«Generative AI and its possibilities (5/5). Artificial intelligence has necessitated the search for new ways to refine, index, and store information. Mathematical vectors are the most suitable objects for storing information from diverse data by rationalising it through mathematical knowledge. This is how the idea of the vector database was born. In this slightly more technical article, we will explore vector databases, also known as vector databases, which play a fundamental role in the storage, search, and manipulation of complex data.
The models used for generative AI are the famous LLMs. The name is explicit in itself: they are computer models produced by processing very, very large volumes of text. Obviously, this is not the kind of exercise a company can reproduce on its own, and it is therefore worth outlining the possible adoption methods.
If a company wants to use generative AI, it has 4 and only 4 options:
We will review these 4 options:
Using a model in its current state is akin to discussing «prompt engineering». In layman's terms, it's really the art of asking a model a (good) question. This is, by design, the «low cost» solution for using a model as it only requires work on how to use the solution, rather than particular customisation work. Two sub-options are then available to you:
The art of «prompt engineering» will then be to provide a question with enough context for the generic model to answer your specific question.
Each business area is different, making customisation necessary. The procedure then involves fine-tuning the system by submitting examples. The complexity then lies in providing the appropriate documentation or the example responses desired in a given context.
Training must then be carried out which will lead to the generation of a new, more specific model. The burden (and cost) of such an exercise will depend on the complexity of the content and the level of accuracy expected, but will provide an infinitely reusable basis.
The combined model aims to strike a balance between the cost of training on stable data assets and adaptation to a particular context.
Generative AI is not limited to text-based work. JEMS masters a comprehensive catalogue of solutions that can address other uses related to sound, image, and more.
Depending on your company's needs, a particular technology will be considered for improving your productivity (Model 1) or creating new services (Models 2-4). JEMS is technology agnostic but possesses unique expertise in the methodology of creating data assets. This method has been proven with some of the largest CAC40 clients. JEMS has the capabilities and know-how to urbanise products based on generative AI within a more global vision.
In essence, JEMS's ability to adapt LLMs to your needs, in addition to its expertise on other types of artificial intelligence models, will enable you to be best supported in bringing your ideas to fruition, from the simplest to the most complex.
