ChatGPT-5 launch: Altman felt ‘useless' testing AI, compares its impact to the Manhattan project
OpenAI CEO Sam Altman took to X to share that the livestream will run for close to an hour, longer than usual. 'We have a lot to show and hope you can find the time to watch!' he wrote.
ChatGPT-5: The 'here it is!' moment
Speaking on a recent episode of 'This Past Weekend w/ Theo Von', Altman shared a personal epiphany while testing ChatGPT-5. He described an experience where the new model tackled something he himself could not grasp, which left him feeling "useless". 'I really sat back in my chair and I was just like 'Oh man! Here it is!' he reminisced. 'I felt useless relative to the AI…I felt like I should have been able to do it, and I couldn't, and it was hard. But AI did it just like that. It was a weird feeling.'When asked about the leap in intelligence from GPT-4 to GPT-5 during a Q&A at Stanford University, Altman said that GPT-4 is not 'phenomenal' and that it is 'mildly embarrassing at best.''GPT-4 is the dumbest model any of you will ever have to use, by a lot,' he said.
Also Read: ChatGPT may face capacity crunches ahead of GPT-5 launch: Sam Altman
On AI development
In his conversation with Von, Altman drew a comparison between the development of AI and the Trinity test, which was part of the Manhattan Project, the World War II-era programme that produced the first atomic bomb. He said that there have been moments in the history of science where scientists have looked at their creation and said 'What have we done?' He was referencing the reaction of those who witnessed the sheer scale of the atomic bomb's power.According to Altman, the people working on AI today feel something very similar.'We just don't know. We think it's going to be great. There are real risks. In truth, all we know right now is that we have discovered or invented something extraordinary that is going to reshape the course of human history,' he added.
Also Read: OpenAI in talks to sell employee shares at $500 billion valuation: Bloomberg
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Hindustan Times
14 minutes ago
- Hindustan Times
Haryana, Israel will collaborate on agriculture, AI: CM Saini
Chandigarh, Haryana Chief Minister Nayab Singh Saini said on Thursday that the state will collaborate with Israel in various sectors such as research, healthcare, agricultural technology, advanced irrigation systems, artificial intelligence and wastewater management. Haryana, Israel will collaborate on agriculture, AI: CM Saini According to an official statement, Israeli Ambassador to India, Reuven Azar, paid a courtesy visit and met the chief minister at the latter's official residence here. The two discussed cooperation and several other essential subjects, the statement said. Saini stated that there was a focus on establishing a Centre of Excellence in Haryana. Discussions on further development of the Integrated Aviation Hub in Hisar and enhancing overseas placement opportunities were also held. The state government has set up the Department of Foreign Cooperation to explore overseas employment opportunities and strengthen ties with other countries. The department is consistently working to provide employment to Haryana's youth abroad and to double the state's exports, Saini said. Over 180 youths from the state are employed in Israel through overseas placement initiatives. Additionally, there has been a demand from across the country to recruit 5,000 nurses in Israel's healthcare sector, the CM said, emphasising Haryana is keen to increase its participation in this initiative. Azar and Saini also talked about setting up a global artificial intelligence centre in the state to train youth in modern AI skills and promote technological innovation. The state is also planning on working with Israel on new technologies to reuse wastewater for irrigation and make water suitable for agriculture and drinking purposes, the CM said. During the meeting, he presented a copy of the Bhagavad Gita to the Israeli ambassador. Principal Secretary to the Chief Minister, Arun Kumar Gupta and Commissioner and Secretary of the Department of Foreign Cooperation, Amneet P Kumar, were also present on the occasion. This article was generated from an automated news agency feed without modifications to text.
Mint
14 minutes ago
- Mint
OpenAI ChatGPT-5 Launch Live Updates: Sam Altman teases LIVE5TREAM event; what to expect
OpenAI ChatGPT-5 Launch Live Updates: OpenAI appears set to officially unveil GPT-5, the long-anticipated successor to its GPT-4 model, during a livestream scheduled for Thursday at 10:30 PM IST (10 AM PT). Follow updates here: 07 Aug 2025, 09:25 PM IST OpenAI ChatGPT-5 Launch Live Updates: OpenAI appears set to officially unveil GPT-5, the long-anticipated successor to its GPT-4 model, during a livestream scheduled for Thursday at 10:30 PM IST (10 AM PT). A promotional teaser titled 'LIVE5TREAM' has sparked widespread speculation online, with the stylised '5' in the name widely interpreted as a nod to the model's imminent release.
Mint
14 minutes ago
- Mint
OpenAI GPT-5 reportedly leaked ahead of imminent launch, variants and features tipped
OpenAI's next-generation artificial intelligence model, GPT-5, appears to have been leaked ahead of its official unveiling. According to reports, details about multiple GPT-5 variants were published in a GitHub blog post that has since been deleted. However, screenshots circulated on social media suggest the post briefly revealed information about at least five different model versions, as well as a potential redesign for the ChatGPT interface. The post, first flagged by TestingCatalog, reportedly listed the standard GPT-5 model along with GPT-5 Mini, GPT-5 Nano, GPT-5 Chat, and a Pro version. While OpenAI has yet to confirm the existence of these models, the leak suggests a broader product strategy, possibly aimed at catering to different levels of usage and deployment environments. The GPT-5 Mini and Nano variants are believed to be lighter models, though technical specifications remain unconfirmed. Mini may be optimised for quicker response generation with lower latency, while Nano could be designed for use in edge computing or on-device operations, suitable for localised processing without a cloud dependency. Another notable inclusion in the leak was GPT-5 Chat, reportedly tailored for enterprise-level conversations and GitHub Copilot. Descriptions from the now-deleted post indicate it is intended to deliver more advanced, multimodal, and context-aware interactions. According to information sourced byThe Verge, GPT-5 Chat is designed for use in business applications that require sustained, nuanced dialogue. The leak also referenced a model or feature called GPT-5 Thinking. Though details are sparse, it is speculated this may relate to an API function allowing developers to control the model's 'thinking budget', possibly a way to allocate more computational effort to complex queries. The rollout strategy, as suggested by the leak, will reportedly give users on different ChatGPT plans varying levels of access to GPT-5. Free-tier users are expected to receive limited access to the base GPT-5 model, while those subscribed to ChatGPT Plus may gain features related to advanced reasoning and increased usage limits. Pro-tier subscribers could receive access to a new GPT-5 Pro variant, which a separate leak claims is designed to deliver 'research-grade intelligence.' This particular claim was shared by a user named @scaling01 on X, who reportedly found relevant descriptions within ChatGPT's internal codebase. Additionally, Team plan subscribers are said to be in line for access to both GPT-5 with enhanced reasoning and the GPT-5 Pro model. Alongside the model rollout, a redesign of the ChatGPT user interface is also anticipated. The update is expected to feature a simplified prompt input area, a unified control button for modes and actions, and new background colour themes to enhance usability. While no images of the redesign have surfaced, TestingCatalog reports that the new interface is part of the wider GPT-5 deployment. OpenAI has not issued an official statement confirming the leak or the details therein. However, the information aligns with growing speculation about the imminent release of GPT-5, widely expected to be announced soon.
