AI toolkit for civil and military critical decisions

There is a need for “explainable AI” systems due the need to use AI in safety critical systems.  However, “black-box” style methodologies, such as Deep Learning Neural Networks (DLNN) are currently the principal AI approaches being currently investigated. There has been extensive work to make Neural Networks more explainable and trustworthy, with some progress being made. However, these primarily rely upon a second system to be made to attempt to explain the black-box AI.

Additionally, the black box AI research does not extend to interpretability, a key requirement to realizing the full potential of formal verification capabilities. Thales GFT technology is an AI/ML solution that can be inherently fully understood, trusted and formally verified.

Formal verification uses rigorous methods to specify, design, and analyze systems, ensuring algorithms meet defined safety specifications. A key feature of the Thales AI Toolkit is its ability to mathematically guarantee compliance with such constraints.
These specifications may include design requirements, standard procedures, or operational rules — for instance: “The AI must never ___” or “If conditions are ___, then output must remain within ___”.
Many aerospace and defense failures could have been avoided through formal verification.
Thales integrates these tools directly into the GFT AI development process. When issues arise, input and intermediary traces are provided. Thanks to the interpretable nature of the Fuzzy Tree, corrections can be made cleanly during development, with full visibility on their impact.
The Thales AI Toolkit blends numerical testing with formal verification to ensure a highly trustworthy AI system.

AI systems are now capable of superhuman performance in areas like autonomous vehicle control, decision-making, planning, and predictive analytics. However, their adoption in mission- and safety-critical applications has long been hindered by a lack of trust and explainability.
The Thales AI toolkit addresses this challenge by enabling the development of high-performance AI systems that are fully trustable and explainable, making them suitable for critical deployment. This breakthrough opens up ethical, “first to market” opportunities across multiple domains, offering improvements in savings, efficiency, performance, safety, mission effectiveness, or lethality over existing systems.
The toolkit has already been used to successfully implement Genetic Fuzzy Trees in real-time control, sensor fusion, predictive analytics, and decision-making. Use cases range from superhuman AI for beyond visual range air-to-air combat to treatment prediction for bipolar disorder — proving its value across both defense and healthcare sectors.

Thales AI toolkit encompasses GFT assets in a framework specifically designed for customers that are interested in developing their own AI system. GFT AI toolkit thus includes three unique dedicated software development tools:

• to create Fuzzy Tree AI systems
• to train and optimize the AI solution
• to formally verify the AI solution

Thales also offers professional services: from training which enables fully independent AI development through development support up to full turn-key project development, for customers that would like to delegate some or all of the development process to Thales.

This AI toolkit is Thales’s domain & platform-agnostic Machine Learning software of choice to develop verifiable and certifiable AI systems.