OpenAI CEO Sam Altman joined by several other company executives recently hosted an ask-me-anything (AMA) session, offering insights into the company’s operations and vision. The session took place on the popular social platform Reddit. Participants were encouraged to inquire about the AI company’s offerings like ChatGPT, as well as pose any general questions related to artificial intelligence (AI) and artificial general intelligence (AGI). Two line breaks are needed after “Reddit” and “intelligence (AGI). ” During the session, Altman mentioned that GPT-5 is not scheduled for release this year. Nevertheless, the company aims to unveil “some very exciting releases” before the conclusion of 2024.
OpenAI Staff Host AMA on Reddit
The AMA session was hosted on the ChatGPT subreddit. Referring to it as our “Reddit launch”, numerous executives took part in the question-and-answer post, including Altman, OpenAI’s CPO Kevin Weil, SVP of Research Mark Chen, VP Engineering Srinivas Narayanan, and Chief Scientist Jakub Pachocki. The official X (formerly known as Twitter) handle of OpenAI also posted about the Reddit AMA. When asked about the timeline for the release of GPT-5 or a similar model, Altman expressed, “We have some exciting releases planned for later this year! However, there won’t be anything labeled as GPT-5. ” This aligns with the expectations of many, as reports suggest that OpenAI is likely to introduce their next flagship model around 2025.
Another user inquired about the benefits of SearchGPT or the ChatGPT Search feature. Altman expressed that he sees it as a quicker and more convenient method for accessing information. He also emphasized that the web search feature will be particularly beneficial for advanced research.
“I’m excited about the prospect of a future where a search query can generate a personalized web page on-the-fly,” he shared. Weil answered a variety of questions from users. One inquired about the delay in Sora. The Chief Product Officer at OpenAI explained that the delay was due to the extra time taken to refine the model, ensure safety and prevent impersonation, as well as the necessity to scale compute. He declined to disclose a specific launch date. Weil also emphasized that the ‘o’ series AI models, like GPT-4o and o1-preview, will remain a key feature in the company’s lineup and will continue to be used even once GPT-5 is launched.
Moreover, he also disclosed that the ChatGPT Advanced Voice Mode could be customized to incorporate a singing voice into the AI. One user had a suggestion: “Can we please have a hands-free method to end a voice conversation with ChatGPT? Currently, I use the action button shortcut on iPhone, but I have to manually dismiss the voice chat every time. ” Weil showed enthusiasm for the idea, commenting, “I really love this concept. ” Sharing with the team now!”
OpenAI Senior Vice President of Research, Mike Chen, also addressed an important user question regarding AI hallucination. He described the persistence of hallucinations in AI models as a fundamentally challenging issue. The reason for this is that AI models are trained using human-written text, and since humans can sometimes make mistakes, these errors get incorporated into the primary dataset of large language models (LLMs). “Our models are advancing in their ability to cite sources, enhancing the credibility of their answers. Additionally, we have confidence in the reinforcement learning (RL) techniques we employ.
