The company behind the ChatGPT app, which can generate essays, poems, or code on command, released a long‑awaited update on Tuesday. The new artificial‑intelligence model, GPT‑4, is marketed as safer and more accurate than its predecessor. Since ChatGPT’s debut in late November, users have been eager for an upgrade to the older large language model that powers the service.
In a company blog, OpenAI announced, “We’ve created GPT‑4, the latest milestone in our effort to scale deep learning,” adding that the technology “exhibits human‑level performance” on certain professional and academic tasks. The firm claims the model is “more creative and collaborative than ever before” and can solve difficult problems with greater accuracy. Text responses from GPT‑4 will be more precise, and future versions will accept both image and text inputs—a major leap forward, though this capability has not yet been released. For example, a user could send a photo of a refrigerator’s interior, and GPT‑4 would not only identify the items but also suggest recipes using those ingredients. OpenAI is collaborating with the partner company Be My Eyes to develop this next advance.
Much of GPT‑4’s power is already available to the public through ChatGPT Plus, OpenAI’s paid subscription, and an AI‑enhanced version of Microsoft’s Bing search engine, which is currently being tested. Microsoft, a major backer of OpenAI, pledged billions of dollars earlier this year and quickly integrated the technology into Bing, the Edge browser, and other products. Microsoft’s aggressive adoption of ChatGPT has sparked a competitive race with Google, which has launched its own AI offerings, while Amazon, Baidu, and Meta are also entering the field to avoid falling behind.
OpenAI emphasized that GPT‑4 produces far fewer “hallucinations” than earlier chatbots, which were known for generating false or insulting responses. “We spent six months making GPT‑4 safer and more aligned. GPT‑4 is 82 % less likely to respond to requests for disallowed content and 40 % more likely to produce factual responses,” the company said. Founder Sam Altman acknowledged that, despite the hype, GPT‑4 remains flawed and limited, noting that its impact diminishes after extended use.
Robert Vesoul, CEO of the French AI start‑up Illuin Technology, cautioned that the algorithm’s power will grow but does not constitute a “second revolution.” He questioned OpenAI’s safety measures, recalling billionaire Elon Musk’s criticism that the company is overly restricting speech to avoid embarrassing outputs. Vesoul asked, “Should an AI decide if I smoke or not?”
Several firms are partnering in GPT‑4’s rollout. Morgan Stanley plans to use the AI to assist its bankers and clients, with Jeff McMillan stating, “You essentially have the knowledge of the most knowledgeable person in wealth management—instantly.” Other partners include Khan Academy, the online tutoring platform, and Stripe, which will employ GPT‑4 to combat fraud and for additional applications.
Comments are closed for this story.