OpenAI has made a habit of using its GPT-3 text generator to work for the design of many tools. The company’s latest experiment is based on the idea of providing book summaries of any size.
Concretely, the system works by summarizing small sections of a book, then each of the summaries is assembled to present the whole in a global way. OpenAI describes this process as “ recursive decomposition of tasks ».
When AI suggests reading choices based on your mood
Such a tool, if it turned out to be truly effective, would be likely to appeal to the general public but also and above all to companies which have recourse to a great deal of documentation, for example software developers. This could save time and simplify the task of employees who spend an average of 9.3 hours per week searching for and gathering information related to their work, according to figures from a SearchYourCloud survey, relayed by VentureBeat.
Our colleagues also recall that the company is far from being the only one interested in this technology. Tech giants such as Google, Microsoft, and Facebook are working on AI capable of crafting summaries.
For now, OpenAI recognizes that its system has limitations. While the tool generated summaries with important information, sometimes they also contain inaccurate data.
The whole thing is also not always very cohesive and often reads like a list of events, rather than a well-constructed text. Be that as it may, this work is nonetheless promising for the company which will continue its work.
Remember that the GPT-3 type text generator has given rise to many experiments. In the field of literature, the GPT-3 Books tool thus establishes book recommendations according to your mood of the moment. It is therefore possible to simply enter what you are looking for and the AI takes care of the rest.