The last weekend, OpenAI (a non-profit artificial intelligence research company) announced the launch of an API, which will do to access new models of artificial intelligence developed by the organization.
The peculiarity of the new API, According to the organization, it is that unlike most artificial intelligence systems that are designed for a single use case, this API provides a general purpose interface, allowing users to test it on almost any linguistic task as long as it is in English.
It can be applied, with just a few sample data, to semantic search, summary, sentiment analysis, content generation, translation and more, according to an OpenAI announcement.
The company's new API allows developers to call their artificial intelligence tools on "virtually any task in the English language".
This API, allows you to refine performance on specific tasks by training on a data set (small or large) of examples you provide, or by learning from human feedback provided by users.
In the presence of any text message, the API returns a completion text, trying to match the template you gave it.
What does the OpenAI API do?
If you have a task, OpenAI can automate it. The different capabilities of the GPT-3 family of natural language understanding models are available to developers who can access the private beta.
In the context of semantic search, the API search for documents based on natural language meaning requests rather than keyword matching. For example, you could count on your ability to navigate through a large amount of text at once to answer a question about an article or find a relevant part.
The API can also allow quick discussions, complex and consistent in natural language, within the framework of the chat.
With a short notice, the API generates dialogs that cover a wide range of topics, from space travel to history, according to the company. By exploiting search and chat capabilities, the API generates a natural dialogue to quickly provide customers with relevant information, which could be useful for customer services.
Thanks to the semantic understanding of texts, the API can offer a variety of analysis and productivity tools to better serve customers.
So far, OpenAI says it has partnered with a dozen companiess to test the API before offering it on a larger scale. One of the companies is Quizlet, a learning platform that provides study tools to help people practice and master what they are learning.. Combining cognitive science and machine learning, Quizlet guides students through adaptive learning activities to confidently achieve their learning goals.
One of the most common uses of Quizlet is to learn vocabulary faster. To enable deeper understanding than memorization, Quizlet leverages OpenAI's powerful text generation capabilities to automatically generate examples of how each vocabulary word can be used in a sentence, literally in the OpenAI article.
According to the organization:
“By combining the OpenAI API with your machine learning work, Quizlet will be able to develop sample sentences for people who study vocabulary and languages, just like a tutor does, to help students integrate their knowledge in a fun way and test themselves themselves more fully ”.
The project was widely seen as a great step forward. Users can enter each GPT-2 text message - multiple lines of a song, a short story, and even a scientific article - and the software will continue to write, to some extent, matching style and content.
This year, OpenAI announced a more advanced and 100 times larger version of the system, called GPT-3, which has now become its first commercial product.
The launch of the GPT-3-based API has sparked excitement among commentators who see it as the democratization of technology.
link: https://beta.openai.com