Raytheon to develop AI that can explain its decisions
Industrial technology firm Raytheon is on a mission to develop an artificial intelligence technology that can explain its decisions as it makes them.
The first-of-its-kind neural network will be designed as part of the Defense Research Project Agency (DARPA) Explainable Artificial Intelligence program, or XAI.
According to Raytheon, the XAI program aims to create machine learning techniques that can operate at a high performance level and produce models that are more explainable.
This will help humans to understand, trust, and manage an emerging generation of artificially intelligent partners.
Raytheon BBN's Explainable Question Answering System (EQUAS) will allow AI programs to 'show their work,' increasing the human user's confidence in the machine's suggestions.
Raytheon BBN EQUAS principal investigator and lead scientist Bill Ferguson says there is one main goal.
"Our goal is to give the user enough information about how the machine's answer was derived and show that the system considered relevant information so users feel comfortable acting on the system's recommendation,” he explains.
The EQUAS system will be able to show what data mattered most in an AI’s decision making process. It is designed with a graphical interface so users can explore the system’s recommendations.
Raytheon says that the system is still in its early development phases, but the final product could have wide-ranging uses in defence, security, and healthcare.
"A fully developed system like EQUAS could help with decision-making not only in Department of Defense operations, but in a range of other applications like campus security, industrial operations and the medical field," explains Ferguson.
"Say a doctor has an x-ray image of a lung and her AI system says that its cancer. She asks why and the system highlights what it thinks are suspicious shadows, which she had previously disregarded as artifacts of the X-ray process. Now the doctor can make the call – to diagnose, investigate further, or, if she still thinks the system is in error, to let it go."
Raytheon says that as the EQUAS system is improved, it will develop self-monitoring capabilities and share things that limit its ability to make reliable decisions.
These will help refine AI systems and allow developers to add additional data, or change how data is processed.
Raytheon is heavily involved in the development of solutions for defense, cybersecurity, and civil government. Its 2017 sales reached $25 billion.
The company recently announced a contract with the Australian Defence Science and Technology to help develop advanced electronic warfare technologies for the Australian Defence Force.