The MLC’s first survey of AI and machine learning in manufacturing reveals growing experimentation with the technologies and a sober view of their effect on jobs.   By David R. Brousell

Opinions about the impact of artificial intelligence today range from the apocalyptic to the miraculous. Media darling Elon Musk of Tesla, for example, thinks AI is an “existential threat” to human civilization. Oracle CEO Mark Hurd believes a battle between the United States and China for “AI supremacy” will have important consequences for the global economy. And Ginny Rometty, IBM’s CEO, is convinced that AI has the power to transform industries in positive ways.

Whatever your view of AI, a term coined in 1955 by the computer scientist John McCarthy, the technology is at the forefront of discussions throughout society today, leading a debate about the future of work, jobs, and even what it means to be human. And as the manufacturing industry transitions to the digital era, AI is being viewed as central to leveraging the vast amounts of data that factories and plants will generate to do everything from improving operational efficiency to creating new, competitive advantages.

“Industrial AI can give the Fourth Industrial Revolution a huge boost and take Industrie 4.0 and similar initiatives to the next level,” said Roland Busch, Chief Operating Officer, CTO, and Member of the Managing Board of Siemens AG, in an article posted on the World Economic Forum’s website in January.

In an attempt to separate the hype from the reality of AI, and to take the measure of where AI and its cousin machine learning stand in manufacturing today, the Manufacturing Leadership Council undertook its first ever survey on manufacturers’ attitudes, plans, projects, and expectations with the technology earlier this year.

Chief among the survey’s findings is that, despite the hype, the 64-year old concept is at an early stage in most manufacturing companies. And while many companies expect AI to displace significant percentages of their workforces, they also anticipate that many of the displaced workers will be retrained for other roles in their companies, undercutting the notion that AI will inevitably lead to a vast wasteland of unemployed people. Moreover, a majority believes that while AI and machine learning are significant, they will not be transformative for the manufacturing industry.

PART 1: CHALLENGES TO AI ADOPTION

1 65% See Workforce
Changes Stemming from AI

Q: What percentage of your current workforce headcount do you expect will be replaced or removed by 2025 as a result of AI adoption?

2 But 60% Also See
Retraining for Those Displaced

Q: What percentage of the workforce displaced by AI adoption do you expect to be retrained for other roles in your company by 2025?

A majority of survey takers believes that AI and machine learning are significant but will not have a transformative impact on the industry.

3 Top 5 Challenges to AI

Q: What do you see as the biggest challenges to AI adoption in your organization today?

4 A Majority Sees AI as Significant But Not Transformative

Q: Ultimately, how significant an impact will AI and Machine Learning have on the manufacturing industry in the future?

Small Projects the Norm 

Digging deeper into what the survey data reveals about the status of AI and machine learning adoption, at an overall corporate level, 20% of respondents indicated that they are experimenting with a range of small-scale pilot projects in their companies and another 12% said single projects have been implemented. The largest group, 40%, are either in the stage of developing awareness of the technology, conducting research, or defining a roadmap (Q10).

The good news is that, over the next two years, survey respondents expect AI and machine learning investments to increase, in some cases substantially. More than 30% of respondents said they anticipate spending increases of between one and 10% in that timeframe, while 22% said 10-25%, and 14% indicated an increase of 25 to 50%.

At a departmental or functional level, manufacturing and production, with 60% of respondents indicating they have begun the adoption of AI, are the leading areas for the technology at present. Supply chain follows, at 30%, and research and development comes in third at 28%. But many other areas of the enterprise, from sales and marketing to quality operations, are also getting involved (Q11).

On the factory floor itself, 24% of survey respondents said they are implementing AI and machine learning on a single-project basis, while 48% are still going down the awareness, research, and roadmap trail (Q12). And among the application areas being addressed, process improvement, production planning, and preventative maintenance are getting the most attention.

PART 2: AI STRATEGY & ORGANIZATION

[tooltip text=”Tooltip Text”]

5 Few Have a Formal AI Strategy Today

Q: How would you characterize your company’s approach to AI and Machine Learning today?

[tooltip text=”Tooltip Text”]

6 AI Importance May Rise Dramatically

Q: How important do you think AI and Machine Learning is to your company in terms of business impact today, and how important will it be in in 2 years?

A Lack of Formality 

As companies proceed with pockets of AI and machine learning activity, they are doing so largely on an informal basis, suggesting experimentation with single or pilot projects to address a specific need or opportunity. Only 12.5% of survey respondents said their companies have a formal plan and strategy in place for the adoption and use of AI and machine learning technologies today (Q5).

But as knowledge of and experience with the technology matures, and as the number of applications increase, the informality will inevitably give way to more structure. And this shift could come in relatively short order as the perceived importance of AI and machine learning grows.

Interestingly, the survey suggests that a possible inflection point in that perception could come in the next couple of years. Today, only 12.5% of survey takers attach a “high importance” to the business impact of the technologies, but over the next two years, this group grows to 41%, a shift, should it occur, that would amount to a dramatic change in attitude (Q6).

Before that happens, though, manufacturers will need to work out some process issues as well as grow their own knowledge bases about the technologies. Right now, for example, fewer than one-third of respondents say their companies have a dedicated budget for AI and machine learning technologies (Q9). And just under 11% say they have a high level of confidence that their companies have the internal expertise to successfully manage and support deployment of the technologies. About 20% of survey respondents say that their software providers function as the primary source of support on AI and machine learning projects today, while just 16% say an in-house AI development team fulfills that important role (Q14).

Process Improvement Focus 

As might be expected at this stage of adoption, many of the anticipated benefits of AI and machine learning tend to center around improving existing processes. Just over 52% of survey respondents identify predictive insights and better decision making, for example, as “high potential” benefits of the adoption of AI and machine learning technologies.

Cost savings, at 45% of the sample, and better planning, at 43%, come in fourth and fifth in terms of having high potential. But respondents also seem to be thinking broadly about the possible business impact of the technologies. Nearly 48% selected increased competitive advantage arising from the technologies as a potential benefit (Q16).

Respondents’ assessments of potential benefits in specific functional areas also tend to focus around process improvements. In production operations, for example, the top three expected benefits are increased uptime of factory assets, production process innovation, and improved predictive maintenance of plant floor equipment (Q17). And in their supply chain operations, survey takers cited better planning, more predictive insights, and increased agility as their most desired improvements (Q18).

But before they can truly understand the effectiveness of AI and machine learning technologies in any area of their organizations, manufacturers will have to get better at measuring them. Right now, nearly 48% of respondents said they do not have metrics established to measure the impact of the technologies; encouragingly, 38% said they do.

[tooltip text=”Tooltip Text”]

7 Manufacturing, IT in AI Driver’s Seat

Q: Who is in charge of AI and Machine Learning efforts in your organization?

[tooltip text=”Tooltip Text”]

8 Much Headroom for Growth of Internal AI Expertise

Q: What level of confidence do you have that your company has the internal expertise to successfully manage and support AI and Machine Learning deployment?

[tooltip text=”Tooltip Text”]

9 Fewer Than One-Third
Have an AI Budget

Q: Does a dedicated budget exist within your company for AI and Machine Learning technologies, training, and education?

PART 3: STATUS OF AI ADOPTION

[tooltip text=”Tooltip Text”]

10 Awareness Building, Pilots Characterize AI Status Today
for Performance Assessment

Q: What is the overall progress level for AI adoption at your company? (Check all that apply)

[tooltip text=”Tooltip Text”]

11 Production Leads Corporate
Functions in AI Adoption
for Performance Assessment

Q: Which of the following corporate functions has begun the adoption of AI? (Check all that apply)

[tooltip text=”Tooltip Text”]

12 Nearly One-Quarter Implementing AI Projects in Factories

Q: What is the progress level of AI adoption in your plants and factories?

Characteristic of the early stage most manufacturers are at with AI, few companies have a formal strategy in place for the technology.

[tooltip text=”Tooltip Text”]

13 Process Improvement,
Planning Key AI Factory Applications

Q: What are the key application areas for AI and Machine Learning technologies in your plants and factories? (Check all that apply)

Adoption Challenges Abound 

Among the most significant and provocative challenges attending AI and machine learning are the effects that these technologies may have on the workforce. There is little question that there will indeed be an impact, perhaps even a dramatic one. But people in manufacturing, who have had to cope with skills shortages and the problem of unfilled job for many years, may have a perspective on the issue that is markedly different from those outside the industry who fear a dark future for the human race.

