, 2025 at 3:00 PM
saving a ts oauth app to my own db and then using those creds to let anyone with access to the network the app is associated with exchange tokens, which claude is remembering and then using to authenticate across sessions. nice because claude doesn't see any oauth creds, just gets a token
June 21, 2025 at 2:59 PM
built an mcp server that works in conjunction with another server i'm running that handles the tailscale oauth flow and allows me to expose the mcp server for claude or other public clients securely! now that claude is offering custom remote integrations it can call my custom tools in the web app
June 20, 2025 at 8:12 PM
mental load of reviewing something generated for you rather than writing from scratch is vastly less. that's the case for me and writing unit tests, as an engineer, and I'd imagine it holds true for email writing. it's really not error-filled crap anymore, unless the topic is esoteric.
June 18, 2025 at 9:12 PM
i think a lot of people use it for work. i've observed this in my own life. when you have to send 1000 stupid emails a day it's a relief to not have compose them by hand yourself.
June 18, 2025 at 7:52 PM
so companies are forcing use of something that meets your criteria of AI but chat gpt doesnt qualify?
June 18, 2025 at 5:43 PM
chatgpt dot com is the sixth most popular website in the world, and internal pushes for AI adoption wouldn't account for that fact. it's popular. we need to be able to agree on the basics to discuss this stuff.
June 18, 2025 at 5:20 PM
they're bad at performing calculations. that's not the same thing as a tutor for math.
June 18, 2025 at 5:17 PM
i think MCP is going to turn out to be the infrastructure evolution that triggered mass adoption and transformation. it's not rlly about llm capabilities anymore, just infrastructure
June 11, 2025 at 3:40 PM
they were also mostly laying on the floor in the last couple minutes
May 22, 2025 at 9:07 PM
eh. the infrastructure, including libraries for better interfaces, will grow and improve over time.
May 11, 2025 at 11:31 PM
consider you are not always the subject of discussion
May 7, 2025 at 7:48 PM
infrastructure to augment llm's with information retrieval systems is starting to emerge, and it's pretty effective. increasingly i think the technical (not ethical) issues with GenAI are ecosystem problems more than anything
April 28, 2025 at 11:43 PM
right because thousands of grant funded academics developed NFTs over the course of decades
April 26, 2025 at 5:53 PM
chat gpt is much better at finding the needle in a thousand lines of logging than i am
April 24, 2025 at 3:33 AM
oh my god yes. repeating a process of any kind that you haven't done in a while and don't have too many romantic feelings about is immensely easier with these tools.
April 24, 2025 at 3:31 AM