Study Examines How Machine Learning Drives Manufacturing | MIT News


Which companies are successfully deploying artificial intelligence (MI) and data analytics for manufacturing and operations? Why are these top adopters so far ahead and what can others learn?

MIT Machine Intelligence for Manufacturing and Operations (MIMO) and McKinsey and Company have the answer, revealed in a first of its kind harvard business review item. The article tells how MIMO and McKinsey teamed up for a large survey of 100 companies to explain how high-performing companies are successfully using machine learning technologies (and where others could improve).

Created by the MIT Leaders for Global Operations (LGO) program, MIMO is a research and education program designed to drive industrial competitiveness by accelerating the deployment and understanding of artificial intelligence. The goal is to “find the shortest path from data to impact,” says Managing Director Bruce Lawler SM ’92.

As such, the McKinsey project encapsulates MIMO’s mission to demystify the effective use of machine learning. The survey studied companies across all industries, probing their use of digital, data analytics and MI technology; objectives (ranging from efficiency to customer experience and environmental impact); and followed. Respondents came from the extensive networks of MIT and McKinsey.

“The study is probably the largest ever in the field: 100 companies and 21 performance indicators,” says Vijay D’Silva SM ’92, senior partner at McKinsey and Company who worked with MIMO on the project.

Overall, those that saw the greatest gains from digital technologies had strong governance, deployment, partnerships, IM-trained employees, and data availability. They also spent up to 60% more on machine learning than their competitors.

One notable company is biopharmaceutical giant Amgen, which uses deep learning image augmentation to maximize the efficiency of visual inspection systems. This technique pays off by increasing particle detection by 70% and reducing the need for manual inspections. AJ Tan PhD ’19, MBA ’21, SM ’21 has been instrumental in the effort: he wrote his LGO thesis on the project, winning the best thesis award last year upon obtaining his diploma.

Lawler says Tan’s work exemplifies MIMO’s mission to bridge the gap between machine learning and manufacturing before it’s too late.

“We saw the need to get these powerful new technologies into manufacturing faster. In the next 20 to 30 years, we’re going to add 3 billion more people to the planet, and they’re going to want the lifestyles that you and I enjoy. These usually require manufactured items. How to improve the translation of natural resources into human well-being? One of the main ways to do this is through manufacturing, and one of the newer tools is AI and machine learning,” he says.

For the survey, MIMO released a 30-page handbook for each company analyzing how it compared to other companies across a range of categories and metrics, from strategy to governance to data execution. This will help them target areas of opportunity or where to invest. Lawler hopes it will be a longitudinal study with a broader scope and playbook each year – a large but impactful undertaking with LGO’s mastermind as its driving force.

“MIT was extremely important and essential to the work and an incredible partner for us. We had talented MIT students on the team who did most of the analysis in conjunction with McKinsey, which improved the quality of the work accordingly. says D’Silva.

This collaborative approach is at the heart of MIMO’s philosophy as an information gatherer and private sector partner. The goal is to drive “effective transformation in industries that not only achieves technical goals, but also business goals and social goals,” says Duane Boning, director of the engineering faculty at MIT LGO and head of faculty at MIMO.

This fusion of research and collaboration is the next logical step for LGO, he says, as it has always been at the forefront of problem solving for global operations. Machine learning is certainly the latest big knowledge gap for many companies, but not the first, and MIMO can teach companies how to apply it.

“[I liken] 30 years ago, when LGO started, when it was all about lean manufacturing principles. About 15 years ago, that was the supply chain idea. It got us thinking – not just for our LGO students, but for the benefit of the industry in general – to understand this big shift, to facilitate it, to do research and make connections with other LGO activities. real research, we need efforts to catalyze that,” says Boning. “This is [MIMO’s] real excitement: What ideas work? What methodologies work? What technologies work? And LGO students, in a way, are the perfect vehicle to experience some of that. »


Comments are closed.