OpenAI, led by CEO Sam Altman, has announced a significant update to ChatGPT. This update is a new voice-enabled ChatGPT for the ChatGPT Plus subscribers.
This new feature improves user engagement as it allows for more conversational and engaging exchanges.
OpenAI has started the gradual release of its new voice chatbot based on the GPT-4o model to a limited number of users of the ChatGPT Plus service.
This new mode combines voice, text, and vision to enable real-time communication with emotional intelligence.
The feature will be made accessible to all ChatGPT Plus users by fall 2024.
The voice feature was initially not available to be used due to the issue of safety measures.
Originally scheduled for May, the product was delayed to enhance the model’s safety measures, so that it would be capable of identifying and declining undesirable content.
OpenAI has also increased the capacity to answer to many users in real time.
To avoid safety and ethical issues, OpenAI performed a series of tests with more than 100 external red-teamers from 29 countries who spoke 45 languages.
These testers assessed the model’s capacity to manage a large number of inputs without endangering the user.
Other measures that OpenAI has put in place are; the model has been trained to use four pre-defined voices which were developed by the company in collaboration with voice talents so that it cannot generate any other voices on its own.
Notably, the ‘Sky’ voice which was accused of mimicking Scarlett Johansson has been pulled out due to public and legal pressure.
In an official statement, OpenAI’s Lindsay McCallum stated that this new model will not imitate people or personalities and will prevent any voice not registered on the model from being produced.
The new voice mode is a major improvement over the previous audio features available in ChatGPT.
The old solution had three models; voice to text, text processing, and text to voice, and this made the conversations choppy and unresponsive, while the new GPT-4o has integrated all these into one model making the conversations smoother and more responsive.
The model can also recognize the emotional tone and hence make conversations more richer.
Although the present deployment is on voice only, OpenAI has shown other features like video and screen sharing that are planned to be added in the future.
During the May demo, the workers leveraged the AI to solve mathematical problems and coding questions using visual input suggesting that there is a possibility of expansion in future uses.
Apart from the recent ChatGPT update, OpenAI is also engaged in various other AI advancements.
Recently, the company has become a member of the Coalition for Secure AI (CoSAI) where leading technology companies such as Google, Microsoft, Nvidia, and Amazon also participate.
Launched at the Aspen Security Forum, CoSAI focuses on developing sound security principles and best practices for AI development and use to tackle key risks associated with AI.
In addition, OpenAI has recently launched SearchGPT that is an AI search engine which can give quick and to the point answers with the source.
At the moment, SearchGPT is being tested with a limited set of users and publishers, and the goal is to bring a significant shift in the way people engage with information online through offering relevant and timely answers.
저작권자 ⓒ 블록테크미디어, 무단 전재 및 재배포 금지