Drive AI Adoption Using Continuous Process Methodologies
Photo by Kelly Sikkema / Unsplash
Artificial Intelligence (AI) is still making headlines for its capabilities for data analysis, idea generation, and coding abilities. In a recent Harvard Business Review article by Sowmyanarayan Sampath, Verizon Consumer Group CEO, a compelling case was made for a decentralized, frontline-driven approach to AI adoption at Verizon. This strategy diverges from top-down centralized management by empowering those closest to the work to lead AI initiatives. I would like to explore how this innovative approach aligns with established continuous improvement methodologies such as Lean, Six Sigma, Agile, Total Quality Management (TQM), and Kaizen, highlighting the methodologies that can enhance AI integration and organizational performance.
Understanding Continuous Improvement
Continuous improvement is a principle in modern business operations, centered around the ongoing effort to enhance products, services, or processes. Among the key methodologies that embody this philosophy are Lean, which focuses on waste reduction; Six Sigma, which targets variance and defect reduction; Agile, known for its flexibility and responsiveness to change; Total Quality Management (TQM), which involves all employees in the improvement process to enhance quality and performance holistically; and Kaizen, a strategy of continuous, small, incremental improvements initiated by the workforce. Each methodology offers distinct advantages but shares some of the same risks; they need leadership support but can easily be derailed by that same leadership. The methodologies have their focus and were brought to the forefront of business by innovators trying to build problem-solving frameworks. Each methodology can be uniquely leveraged to integrate AI.
Lean
Lean methodology is renowned for its focus on eliminating all forms of waste in a process, including time, labor, and materials, thereby enhancing overall efficiency and value to the customer. It emphasizes techniques like value stream mapping, just-in-time manufacturing, and the 5S system. Instead of pushing work into the system, practitioners find a way to pull work in based on customer demand. Unfortunately, Lean is often associated with manufacturing and overlooked by service organizations. By only doing work when needed organizations can reduce work in process (WIP) and maximize cash by not tying it up in inventory or wasted actions.
While the roots of Lean can be traced back to the Toyota Production System (TPS) in post-war Japan, it was Taiichi Ohno and Shigeo Shingo who are credited with developing the core concepts. Toyota had a problem not experienced by US car companies at the time, they had a small market with diverse needs and couldn't generate economies of scale by producing large batches of one car model at a time. This required producing multiple car models on one production line requiring the rapid changing of equipment and processes for the next vehicle.
In AI projects, Lean can help streamline development processes by identifying and removing unnecessary steps, optimizing resource usage, and ensuring that AI solutions are developed and deployed efficiently and effectively.
Six Sigma
Six Sigma (6σ) is known for its data-driven approach aimed at minimizing defects and reducing variability in processes. It uses statistical methods to progress methodically through stage gates. To improve outputs, existing processes use DMAIC (Define, Measure, Analyze, Improve, Control), and to create new processes use DMADV (Define, Measure, Analyze, Design, Verify). Six Sigma can often be combined with Lean (LSS). At Verizon, I earned the LSS Green Belt and eventually Blackbelt through education and completing several projects. Operating at sigma level 6 means only 3.4 defects per million outputs.
Six Sigma was developed by Bill Smith at Motorola in the 1980s to identify and remove causes of defects. It was popularized by Jack Welch at General Electric in the 1990s, where it became integral to the company's corporate culture and operational success.
Six Sigma can be applied to AI projects to improve the accuracy and quality of AI models by rigorously measuring and optimizing every stage of AI development and deployment, thus ensuring that the final product meets strict quality standards.
Agile
Agile methodology is celebrated for its adaptability and flexibility, allowing teams to respond to changes in customer needs quickly through its timeboxed development cycles called sprints. By delivering testable software (or product) every few weeks defects are identified quickly and costly rework is avoided. The small teams are typically composed of cross-functional members, including developers, project managers, and quality assurance professionals, all working together. This structure knocks down silos which fosters a high degree of communication and accountability. Each team member brings their unique skills directly to the table, which streamlines decision-making and problem-solving.
The Agile Manifesto, published in 2001 by 17 software developers including Jeff Sutherland, Ken Schwaber, and Jim Highsmith, sought to promote a more flexible and iterative approach to software development.
Agile is particularly suitable for AI projects due to its iterative nature, allowing for continuous integration and testing of AI and adapting rapidly to feedback. To protect customers and proprietary data, these teams should include legal and data governance members.
Total Quality Management (TQM)
TQM is a comprehensive management approach that focuses on long-term success through customer satisfaction and involves all members of an organization in improving processes, products, services, and the culture they work in. It emphasizes quality management principles and tools, continuous training, and customer-focused planning. Like Six Sigma, TQM uses data-heavy tools such as flow charts, control charts, histograms, and Pareto charts to make evidence-based decisions.
TQM was popularized by management consultants in the United States in the 1980s such as W. Edwards Deming and Joseph Juran, who are credited with laying the foundations for quality management practices in Japan and globally.
TQM can guide AI initiatives by fostering a culture of continuous improvement and cross-departmental collaboration, ensuring that AI solutions are consistently aligned with customer, user, and organizational needs.
Kaizen
Kaizen "change for the better" is known for its principle of continuous, incremental improvement involving every employee from managers to the frontline. The approach emphasizes small, incremental changes in a process to solve a particular problem or reach a desired goal.
Like Lean, Kaizen was made prominent through its application at Toyota. Masaaki Imai further popularized the concept globally with his book "Kaizen: The Key to Japan's Competitive Success" in 1986.
Kaizen can be implemented in AI projects to foster a culture of continuous learning and adaptation, allowing teams to make small, regular improvements to AI algorithms and applications, which helps in refining their functionality and increasing their impact over time.
Conclusion
One common thread throughout all these methodologies is going to "see" the process, assess the problem, and develop a solution. This means working with the people doing the work and as Sampath put it "You need to understand how stuff gets done. Czars rarely figure that out, because they are sitting too far away from the supply line of information where the work happens."