Rapid innovation and adoption of artificial intelligence (AI) over recent years is transforming the nature of work across every industry and reshaping how people interact and experience the world around them. AI technologies not only help businesses deliver smarter products and services to their customers, but they also have enormous potential to drive broader positive change by powering solutions to longstanding consumer and community challenges in spheres like health care, fraud detection and data security, and the clean energy transition.
Business Roundtable Member Company Case Studies
AI helps Visa stay ahead of fraudsters by analyzing data and evaluating the risk of transactions in real time.
Visa leverages the power of data and data-driven tools, such as AI, for a wide variety of purposes across its business. Visa pioneered the use of AI models in payments in 1993, using neural network-based technologies for real-time, risk-based fraud analytics.
Today, Visa’s Advanced Authorization (VAA) combines Visa’s proprietary online model with offline neural network-based machine learning to evaluate fraud risk across its entire network (VisaNet) in real-time, helping to mitigate potential fraud risk before transactions go through. In 2022, VAA helped prevent approximately USD 27 billion in fraud.
Visa’s fraud solutions like VAA help keep the payments ecosystem safe for consumers and provide a trustworthy foundation for global commerce. Visa believes that strong guardrails for data are inextricably linked to responsible AI governance; and invests in risk and governance processes for the benefit of consumers, clients, and the larger payments ecosystem.
SAP AI enables an end-to-end, automated platform for COVID-19 financial aid applications.
During the COVID-19 pandemic, the cultural sector in Germany was widely affected by government health restrictions as theater performances, concerts, and cinema screenings were all temporarily ceased. To ensure that arts and culture organizations could survive the shutdown, the government created a short-term financial aid fund.
SAP was engaged to build an aid application platform that could manage complex processes, comply with regional regulations, and minimize time-to-delivery. SAP’s Document Information Extraction and Business Entity Recognition tools provide authorities with a highly automated and streamlined process for evaluating aid applications. For example, these AI-based tools can verify whether detailed data uploaded by businesses, such as proof of event cost, revenue, identity, etc. matches information on the aid application.
As a result, the German government was able to implement and operate a fully digital, end-to-end process within three weeks, evaluate over 160,000 documents, and provide more than €2.5 billion of aid to important cultural institutions. Since the initial launch, more than 2.7 million documents have been automatically evaluated and classified from the 47,000 cultural aid applications filed on the platform.
SAP AI solutions like the Cultural COVID-19 Aid Application Platform are developed in accordance with SAP’s Global AI Ethics policy, which operationalizes and enforces the company’s guiding principles of responsible development, governance, and use across their AI applications.
Johnson Controls’ digital platform for Smart Buildings improves efficiency and reduces environmental impact.
Buildings account for nearly 40% of the world’s greenhouse gas emissions, and there is no decarbonizing our future without decarbonizing buildings through digital transformation. Johnson Controls’ OpenBlue digital platform is a complete suite of connected solutions that allows customers to meet their net zero carbon emissions goals and create more sustainable buildings.
OpenBlue uses key AI tools like advanced machine learning to manage energy consumption and improve occupant productivity and comfort while minimizing the impact on carbon emissions. The platform drives up to a 50% reduction in energy usage and reinvents buildings to be a strategic asset. Machine learning—the only available technology capable of processing this much real-time data—helps deliver on the promise of autonomous buildings.
By leveraging AI, the OpenBlue platform enables customers to achieve strategic objectives while ensuring that the highest standards are met with respect to privacy, security, and trust. Just as important as delivering on customer outcomes is Johnson Controls’ foundational commitment to privacy and security. Johnson Controls is committed to “privacy by design,” and their Global Privacy Office is involved from the start in product development so that customers can have the utmost confidence that their privacy is respected and their data is protected.
Accenture helps to ensure Every Name Counts.
Accenture partnered with the Arolsen Archives - International Center on Nazi Persecution, which is part of UNESCO’s Memory of the World program and maintains the world’s most comprehensive archive on the victims of Nazi persecution.
The Center holds 110 million documents and objects, spanning more than 16 miles of shelving, with information on 17.5 million people. Making this archive accessible online is a massive and complex challenge. These documents are used to trace victims and survivors, reunite families torn apart, and support research.
When relying solely on the manual indexing of documents, volunteers can only transcribe about four documents per hour - a rate that could take decades, or longer, to make the collection entirely available.
Working alongside volunteers and historians, Accenture is implementing patent pending AI and optical character recognition solutions to enable the Archives to accelerate the speed of extraction at an estimated rate of 40x.
Accenture has facilitated the processing of 1.5 million documents with an estimated 2.08 million names. Over 2,300 Accenture volunteers from 115 cities have participated by validating the metadata and training the AI. Estimates suggest that without the AI, Accenture volunteers would have only been able to process an estimated 6,400 documents. Accenture is helping to make sure Every Name Counts.
Dell Technologies deploys AI systems to help manufacturing employees work better and smarter.
Dell has developed integrated AI systems that complement their labor force and enable teams to focus on value-added work in the manufacturing process.
