In 2018, craft ai has been selected among the best AI solutions by DGA & Dassault Aviation through the Man Machine Teaming (MMT) call for projects to apply its explainable AI solution to the Cognitive Air Combat System.
How can explainable AI meet defense needs?
AI is becoming a major technology issue for the Defense sector and contributes to maintaining operational superiority. AI helps defense forces carry out their missions in an operational environment that grows more complex every day. AI is a great support to help make more effective decisions, better understand and anticipate threats in order to act more quickly, accurately and confidently.
To apply AI in its strategy, the Defence sector needs to maintain Human control. craft ai explainable AI API fits to the needs of the Defence operational teams. Based on 3 major features craft ai provides:
- White box machine learning, because operational teams need to ensure trust and control of their AIs, not black boxes. Every decision is provided with explanations and can be challenged by the crews if they consider it no longer relevant, resulting in the update & the reinforcement of the predictive model.
- Automated Continuous Learning because nobody can afford to depend on regular updates by technical experts.
- Learning at the Individual level, to know the habits and needs for each crew member in order to provide personalized decisions, adapted to each mission context.
A proactive assistant for better contextualized decision…
To carry out their missions successfully, pilots must perform a certain number of actions to configure their user interface (visualization of the tactical situation, sensors, communication…). This can be tedious and critical during the intensive phases of the mission as the amount of data grows.
Therefore craft ai proposes to simplify the configuration of the pilot interface through its ASPIC assistant and provide a synthetic vision to the pilot with just the useful information, by:
- minimizing the information displayed in the tactical visualization according to the context;
- suggesting sequences of actions to be carried out during the mission based on past missions and the context;
- recommending the best configuration according to the context and the evolution of the situation
…that learns on his own to become even better!
The ASPIC assistant is built from predictive models generated by Machine Learning. These models enable to create dynamic decision rules that are used to recommend the most appropriate UI configurations at the right time, based on the current context and past situations.
craft ai is explainable by design, at the individual level and based on continuous learning. This makes ASPIC capable of differentiating each crew by learning their own habits and sharing experiences. The data collected from new missions will be automatically digested to improve craft ai models. Finally, every recommendation and decision will be provided with a confidence rate and the underlying reasons, to maintain control and ensure acceptance of the concept.
craft ai is a dual, explainable AI technology
ASPIC is a good example of an application of explainable AI on a predictive UX case. craft ai is also applied to similar cases in other sectors including banking, healthcare and human resources. craft ai is already deployed by a dozen customers from various industries such as energy, IOT, industry 4.0, utilities… . We can mention major leaders such as Dalkia, Total Direct Energie, Medclinik, Paris City Hall….. that apply craft ai explainable AI on high added value use cases ranging from business process automation, predictive maintenance, coaching to predictive UX… . MORE +
If you want to learn more about craft ai solutions :
Note: Article made with by craft ai, based on a real use case.