A powerful majority of survey respondents, 65%, does indeed believe that AI adoption will result in workforce headcount level changes by 2025. That number breaks down to 39% saying the impact will translate to a one to five percent replacement or reduction of their current workforces in the next six years. Another 18% expect the impact to range between five and 10% and nearly seven percent see a 10 to 20% impact. Fully one quarter see no impact at all (Q1).

But a noteworthy percentage of respondents, 60%, expect that those displaced will be retrained for other jobs. That number breaks down at about 18% expecting that one to five percent of those displaced will find other jobs, another 18% anticipating five to 10%, , nearly seven percent expecting an offset of 10-20%, and about 16% foreseeing 20% or more being retrained (Q2).

In addition to the workforce issue, there are a number of other significant challenges associated with the adoption of AI and machine learning technologies.

Chief among these are understanding the technologies, at 67% of respondents; understanding the business case for them, at nearly 56%; and data issues, at 53%. The need to upgrade legacy technology systems in order to use AI and machine learning, cited by nearly 49%, is also a substantial challenge for many companies (Q3).

And on the critical question of what impact overall AI and machine learning will have on the manufacturing industry in the future, an interesting but not unusual schism has occurred in the survey data. A majority, 53%, say that AI and machine learning, while significant, will not add up to a force so powerful as to transform what they do. On the other side of the isle, 39% do indeed see AI as not only a game changer for their companies, but also amounting to a new era of technology affecting the business (Q4).

MLC surveys on the impact of Manufacturing 4.0 have revealed a similar dynamic. Several years ago, survey data was pretty much evenly split between those who thought M4.0 was significant but not transformative and those who thought it was truly a game-changer for the industry. But those numbers have slowly shifted over the years toward the more imaginative view as experience and knowledge have developed about the potential of digitization.

Could a similar route be traveled by AI?

The Road Ahead 

The answer to that question will, of course, come with the passage of time, but, in the interim, those manufacturers who are trying to educate themselves about the technology, undertaking research, and even engaging in some pilot projects would be well advised to move ahead deliberately and with a sense of urgency.

Artificial intelligence is a force to be reckoned with. It will come at manufacturing from many directions and affect many functions within the manufacturing enterprise. AI will be part of many different types of application software products, to ERP and supply chain systems, to quality and maintenance systems, and customer-facing systems. It has the potential to be a pervasive influence on those systems, the processes supported by them, and job functions and roles. It could, as Roland Busch of Siemens said, take manufacturing to a new and better level. It could also cause unwanted disruption.

But it is not a force unto itself. People can and should remain in conscious control of deciding when to use it and how much to use of it. Like any technology, and certainly as we have learned with social media, technology can be used wisely or not so wisely.

The decision rests with us.M

[tooltip text=”Tooltip Text”]

14 Manufacturers Tap Broad
Array of AI Expertise

Q:What is the primary source of support for the development of AI & Machine Learning competencies in your organization?

[tooltip text=”Tooltip Text”]

15 Growth in AI, Machine Learning Investments Foreseen

Q: What level of increase in AI and Machine Learning investment do you plan, or expect to see, in your
manufacturing operations over the next 2 years?

PART 4: BENEFITS OF AI ADOPTION

[tooltip text=”Tooltip Text”]

16 Top 5 Potential Benefits Foreseen

Q: How would you assess the potential benefits of AI adoption for your overall business? (% of those indicating high potential benefit)

[tooltip text=”Tooltip Text”]

17 Top 3 Production Benefits Expected

Q: How would you assess the potential benefits of AI adoption for your production operations? (% of those
indicating high potential benefit)

[tooltip text=”Tooltip Text”]

18 Top 3 Supply Chain Benefits Desired

Q: How would you assess the potential benefits of AI adoption for your supply chain? (% of those indicating high potential benefit)

[tooltip text=”Tooltip Text”]

19 Most Lack Metrics on AI Effectiveness

Q: Do you use a specific set of metrics to measure
the effectiveness/impact of your AI & Machine
Learning deployments?

Opinions about the impact of artificial intelligence today range from the apocalyptic to the miraculous. Media darling Elon Musk of Tesla, for example, thinks AI is an “existential threat” to human civilization. Oracle CEO Mark Hurd believes a battle between the United States and China for “AI supremacy” will have important consequences for the global economy. And Ginny Rometty, IBM’s CEO, is convinced that AI has the power to transform industries in positive ways.

