1/ Summary of the OpenAI livestream (tl;dr - facts and figures)
A great format from Sam and Jakub, in which we received a lot of information – and every time I notice what a great person and scientist Jakub is: pleasant manner, on point, chill. Please more in this format
@OpenAI
Here are the most important facts:
- OpenAI is working on a personal AGI (the AI device next year?)
- The company believes that Deep Learning could bring humanity to superintelligence in less than ten years.
- September 2026 → Automated AI research interns.
- March 2028 → Fully automated AI researchers.
- 2030s → The impact on the world is impossible to predict.
- “The future will probably be very bright.”
- Extremely optimistic outlook on this timeline.
- Currently, there’s around $1.4 trillion invested in data centers, with plans to build 1 GW factories per week — so the $7 trillion goal no longer seems far-fetched.
- The internal models aren’t vastly superior to the released ones yet, but OpenAI is highly optimistic about a major leap by September 2026 (Deep Learning seems to be scaling well).
- Additionally, major progress is expected over the coming months and years.
- The cost of the models has fallen by a factor of 40 on average; the trend is expected to continue. (40x / year)
- AGI: "When will AGI happen?" Jakub/Sama: It will be a process of years in which we are in. However, Sama points to the year 2028, when science will be completely AI automated.
- "The internal models give us great hope and there is a very realistic possibility that we will see a huge leap in the quality of the models by September 2026"
- "AI system will be capable of pushing scientific forward; that will be the legacy of AI"
- Automation of human labor will increase significantly in the coming years; Sam and Jakub did not (!) deny the possibility of “massive job losses.”
- GPT-6: No date, but they want to update the models more often.
- Big leaps in modeling skills in less than 6 months
- More important information: