The COALA project (COgnitive Assisted agile manufacturing for a LAbor force supported by trustworthy Artificial Intelligence) was to create a digital intelligent assistant (DIA) powered by AI to train and guide workers in industrial environments. It will transform how workers perform their jobs and allow companies to increase their production process and product quality.
The EU-funded COALA project has been launched to design and develop a pioneering intelligent digital assistant for the manufacturing sector. At the heart of this system is the Mycroft voice assistant, an open-source software, privacy focused.
That core was complemented with a chatbot, a Prescriptive Analytics, a Cognitive Advisor and explainable AI components creating an infrastructure flexible and powerful in terms of Natural Language Understanding and Dialog Management for field training support functions. An AI-focused change management process and guidelines for the professional training of workers have completed the technical work.
The ANITI team members involved in the project - Joao Marques Silva, Yacine Izza, Xuanxiang Huang, Ricardo Passos, Nicholas Asher and Céline Castets - Renard have contributed for 2 subjects: Trust and Ethics in AI.
For Trust, ANITI introduced XAI over the classifier models that presented recommendations to users. After the theory, the teams developed and implemented a component (named WHY Engine that “explains” the results/suggestions/recommendations from the used machine learning models in the Digital Intelligent Assistant (DIA).
For "AI Ethics" ANITI ANITI coordinated the management process of avoiding ethical issues derived from the usage of AI models, in the person of Nicholas Asher. An Ethics Board managed and analysed the ethical impacts of using an AI-assisted digital assistant from different perspectives. Topics such as Trust, Privacy, Worker autonomy, Balance of Power, Bias and Accountability were addressed throughout the project.
The project is now over and time is to take stock and study the possibilities for the WHY Engine for becoming a real product.