o3-mini is now available: https://t.co/CkyJ9RCrXk
I’m using the o3-mini-high version, and it definitely feels faster than o1. The output seems similar (a promising sign given this is just the mini version)
Looking forward to the full #o3 release—I expect a much bigger leap
ChatGPT o1 vs. DeepSeek R1: Interactive Tools. I gave both AIs the same prompt. o1 repeatedly failed whereas DeepSeek took longer but worked. This isn’t an isolated example.
A good reminder to continue mapping the jagged frontier, which is moving over time.
#AI#DeepSeek
I've been preparing examples to show my students where AI can struggle… and I came across this one.
I asked ChatGPT4o to generate an image of a glass COMPLETELY full of water. Sounds simple, right? Turns out, it’s surprisingly hard.
#AI#TeachingMoments#AIChallenges
The new OpenAI Operator lets AI control tools we use daily. Still early and buggy, but a glimpse into a future where AI handles email, posts, and more. I tested it, and it even published a LinkedIn post for me (later taken down). Exciting yet unsettling. #AI#Operator#OpenAI
The 'Stargate' AI Initiative seems to make two things clear:
(1) OpenAI is committed to be being truly terrible at naming things
(2) The investment is massive bet on AI as the future of the American economy.
#stargate, #AI, #LLMs
https://t.co/N0GjHlmY1L
Going by the Book: Valuation Ratios and Stock Returns
https://t.co/cuH9jwYztH
"our findings suggest some institutions
have been slow to adapt to the declining relevance of book values for selecting value stocks"
via my prof buddies @professorso and @charles_cy_wang
Follow them!
My favorite document for combining the ideas of market (in)efficiency and value investing: Alphanomics by Charles Lee and Eric So. Worthy of study.
https://t.co/m20KRcxOTi