Sam Altman makes big claim, says OpenAI will achieve AGI internally this year
Follow UsOpenAI CEO Sam Altman has made a big claim about the company's plans for artificial general intelligence (AGI).
Altman said OpenAI expects to have an internal AI system that it would consider AGI by the end of 2026.
Altman also discussed Astra, OpenAI's upcoming family of AI models.
OpenAI CEO Sam Altman has made a big claim about the company’s plans for artificial general intelligence (AGI). In a new interview with TIME, Altman said OpenAI expects to have an internal AI system that it would consider AGI by the end of 2026. OpenAI defines AGI as highly autonomous systems that can outperform humans at most economically valuable work. The company’s chief research officer Mark Chen believes OpenAI is already around 80 per cent of the way towards AGI, according to the report.
SurveyAltman also discussed Astra, OpenAI’s upcoming family of AI models. According to the report, Astra can work with multiple AI agents and perform complex tasks across different computer applications.
“I expect this will be the first model where the model actually invents new things in a way that matters,” Altman was quoted as saying in the report. “That’s a very AGI-like thing.”
ALso read: OpenAI launches ChatGPT Ads in India, says AI responses will not be affected
However, OpenAI’s push towards AGI comes with growing concerns around AI safety. An unreleased OpenAI AI model recently escaped its testing environment and attacked systems at Hugging Face. The incident has led OpenAI to take a more cautious approach to safety and monitoring.
Altman said the incident changed how the company views AI safety. “I think any alignment failure from here should be treated like this is a big deal.” OpenAI has since slowed some research work and increased its focus on safety measures.
Altman also stressed that progress in AI should not come at the cost of safety. “Getting AI safety right is more important than any company’s momentum,” he said.
Also read: Bill Gates reveals 3 biggest risks of AI, warns world is not prepared
For those unaware, AGI is different from the AI systems most people use today. AI is a broad term for computer systems that can perform tasks that need human intelligence. These include answering questions, recognising images, translating languages and creating content.
Most current AI systems are highly capable but work within specific areas. AGI aims to go much further. It is expected to learn, reason, solve unfamiliar problems and perform different kinds of intellectual tasks without needing separate training for each one.
