OpenAI team member teasing about upcoming launches

Leopold Aschenbrenner, a researcher at OpenAI focusing on Superalignment, recently posted a tweet regarding the one-year anniversary of GPT-4’s release. Here is the tweet.

But first, what’s superalignment?

Superalignment is a concept in AI safety and governance that focuses on securing extremely intelligent AI systems (often referred to as superintelligence) to act according to human values and goals. Knock knock, Gemini.

Aschenbrenner’s statement suggests that the past year may have seen a slower pace of advancement in the field of AI compared to what might be expected in the near future. Aschenbrenner’s connection to the Superalignment team could suggest a potential link to AI advancements.

Is he talking about GPT-5? Likely so, but we can’t be 100% sure until we hear about it. But GPT-5 has been in commotion from time to time. Recently, Microsoft mistyped ‘GPT-5 Turbo’ instead of GPT-V Turbo on the Copilot Pro subscription page, which, at least, got me excited. Perhaps, a beautiful mistake?

But not just this, earlier OpenAI CEO Sam Altman said that GPT-5 is going to be “better at everything,” and Micorosft’s Mikhail Parakhin claimed that GPT-5 could be capable of deciphering a 3000 years old language that has never been deciphered.