• Share:
Back to Stories
05/14/2021

Advancing technologies with Artificial Intelligence & Machine Learning

Megatrend Spotlight: Earlier this year, 3M shared the megatrends we see shaping the next 5-10 years of life on this planet. To understand the source of these trends, and their impact around the globe, we are taking a deeper dive into their key themes and the ways our people, technology and solutions are working to improve lives.


Technology has long been a measure of human advancement and power. When breakthroughs in innovation create fundamental shifts in society, we experience new levels of technological progress.

Part of what makes today unique and exciting is how frequent these breakthroughs occur. Massive leaps in innovation are happening exponentially.

One of the more exciting and interesting technological breakthroughs of today can be found in Artificial Intelligence and Machine Learning (AI / ML), which has the power to drive entire industries and society forward.

Artificial Intelligence can be defined as the development of computer systems able to make human-like decisions. AI takes in diverse information and stimuli to learn, adjust and make changes in real-time—ever-evolving an older version of itself.

It might sound like something out of a science fiction movie reserved for the most sophisticated tasks, but it is all around us working to improve, evolve, understand and solve human problems, both simple and complex.

At 3M, we are constantly studying, developing and implementing the latest technologies to improve lives through scientific innovation and the role AI plays in this mission is no exception.

We sat down with David Frazee, vice president, and Mani S. Upadhyaya, vice president, Data & Analytics for 3M’s Digital Transformation team, to better understand AI’s role in society and how it is being implemented here at 3M.

Technology is moving at an incredible pace. What are some of the ways the world of AI has changed in recent years?

DF: Most significantly, it is the availability of AI to a much larger audience. This has been enabled by two major factors: the dramatic increase in computing power from the cloud platforms at very low cost, and the widely available set of development tools that abstract the complexity of the compute infrastructure from the AI builders, developers and users. What used to be the province of a select few exclusive and well-financed universities or companies, is essentially available to anyone.

MU: AI has become more powerful, useful and accessible. Like David mentioned, it used to be that you would need a data scientist with a PhD, a powerful platform costing millions of dollars in upfront investment, and more money in operating costs to build any algorithms. But in recent years, with more open access online courses, open source platforms, easy-to-use tools and the falling cost of cloud computing, the barrier to entry has lowered and enabled citizen data scientists to apply AI to business problems.

What do you think are some of the more exciting aspects of AI’s current uses or potential uses?

DF: Using AI to improve lives. It’s very clear, and much more will be written about it in the future, that the vaccines developed in 2020 for COVID-19 could not have been created without AI. That is truly revolutionary. The ability to run large experiments with big data and perform AI analysis to learn quickly about many aspects of the vaccine programs was one of the most important technologies used. The use of AI in the broader healthcare industry is just the beginning. AI may be used to find patterns and then predict individual or population health events, which could enable therapy or treatment almost immediately.

MU: There are many diverse industries that AI can improve. Real-time AI and machine learning are becoming very sophisticated. We’ll see improvements in user experience, things like individual recommendations for users of Tik-Tok and Netflix , Uber pricing. Then there are advancements in autonomous vehicles, industrial machinery maintenance, which self-diagnose and self-heal—things like aircraft engines, manufacturing plants. AI will help with planning and optimization software to sense demand, plan accordingly and manufacture efficiently. AI can help improve customer experience by interacting intelligently with customers and reducing issues with their orders. The discovery of new medicines in pharmaceutical industry, with the ability to run through millions of combinations of molecules and shortlist a few promising candidates. Identification of risks in supply chain, for example, a traffic incident or severe weather, and then taking preemptive actions. The list goes on and it’s very exciting.

What types of barriers or challenges does the AI industry face in this time?

MU: A big thing we see is a shortage of talent right now as it’s a growing and evolving industry. Data scientists and translators are in short supply. They are the people with the ability to translate business requirements into data science algorithm requirements (and vice versa). This also leads to data engineers who cleanse and transform the data for data scientists to build algorithms. Another part of it is data and systems. Availability of data is crucial, as we will need good and well-connected data to make the leaps in advancements we mentioned before. Development tools and platforms are becoming widely available but remain complex to use, which is a barrier to entry. Lastly, there are network connectivity advancements. The optimal way to process very large amounts of data is to move it into cloud. Network bandwidth becomes a constraint so 5G networks are a promising technology

What are some things we must be aware of as AI grows and evolves?

DF: There are two things that come immediately to mind. The first is the unethical use of AI (and data) by design or unknowingly. Things like cyberattacks, but also finding gaps in physical security or planning illicit activities. Unintended bias in the AI models can also have deleterious consequences. Secondly, finding the right balance of privacy and use of data. The US is the last developed country with no modern federal privacy statute, which is important when viewing the role of AI in the future. These are things we take very seriously at 3M and are at the forefront of all our developments.

What are some of the ways 3M is using AI (in your area of the company) to improve lives or industry and what are some of the complex problems 3M is trying to solve using AI in your area?

DF: We have many examples of AI right here at 3M. Our Health Information Systems Division uses AI to quickly analyze records for optimal care and economical treatments. Our Transportation Safety Division and Personal Safety Division are looking at road, building and environmental data to optimize the safety of workers and equipment in real-time. And our Construction and Home Improvements Market Division is using AI to predict furnace filter life and advise the homeowner how to maximum home air quality. A big advancement area is also in materials discovery. How does 3M improve its ability to develop and then manufacture new materials and processes 3, 4 or even 10x faster? Deploying sophisticated AI tools to the 3M factory can increase quality and uptime significantly.

MU: We use AI in a variety of commercial products as well as in the enterprise. New advanced algorithms for improving the global supply chain as well as improving the quality of production. Predicting customer disputes and resolving them. We are using AI to improve sourcing, meet customer service needs, predict sales run rates and improve financial operations as well.

What areas or industries in our world do you see AI making a major impact either currently or in the future?

MU: There is a long list of advancements and we at 3M are constantly keeping our finger on the pulse of various industries. Some that come to my mind: improving automobiles, predictive maintenance, optimizing manufacturing & supply chain operations, improving customer service and customer experience, better inventory and sourcing management, improving safety and security, traffic management, delivering better public service, lowering energy consumption through efficient operations. And ushering in big improvements to operational efficiency, customer experience and cyber security.

In what ways do you see artificial intelligence shaping the next 5 years?

DF: The majority of AI will be invisible to us. Voice, facial and other recognition technology will be even more ubiquitous. A simple example can be found in the future of shopping. For shopping, you get personalized recommendations, complete checkout, pick up your goods and walk out, your account is charged because it knows you and all the items you purchased with no checkout line at all. Diet and exercise can be tuned to an individual’s needs. Depending on your health data, it may recommend less carbs for this meal, and which proteins are optimal. And then learn what works for you. Sports will change. Professional and college teams will use AI to keep players safer and create new ways to compete. Major League Baseball has already changed as they incorporate more and more data and analytics to hitters and to position fielders. It will only get more sophisticated. Robots will use AI to learn how to improve their usefulness. Self-learning of these will lead to improved quality of manufacturing, automated cleaning, lawn care, replacement of firefighters... Why send people into burning buildings when you can send a dozen robots who can go where people cannot? The opportunities for AI are almost limitless.


Multimedia Files:

Back to Stories