Learning Library

← Back to Library

The 12 Days of OpenAI

Key Points

  • The speaker outlines OpenAI’s “12 days” of releases, from the debut of GPT‑4o and reinforcement fine‑tuning to Sora, Canvas, Apple‑integrated AI, advanced voice/video, Projects, ChatGPT Search, developer tools, seamless app integrations, and the landmark GPT‑4o‑mini (referred to as “03”).
  • He criticizes the premature, unpolished rollout of GPT‑4o‑mini, arguing that releasing something approaching artificial general intelligence without a consumer‑ready experience is a misstep.
  • The speaker likens OpenAI’s release strategy to historic Bell Labs—focus on rapid innovation over polished products—contrasting it with Apple’s consumer‑first approach and suggesting that OpenAI may not be the one to monetize the technology.
  • He notes that while many expect a cohesive, market‑ready product from OpenAI, the company’s track record suggests other firms will likely assemble the building blocks into successful consumer applications, and invites listeners to read his longer analysis on Substack.

Full Transcript

# The 12 Days of OpenAI **Source:** [https://www.youtube.com/watch?v=VS0BsDw9kaY](https://www.youtube.com/watch?v=VS0BsDw9kaY) **Duration:** 00:03:01 ## Summary - The speaker outlines OpenAI’s “12 days” of releases, from the debut of GPT‑4o and reinforcement fine‑tuning to Sora, Canvas, Apple‑integrated AI, advanced voice/video, Projects, ChatGPT Search, developer tools, seamless app integrations, and the landmark GPT‑4o‑mini (referred to as “03”). - He criticizes the premature, unpolished rollout of GPT‑4o‑mini, arguing that releasing something approaching artificial general intelligence without a consumer‑ready experience is a misstep. - The speaker likens OpenAI’s release strategy to historic Bell Labs—focus on rapid innovation over polished products—contrasting it with Apple’s consumer‑first approach and suggesting that OpenAI may not be the one to monetize the technology. - He notes that while many expect a cohesive, market‑ready product from OpenAI, the company’s track record suggests other firms will likely assemble the building blocks into successful consumer applications, and invites listeners to read his longer analysis on Substack. ## Sections - [00:00:00](https://www.youtube.com/watch?v=VS0BsDw9kaY&t=0s) **The 12 Days of OpenAI** - The speaker rapidly outlines twelve recent OpenAI releases—ranging from 01 Pro and reinforcement fine‑tuning to Sora, Canvas, Apple integration, advanced voice/video, Projects, ChatGPT Search, dev tools, seamless app links, and the groundbreaking GPT‑4o (03)—and concludes by warning that unleashing such a leap toward artificial general intelligence is irresponsible. ## Full Transcript
0:00YouTube is going to hate me for this I'm 0:01just going to give you the 12 days of 0:03open Ai and that's it and if you one 0:05take at the end and we're going to be 0:07done number one was 01 and 01 Pro big 0:12deal first release of reasoning models 0:14number two reinforcement fine-tuning 0:16really aimed at sort of 0:18Corporations number three was Sora which 0:22was a texted video model aimed at the 0:24consumer that had been promised for 0:25months number four was canvas which had 0:27been in beta for a while number five was 0:29Apple intelligence so they sort of 0:31rapchat GPT into the iPhone that was 0:33when they had the massive outage I think 0:35the two are related uh then we had 0:37advanced voice with video which Google 0:40sort of stole their Thunder on because 0:42they did the same thing with Gemini then 0:44we had projects which is what Claude 0:47already has so you can organize your 0:49files and stuff then we had chat GPT 0:51search which again some of us already 0:53had in beta then we had Dev tools for 01 0:56which is cool which is 0:58necessary 11th we had Seamless app 1:02Integrations people cannot make up their 1:04minds what that one is like people are 1:06just giving up linking to that when 1:08they're building their listicles because 1:09it's like important but like the details 1:12are very not not there um and then last 1:17but not least 03 which is a freaking 1:19huge deal and is an entirely new level 1:22of 1:24intelligence here's my comment at the 1:26end that is no way to release 1:30artificial general intelligence you do 1:33not do that and then expect to be able 1:38to get the Public's 1:40attention and I think that the reason 1:43why I'm frustrated many of the rest of 1:44us are frustrated is that we're used to 1:46the apple style of releases where it's a 1:48consumer facing release and it's 1:50polished and it's focused on what the 1:52consumer needs that's not how this works 1:56this goes back to an earlier tradition 1:58of American innovation in b s Bell Labs 2:01just built stuff and they released it 2:03and they kind of didn't care what order 2:04it was in and they didn't care if it was 2:05polished and they didn't care if it was 2:06ready for the consumer and like 03 2:09wasn't ready like it's not actually out 2:11they're asking for safety researchers 2:12Sam Alman is Shilling it for safety 2:14researchers actively on 2:16Twitter 2:17so I think we should think about open AI 2:21more as Bell labs and less as Apple 2:23because that actually explains a ton of 2:25how they're handling 2:26releases and it implies that you should 2:29not necessarily assume that the company 2:32that delivers the Innovation is the 2:33company that monetizes it the company 2:35that figures out the massive hit 2:37consumer application I see people all 2:39the time assuming that open ey has a 2:41grand strategy and they're going to put 2:43it all together into one seamless 2:44product but I don't see a track record 2:46of that somebody else may take their 2:47building blocks and do that we will see 2:52that's my comment those are the 12 days 2:53of open AI is quite a hodg Podge if you 2:56want a longer discussion of it it's on 2:58my substack cheers