The company future in the last few hours, it announced that it had launched RadioGPT, the first station in the world to have announcers powered by artificial intelligence (AI).
This new station is characterized by having a system that generates audio for transmission and that according to the portal Enter, it is based on what is happening in real time in a specific area.
This new station features GPT-3 technology, which features a social content system, as well as AI-driven story discovery. This new system that also has TopicPulse technology will be in charge of tracking information, where they will highlight what is trending on social networks.
Likewise, they will have the ability to create a script and once ready, the voices of artificial intelligence will convert it into audio. They also have the ability to create blogs, social posts, and other content for digital platforms, which will be related to the content that is on the air.
The TopicPulse system will also be able to create short, AI-powered videos on topics that are relevant to society. “As pioneers of artificial intelligence in the broadcast space, it’s only natural that we bring the incredible power of GPT-3 technology, along with innovative technology like TopicPulse, to radio,” the company said in a statement.
On the other hand, they stated that they are now developing artificial intelligence for television, in addition to the creation of chatbots, with the purpose that viewers and listeners can communicate.
“GPT-powered chatbots can provide real-time responses to queries, provide updates on current events, or even strike up a small conversation with the listener or viewer,” they added.
ChatGPT: this is the way in which the chatbot will be integrated into more ‘apps’
Days ago Open AI announced that it is now available the application programming interface (API) for ChatGPT and Whisperwith which developers will be able to integrate the chatbot and its AI-powered speech-to-text technology into their applications and services.
It should be remembered that the ChatGPT chatbot is a conversational AI tool capable of chatting with users, answering their questions and in developed text format. This tool has surprised users due to the nature of its responses and its ability to generate and link ideas.
However, the company announced in a statement on its website the arrival of the API for developers, who will have access to the new model family called gpt-3.5-turbo.
In this senseOpen AI defines gpt-3.5-turbo as “the best model for many non-chat cases”which means that the ChatGPT API will be able to be used for more functions, in addition to creating an AI-powered conversational interface.
Also, as specified by Open AI, gpt-3.5-turbo is priced at $0.002 per 1,000 tokens. That’s about ten times cheaper than their existing GPT-3.5 models. This decrease is based on “a series of optimizations throughout the system” with which they have achieved a 90 percent cost reduction.
The tokens are used to pay for the use of the OpenAI neural network, the company explains. They can be thought of as “word pieces”; Before the API processes the requests, it divides them into tokens to measure how much is being used in words.
For example, a token equals four English characters. However, the company also provides the ‘Tokenizer’ tool to check how many tokens will be needed to interpret a text string.
That being said, the ChatGPT API also offers dedicated instances for developers. Specifically, the Open AI API runs on Azure, and with Dedicated Instances, developers will pay for a period of time on a reserved compute infrastructure allocation to serve their requests.