Powered by OpenAI’s innovative mannequin, the voice mode seamlessly integrates voice, text, and vision capabilities. Initially, the corporation will roll out its suggested-assembled chatbot to a small cohort of premium customers subscribed to ChatGPT Plus, before expanding availability to all paying users by autumn. The ChatGPT Plus subscription costs $20 per month. OpenAI plans to inform participants in the initial rollout group within its ChatGPT app, providing guidance on utilizing the new model.
The delayed launch of our innovative voice function, originally scheduled for this month, is now set to arrive a month later, following an extended development period aimed at bolstering its robust security features, including advanced capabilities to detect and block unwanted content. The corporation further announced that it is gearing up its infrastructure to provide instantaneous responses to hundreds of millions of clients.
According to OpenAI, the company has conducted an exhaustive assessment of its mannequin’s voice capabilities by engaging more than 100 external “red-teamers,” a diverse group of individuals tasked with identifying and exploiting potential vulnerabilities in the technology. According to OpenAI, these testers collectively spoke a total of 45 languages, hailing from 29 international locations.
The corporation claims to have implemented various security measures. To prevent the mannequin’s misuse in generating audio deepfakes, such as mimicking famous individuals’ voices, the system has developed four preconfigured voice profiles in partnership with professional voice artists. GPT-4 will not impersonate or generate others’ voices.
When OpenAI launched GPT-4, it faced criticism for using a voice called “Sky” that bore an uncanny resemblance to actress Scarlett Johansson’s. Scarlett Johansson revealed that she was approached by the company to permit the utilization of her voice for the artificial intelligence-powered doll, an offer which she refused. The artist revealed her astonishment upon hearing a voice “uncannily similar” to her own emanating from the mannequin’s audio sample, leaving her deeply unsettled. OpenAI had previously utilized Scarlett Johansson’s voice for its technology called Sky, but it has since halted its use.
The corporation may become entangled in multiple legal disputes regarding purported copyright violations. OpenAI claims to have implemented filters that recognize and prohibit requests for generating music or other copyrighted audio content. OpenAI claims that it has deployed the same robust security measures used in its text-based models for GPT-4, designed to prevent the AI from generating harmful or illegal content.
OpenAI intends to expand its capabilities by incorporating advanced features, including video and screen sharing capabilities, that will further enhance the assistant’s utility. During the In Its Might demo, participants held up their phone cameras and asked the AI model to help clarify complex math equations written on a piece of paper. The team members simultaneously shared their PC screens with the instructor, who guided the life-sized mannequin in resolving coding errors. Although OpenAI indicates that these options will not be available for the time being, they may become accessible at a future, yet to be determined, point in time.
