Google’s and OpenAI’s mystery new product, slated to be revealed just a day before Google’s I/O conference, is causing speculation. The latest guesses? A multimodal AI assistant.
OpenAI is determined to steal the limelight by hosting a livestream and unveiling new products, promising “demonstrations of magical ChatGPT and GPT-4 updates.”
So, what’s this mysterious new product? Speculations of GPT-5 and a revamped search engine have been personally debunked by OpenAI CEO Altman.
According to recent reports, the upcoming product from OpenAI might be an AI assistant integrated into smartphones.
According to Latest reports that OpenAI plans to introduce a multimodal AI model. This model would possess both visual and auditory capabilities, able to engage in conversations, identify objects, and boast superior logical reasoning compared to current chatbots. OpenAI has already showcased this model to select clients.
OpenAI has developed models capable of transcribing audio and converting text to speech. The new model, reportedly, is a fusion of these models but with greater accuracy and faster response times. It can aid AI assistants in discerning tones, better understanding semantics, and, “in theory,” assist students in learning mathematics or translating real-world gestures.
However, despite the potential of surpassing GPT-4 Turbo in “answering certain types of questions,” there are still perceptual challenges.
Developer Ananay Arora suggests that OpenAI might roll out a service featuring built-in ChatGPT functionality for making phone calls on smartphones. Arora posted screenshots of the code related to such calls on social media, also discovering evidence of servers configured for real-time audio and video communication by OpenAI.
Using AI to make calls can save users waiting time, potentially positioning this service as one of the functionalities of AI assistants.
AI assistants are also a feature Google has been developing. Reports suggest that the Google Pixel 9 series will feature a new exclusive AI assistant called “Pixie,” capable of using the device’s camera to identify items and perform tasks like instructing purchases or providing item instructions.
Altman previously revealed in an interview with Salesforce CEO Marc Benioff that his favorite AI movie is “Her“, a story about a man falling in love with his AI virtual assistant, noting the incredible prescience of language interface conversations.
The Information reports that Altman envisions developing a virtual assistant akin to that in the movie, capable of rapid responses, to support existing voice assistants like Apple’s Siri with such technology.
Notably, according to insiders, Apple is reportedly set to strike a deal with OpenAI to incorporate its technology into the next generation iOS operating system. The terms of an agreement have been in the works, aiming to integrate ChatGPT functionality into Apple’s upcoming iPhone operating system, iOS 18.
The new model relies on cloud computing and could potentially be included in a free version of ChatGPT in the future.
OpenAI believes that AI assistants with visual and auditory capabilities could bring about transformative changes akin to smartphones. They could observe users’ environmental information, offer suggestions, and serve various purposes like acting as home tutors, translating traffic signs, or assisting in car repairs.
However, the current hardware requirements for such technology are too high to run on personal devices. Media analyses indicate that the new model relies on cloud computing and requires internet connectivity to function. Shrinking complex AI conversations with visual and auditory capabilities enough to run on personal devices like smartphones may take several months or even years.
It’s currently unclear when OpenAI will provide these new features to paying customers, but individuals who have tested the voice assistant reveal that OpenAI’s ultimate plan is to incorporate these features into a free version of ChatGPT, aiming for lower operating costs compared to its most advanced model, GPT-4 Turbo.
OpenAI has not responded to these speculations.
The revelation of OpenAI’s ultimate product will be unveiled next week. OpenAI has announced a livestream on its official website on May 13 at 10 a.m. Pacific Time, where they will demonstrate some ChatGPT and GPT-4 updates.
