The rumours continue as to what OpenAI is set to reveal on Monday at 10 am PT/1 pm ET at their “Spring Update”. One possibility that is getting some traction is a new AI model that can talk and recognize objects.
This multimodal AI model, was reportedly shown to some of OpenAI’s customers, is rumoured to be more adept at processing images and audio than its predecessors.
While this seems simple in concept, it has huge implications for commercial usage. It could, for instance, offer potential improvements in customer service, not only answering questions but by detecting nuances such as sarcasm in people’s voices.
Visual abilities could assist in everything from aiding students with math by recognizing written formulas to translating signage, pictures and even screen displays.
There is also speculation that OpenAI might also introduce capabilities for making phone calls within ChatGPT. This rumour was fuelled by a developer who shared screenshots suggesting OpenAI has prepared for real-time audio and video communication.
This timing of this big revelation comes just as Google is prepared to launch multi-model AI functionalities with “Pixie,” a Google Assistant that could interact with real-world objects through camera input. It’s unlikely that this timing is an accident as it comes just a day before Google’s big conference Google I/O that opens on May 14th.
While there’s speculation about new AI tools and services, OpenAI’s CEO Sam Altman has clarified that the upcoming announcement does not pertain to GPT-5 or a new AI-powered search engine, keeping open future potential this year that could once again overshadow advancements by rivals like Google.
You can watch this live at OpenAI.com.