OpenAI may soon introduce a digital assistant with support for multimodal artificial intelligence. According to The Information, the new OpenAI assistant can even recognize sarcasm.
According to sources from The Information, OpenAI has demonstrated a new multimodal AI model to a number of clients, capable of engaging in dialogue with users and recognizing objects.
The new model provides faster and more accurate interpretation of images and audio compared to existing AI models for transcription or text-to-speech conversion. As noted by The Information, it can help customer support staff "better understand the intonation of callers' voices or whether they are being sarcastic."
"In theory," the model could help students solve math problems or translate real-world notation. According to sources, the new AI model may outperform GPT-4 Turbo in "answering some types of questions," but will still be more prone to making erroneous decisions.
According to developer Ananay Arora, OpenAI is also preparing to add a phone call function to the ChatGPT chatbot. Arora claims to have found evidence that OpenAI has already considered using servers for real-time audio and video communication.
It should be noted that this feature will not be available in the GPT-5 model, if it is unveiled next week, as CEO Sam Altman denies that the upcoming event will be related to the successor of the GPT-4 AI model. He also denies rumors that the company will unveil a new AI-based search engine next week.
Follow NEWS.am Tech on Facebook and Twitter