Bluesky Thread

i posted this to linkedin and i’m very worried people are taking it seriously...

February 10, 2025 View original thread

i posted this to linkedin and i’m very worried people are taking it seriously www.linkedin.com/posts/tim-ke...

Alt Text:
A LinkedIn post by Tim Kellogg under the AI & Engineering category titled “AI Cheat Sheet.” The post humorously attempts to clarify AI model naming conventions, listing various models and their definitions:
• o1 – A reasoning model from OpenAI
• o2 – Skipped due to naming conflicts
• o3 – A better reasoning model, not yet released
• o3-mini – A smaller, faster, cheaper reasoning model similar to o1
• 4o – Short for GPT-4o, previously the state-of-the-art (SOTA) multimodal LLM
• o4 – An unreleased reasoning model
• r1 – A reasoning model from DeepSeek, on par with o1
• s1 – A reasoning model that is “not terribly important except that it CHANGES EVERYTHING”
• t1 – A reasoning model that does not exist but is theorized to follow t-1, which also does not exist
• Deep Research – An OpenAI agent that uses reinforcement learning
• Deep Research – A Google agent that does not use reinforcement learning

The post ends with: “Hopefully that clears things up!”

At the bottom, reactions from David Stone and 2 others are visible, along with engagement statistics showing 38 impressions and 1 repost. Interaction buttons for Like, Comment, Repost, and Send are also displayed.

36 2

oh no, someone just did the “insightful” reaction

10

12 hours later

it’s heavily backfiring. people are even commenting how helpful it is

5

More like this

one one hand, i want to be appalled that Musk is slanting political discourse

mental health — i’ve noticed that with agents, i’m shifting my attention betw...

i’m looking at the ChatGPT & Gemini apps, reverse engineering them