March 2022CIOAPPLICATIONS.COM9Ensure equity and mitigate bias via inclusivityWhen designing solutions, we will ensure sufficient diversity and inclusion in their data, learning and testing to avoid biased training data, models, decisions, and outcomes. Diversified teams (e.g., SMEs, end users, security/safety experts) are needed to avoid unbalanced focus on system performance and optimization at the expense of system reliability, fairness, and/or safety. We aim for AI development and deployment teams that are inclusive across multiple kinds of diversity.Create symbiotic human machine interactionsWe aim to design solutions that effectively distribute tasks and workloads across humans and machines, ensuring appropriate roles and responsibilities (e.g., oversight over key decisions by humans, delegation of dull, dirty or dangerous tasks to machines). We seek to design effective human machine interfaces (Maybury and Wahlster 1998) including affordances that enable graceful degradation. Be aware (of self and others) and beneficentWe will strive to ensure that AI solutions enforce appropriate self-limitations and design machine intensions and actions that are task and situation appropriate. Be resilientWe will design solutions that are resilient by using methods such as diversified training and test sets that avoid classification failures, utilization of V&V methods and stress testing appropriate to the application and when switching environments. After systems are fielded, we will strive to actively monitor their performance in various deployment contexts and continuously assess their performance with their (reinforcement) learning over time.SummaryStanley Black & Decker is leveraging artificial intelligence and machine learning enhance products, accelerate manufacturing agility, and grow profitability. We are committed to equity, inclusion, safety, and sustainability. Given the criticality of AI and ML in work, home, and play, we aim to ensure our values are reflected in the realization of responsible products and services. Enabling responsible AI enables us to better serve humanity and be a force for global good for those who make the world. was deployed in 3 months in our Fort Mill Distribution Center with an 18-month payback. We employ cooperative (50+), mobile (60+), and traditional robots in addition to computer-numerical control (CNC) machines. Robots enhance worker safety and productivity by delegating and automating dull, dirty, and dangerous operations to machines and can improve productivity by double digits with rapid payback. Finally, digital twins can accelerate innovation. For example, in a few months Stanley Black & Decker saved $2M through simulation to refine a new US Craftsman plant design, removing expensive capital equipment when simulation revealed an opportunity for process and production streamlining. Responsible AI For Those Who Make The WorldTMOur deployment of AI and products and manufacturing is driven by our purpose of supporting the makers of the world, focused on building, creating and securing a better world for our customers, employees and communities. Artificial Intelligence (AI), which includes Machine Learning, promises to help us elevate that mission to make our lives more productive, safer and more joyful. Appreciating that new technologies and methods can pose new risks to society, we have developed the following principles that will guide the development of safe AI systems (IEEE 2021) at Stanley Black & Decker in alignment with our purpose as illustrated in Figure 3. Figure 3. Responsible Artificial IntelligenceBe truthful and transparent We will apply standards and protocols to create clarity and ensure understandable explanations (Maybury 2004) of the basis for and reasoning underlying an AI system's decisions and behaviors and ensure user awareness of conditions that could lead to adverse outcomes (e.g., a biased decision, a risky action).Ensure privacy and securityWe will design solutions that use methods such as data and information minimization, aggregation, and deidentification to reduce privacy attack vectors. We will apply digital product security practices that ensure confidentiality, availability, and integrity. Artificial Intelligence (AI), which includes Machine Learning, promises to help us elevate that mission to make our lives more productive, safer and more joyfulMark T. Maybury
< Page 8 | Page 10 >