Articifial Intelligence is a technology that belongs to the field of computer science and aims to create systems and algorithms that run in a dynamic environment, based on the collection and processing of data. These computer programs must be able to simulate human intelligence. The main objective of AI is to create intelligent machines that can help solve complex problems in many fields.
There are several categories of Articifial Intelligence (AI), which can be classified according to their capability and level of sophistication. Here are some of the most common categories of Artificial Intelligence :
To get as close as possible to human behaviour, Articifial Intelligence needs a lot of data, as well as a processing and learning capacity. To achieve this, three components are needed:
To enable computers to learn from data, Articifial Intelligence relies on Machine Learning models (a method that aims to teach machines to learn from data and improve with experience). There are 3 learning methods used:
It is important to note that AI is a constantly evolving field of research, and that definitions and distinctions between different types of Artificial Intelligence may change.
The main goal of AI is to create intelligent machines that can help solve complex problems in many fields, such as medicine, engineering, finance, security, social sciences, gaming, etc. Articifial Intelligence is seen as a key technology for the future and has the potential to transform the way we live and work.
These examples are just a small portion of the application areas for AI, and new uses are regularly discovered as the technology continues to advance. AI can therefore be used to improve efficiency, accuracy, safety, and quality in many different areas.
One of the most popular supervised AI at the moment is Chat GPT (Chat Generative Pre-trained Transformer). This conversational tool aims to help its users solve problems, answer questions, and provide information in various domains. It therefore generates text from input data (questions, queries, etc.). It is based on Natural Language Processing technology (NLP) and uses unsupervised learning. It has been trained on a very large corpus of text to learn how to generate consistent and relevant answers based on user queries. This is possible because it has access to huge amounts of text, from various sources.
As a pure player in the R&D ecosystem, FI Group France has created a Scientific Department back in 2019. They lead research in Artificial Intelligence and NLP mostly. This department is composed of seven Researchers (including two Industrial PhD CIFRE). One PhD student is conducting a thesis on data extraction and the construction of algorithms to evolve their grouping by theme and subject. The second PhD student is doing a thesis on unsupervised learning on data flows. She is developing methods capable of clustering data continuously.
The objective of this department is to allow the realization of a regular scientific and technological watch to propose new approaches, and thus to support R&D Financing Consultants in their daily missions.
These projects are possible thanks to the development and experimentation of techniques in Machine Learning and Automatic Language Processing. These processes facilitate the search for information in a large volume of data. A third research topic concerns the acquisition of new knowledge and the involvement of collaborators, via Gamification processes and serious games.
One of the projects supported by FI Group is called NASA. This AI makes it possible to search for scientific articles based on various concepts.
For each query, the articles published between 2019 and 2021 (about 13 million) are used to represent this knowledge in the form of a graph of concepts. It is then possible to display 10 scientific articles published for each concept.
The first prototype of NASA “First STEP” (Scientific Taxonomy Exploration Prototype) was launched in March 2022. The second and the third were respectively put online in September 2022 and February 2023. This “Third STEP” proposes improvements in performance, quality and user experience.