3.3 Awareness of Augmented Worker projects and solutions
COMPANIES’ DIFFICULTIES WITH DATA SHARING IS AN OBSTACLE TO THE AUGMENTED WORKER
At a more macroeconomic level, companies themselves do not know the volume and quality of the data they generate. And while they do know the quantity and quality of data produced, they are not aware of the possible uses. Moreover, they do not want to share their internal data for fear of disclosing sensitive information.
Study centers, data marketplaces and open-data platforms are trying to provide an initial response to this problem. But the desire not to share data (intellectual property, confidentiality, non-structuring of data) remains a real problem in the development of Augmented Worker solutions. Companies must therefore go beyond short-term competitive logics and define common languages to enable the development of multi-sector Augmented Worker solutions.
A MARKETPLACE OF MODELS TO ENCOURAGE SHARING
A marketplace of AI models could be imagined to obtain more accurate results. Concretely, an algorithm would gather data from multiple companies in order to capitalize on a large volume of data. The law of large numbers justifies this business intuition. The final goal is to share the lessons learned from processing data from several companies by mutualizing them, to erase the learning period of models thanks to turnkey models and to guarantee data confidentiality and a high level of security.
The data used to perform model learning (ML*) must respect a common nomenclature (segmentation/semantics) and the use of turnkey models requires clients to have a structured database.