OpenAI CEO Sam Altman and several other company executives hosted an ask-me-anything (AMA) session on Thursday. The session was held on social networking platform Reddit and asked users to ask general questions about the AI firm’s products such as ChatGPT or artificial intelligence (AI) and artificial general intelligence (AGI). During the session, Altman said that GPT-5 will not be released this year, however, the company plans to introduce “some very nice releases” before the end of 2024.
OpenAI staff hosts AMA on Reddit
The AMA session was hosted on the ChatGPT subreddit. Calling it “our Reddit launch”, a number of executives participated in the question-and-answer post, including Altman, OpenAI CPO Kevin Weil, SVP of Research Mark Chen, VP Engineering Srinivas Narayanan, and Chief Scientist Jakub Pachocki. OpenAI’s official
Responding to a question about the timeline for the release of GPT-5 or its equivalent, Altman said, “We have some really cool releases coming later this year! However, it’s not something we’re going to call gpt-5.” This appears to have been confirmed by multiple reports that OpenAI will release the next flagship model sometime in 2025.
Another user asked about the importance of the SearchGPT or ChatGPT search feature, Altman said he found it to be a faster and easier way to find information. He also highlighted that web search functionality would be more useful for complex research. He added, “I also hope for a future where a custom web page can be dynamically rendered in response to a search query.”
Weil also asked users a number of questions. One asked about the delay in Sora, to which the OpenAI CPO said the delay was due to the extra time it took to get the models right, get security and impersonation right, and need to scale the compute. However, he did not reveal its launch date.
Weil also highlighted that the ‘O’ series AI models, such as the GPT-4O and O1-Preview, will become a mainstay in the company’s lineup and will be visible even after the release of the GPT-5. Additionally, he also revealed that the ChatGPIT Advanced Voice mode can be modified to add singing voices to the AI.
One user suggested, “Can we get a hands-free way to end voice conversations with ChatGPT? Using action button shortcuts on iPhone, but having to manually dismiss voice chat every time. Weil seemed impressed by the idea and said, “I like the idea. Sharing with the team now!”
OpenAI SVP or Research Mike Chen also answered an important user question about AI hallucinations. Explaining why hallucinations have not been completely eliminated from AI models, he described it as a fundamentally difficult problem. This is because AI models learn from human-written text, and humans can often make errors, which are then added to the original dataset of a large language model (LLM).
“Our models are improving at quoting, which bases their answers on reliable sources, and we also believe that RL will also help with hallucinations – when we can programmatically check whether models hallucinate. If so, we can reward him for not doing so,” Chen said. ,
