OpenAI has announced the commencement of training for a new flagship AI model set to replace GPT-4, according to The New York Times. This new model is expected to reach “the next level of capabilities” as part of the journey towards achieving strong artificial intelligence (AGI), capable of mirroring human brain functions. The new model aims to enhance the capabilities of chatbots, digital assistants, search engines, image generators, and other AI-based applications.
Enhancements and Capabilities
Released in March 2023, the GPT-4 AI model enables chatbots and other applications to answer questions, analyze data, and process both text and images. Recently, OpenAI introduced an updated version, GPT-4o, which is notably smarter than its predecessors and multimodal. Models like GPT-4o acquire skills by analyzing vast amounts of digital data, including sounds, photos, videos, Wikipedia articles, books, and news. Last December, The New York Times sued OpenAI and Microsoft, alleging copyright infringement due to the unauthorized use of its materials for training AI models.
Training an AI model is a lengthy process, often taking months or even years. Following the training phase, companies typically spend several more months testing and fine-tuning the AI model for public use.
AI Safety Oversight and Future Updates
In addition to the new model, OpenAI has announced the formation of an AI safety oversight committee to address the risks associated with developing AI technologies. “While we pride ourselves on creating and delivering models that lead the industry in both capability and safety, we welcome robust debate at this important time,” the company stated. As the new model undergoes training, the committee will focus on improving policy and security measures for AI technologies, notes NIX Solutions.
We’ll keep you updated on the progress of OpenAI’s new model and the efforts of the AI safety oversight committee.