OpenAI, is gearing up for its "Spring Updates" event on Monday, where it is expected to showcase a groundbreaking new AI model that combines state-of-the-art capabilities in speech recognition, text-to-speech, and image analysis.
While CEO Sam Altman and President Greg Brockman were quick to dismiss rumors and clarify that OpenAI will not be releasing GPT-5 or a search engine at the event, the company is set to demonstrate significant advancements in their AI technology. The Information is reporting that OpenAI has developed an AI voice assistant that could rival industry giants like Google Assistant and Apple's Siri.
What's special about this new AI model is its ability to integrate and build upon OpenAI's existing cutting-edge models. The company has already developed Whisper for speech recognition, Voice Engine for text-to-speech and GPT-4V(ision) for image analysis. The new model will combine these capabilities into a single, more powerful system that outperforms its predecessors.
Sources familiar with the technology reveal that OpenAI has demonstrated its new AI to some customers, showcasing its improved logical reasoning abilities compared to the company's current offerings. By understanding intonation and sarcasm, the new model can provide more accurate and contextually appropriate responses. It could potentially act as a tutor, helping students with papers or math problems, or even provide real-time information about a user's surroundings, such as translating signs or offering car repair advice.
Although the new tech is currently too resource-intensive to run on personal devices, cloud-based versions could be utilized by customers to enhance features already powered by OpenAI's software. The audio capabilities of the integrated model could help customer service agents better understand callers' intonation or detect sarcasm, leading to more accurate and efficient interactions.
OpenAI has been working on GPT-5, which Altman says will be significantly "smarter" than its predecessor, GPT-4, which was released more than a year ago. The company hopes to complete and publicly release GPT-5 by the end of the year, according to a person who has discussed the project with OpenAI leaders.
In addition to its work on AI voice assistants and GPT-5, OpenAI is also exploring new pricing models to better serve its customers. The company plans to offer discounts of up to 50% for customers who prepay to reserve tokens, which could help OpenAI compete with rival model developers and AI server resellers focused on cost efficiency.
Ultimately, this integrated AI model represents another important step forward in the company's mission to develop artificial general intelligence. So, go ahead and mark your calendar for Monday, May 13th at 10AM PT. You won't want to miss it!