Specifically, Dell works with site managers to identify and diagnose specific points along the productions process where there is potential to optimize human performance by integrating machine learning models. Dell has deployed high-resolution cameras along manufacturing lines to feed images of materials into “deep learning” models. With the help of human trainers, these integrated systems learn how to spot product errors and abnormalities more quickly and accurately than human workers.
This AI-enabled process does not eliminate the need for humans in the manufacturing process, but rather frees them up to do what they do best: conduct careful, pinpoint analysis of specific defects, while leaving the high-frequency, mundane work of flagging defects to the AI system.
The result is that manufacturers can achieve higher quality, more consistency, and increased productivity without sacrificing job satisfaction or increasing the stress and strain on their workforce.
Honeywell’s cloud-based machine learning solution helps optimize building energy efficiency leading to opportunities to reduce costs.
The Honeywell Forge Energy Optimization solution uses AI capabilities and machine learning to help building operators meet their efficiency goals and cut costs while making meaningful progress towards sustainability objectives.
The solution uses reinforcement learning based on near-real-time weather, temperature, building occupancy, and energy cost data to provide insights into heating and cooling demand and fine-tune building HVAC system settings around the clock.
Energy Optimization enables visibility into HVAC systems to help maintain peak efficiency without sacrificing occupant comfort by modeling heating and cooling demand and adjusting to changing conditions, such as weather patterns or occupancy increases. This increased efficiency can result in reduced energy consumption, which produces cost savings for consumers and can help reduce greenhouse gas emissions.
The SAS AI platform supports the transparency, explainability, and auditability of AI systems.
The SAS Viya platform gives organizations the tools and capabilities they need to build employee and customer confidence in AI-supported systems by providing robust oversight tools for both models and outcomes. The Viya platform detects biases, assesses model fairness, provides explainability, monitors system performance, and supports data privacy.
For example, SAS Viya helps healthcare providers ensure medical integrity in cancer detection through its privacy and decision auditing capabilities. It has also been used to identify inequities and flag potential biases in home loan and foreclosure policies.
By making an array of trustworthy AI tools and functions widely available to diverse organizations across different sectors of the economy, SAS educates and enables organizations to innovate responsibly on a trusted platform and operationalize Responsible AI principles into practice.
Booz Allen helps clients to develop and deploy secure, responsible AI solutions at scale.
Many organizations struggle with operationalizing the principles of RAI as they work to keep up with rapidly evolving best practices and regulations. There is a need for enterprise-grade AI that helps organizations meet multiple mandates: ensure compliance with RAI best practices, reduce time-to-mission, streamline collaboration and ease continual learning and refinement of deployed solutions.
Booz Allen’s aiSSEMBLE™ brings a holistic approach to operationalize RAI governance. Leveraging a set of pre-engineered adaptable tools, Booz Allen develops and deploys AI solutions at enterprise scale with embedded RAI principles and practices (e.g., transparency, traceability, auditability, dynamic data and model drift detection) into AI system design, execution, and monitoring—ensuring that RAI is built-in from the start.
The aiSSEMBLE™ framework has helped clients achieve maximum mission impact through an AI reference architecture, proprietary Responsible AI toolkit, established data delivery and machine learning patterns, and reusable software components. Together, these integrated features provide a consistent approach to achieve RAI by applying standardized AI principles and embedding them in the execution process, instead of engineering AI systems from scratch every time.
GM’s enterprise-wide framework promotes responsible use of AI in alignment with GM values.
GM understands that AI will power its vision of a Zero Crashes, Zero Emissions, and Zero Congestion future, as well as many aspects of the company’s digital services and customer experiences. The Responsible Use of AI @ GM initiative has incorporated Responsible AI principles into processes and practices throughout the company to ensure that AI deployment is aligned with GM values. The initiative was developed through an iterative crowd-sourced process, drawing on extensive input from diverse stakeholders, engagement with GM employees, and lessons learned from other organizations.
GM’s framework includes detailed principles that have improved enterprise-wide understanding around Responsible AI use practices. It has also encouraged engagement with AI vendors to ensure the systems and services GM procures also support company values.
GM is proactively incorporating responsible use into the foundation of AI innovations, building trust with customers and maintaining the integrity of the GM brand.
IBM’s Ethics by Design playbook integrates technology ethics across the AI development pipeline.
IBM’s Ethics by Design (EbD) Framework integrates tech ethics into the organization’s full tech development pipeline, including AI, and embeds responsible governance across the organization. By embedding RAI in existing governance pathways and processes, and by getting buy-in from every business unit, the EbD Framework centers RAI at the outset without acting as a break on innovation.
The EbD Playbook makes the Framework accessible and practical for AI developers and data scientists by including the tools, methods, and best practices needed to integrate technology ethics into their everyday jobs. Drawing on expertise and input from across business units, the Playbook helps organizations operationalize IBM’s tech ethics—enabling fairness assessments, supporting transparency documentation, facilitating explainability, and ensuring privacy.
The EbD Playbook has been used by hundreds of IBM developers and data scientists. As a result, IBM continues to be a trusted partner to its clients as it increasingly integrates AI and is a resource for clients that are interested in embedding AI-oriented ethics into their own organizations.