Whatever your view of AI, a term coined in 1955 by the computer scientist John McCarthy, the technology is at the forefront of discussions throughout society today, leading a debate about the future of work, jobs, and even what it means to be human. And as the manufacturing industry transitions to the digital era, AI is being viewed as central to leveraging the vast amounts of data that factories and plants will generate to do everything from improving operational efficiency to creating new, competitive advantages.

“Industrial AI can give the Fourth Industrial Revolution a huge boost and take Industrie 4.0 and similar initiatives to the next level,” said Roland Busch, Chief Operating Officer, CTO, and Member of the Managing Board of Siemens AG, in an article posted on the World Economic Forum’s website in January.

In an attempt to separate the hype from the reality of AI, and to take the measure of where AI and its cousin machine learning stand in manufacturing today, the Manufacturing Leadership Council undertook its first ever survey on manufacturers’ attitudes, plans, projects, and expectations with the technology earlier this year.

Chief among the survey’s findings is that, despite the hype, the 64-year old concept is at an early stage in most manufacturing companies. And while many companies expect AI to displace significant percentages of their workforces, they also anticipate that many of the displaced workers will be retrained for other roles in their companies, undercutting the notion that AI will inevitably lead to a vast wasteland of unemployed people. Moreover, a majority believes that while AI and machine learning are significant, they will not be transformative for the manufacturing industry.

Small Projects the Norm 

Digging deeper into what the survey data reveals about the status of AI and machine learning adoption, at an overall corporate level, 20% of respondents indicated that they are experimenting with a range of small-scale pilot projects in their companies and another 12% said single projects have been implemented. The largest group, 40%, are either in the stage of developing awareness of the technology, conducting research, or defining a roadmap (Q10).

The good news is that, over the next two years, survey respondents expect AI and machine learning investments to increase, in some cases substantially. More than 30% of respondents said they anticipate spending increases of between one and 10% in that timeframe, while 22% said 10-25%, and 14% indicated an increase of 25 to 50%.

At a departmental or functional level, manufacturing and production, with 60% of respondents indicating they have begun the adoption of AI, are the leading areas for the technology at present. Supply chain follows, at 30%, and research and development comes in third at 28%. But many other areas of the enterprise, from sales and marketing to quality operations, are also getting involved (Q11).

On the factory floor itself, 24% of survey respondents said they are implementing AI and machine learning on a single-project basis, while 48% are still going down the awareness, research, and roadmap trail (Q12). And among the application areas being addressed, process improvement, production planning, and preventative maintenance are getting the most attention.

A Lack of Formality 

As companies proceed with pockets of AI and machine learning activity, they are doing so largely on an informal basis, suggesting experimentation with single or pilot projects to address a specific need or opportunity. Only 12.5% of survey respondents said their companies have a formal plan and strategy in place for the adoption and use of AI and machine learning technologies today (Q5).

But as knowledge of and experience with the technology matures, and as the number of applications increase, the informality will inevitably give way to more structure. And this shift could come in relatively short order as the perceived importance of AI and machine learning grows.

Interestingly, the survey suggests that a possible inflection point in that perception could come in the next couple of years. Today, only 12.5% of survey takers attach a “high importance” to the business impact of the technologies, but over the next two years, this group grows to 41%, a shift, should it occur, that would amount to a dramatic change in attitude (Q6).

Before that happens, though, manufacturers will need to work out some process issues as well as grow their own knowledge bases about the technologies. Right now, for example, fewer than one-third of respondents say their companies have a dedicated budget for AI and machine learning technologies (Q9). And just under 11% say they have a high level of confidence that their companies have the internal expertise to successfully manage and support deployment of the technologies. About 20% of survey respondents say that their software providers function as the primary source of support on AI and machine learning projects today, while just 16% say an in-house AI development team fulfills that important role (Q14).

Process Improvement Focus 

As might be expected at this stage of adoption, many of the anticipated benefits of AI and machine learning tend to center around improving existing processes. Just over 52% of survey respondents identify predictive insights and better decision making, for example, as “high potential” benefits of the adoption of AI and machine learning technologies.

Cost savings, at 45% of the sample, and better planning, at 43%, come in fourth and fifth in terms of having high potential. But respondents also seem to be thinking broadly about the possible business impact of the technologies. Nearly 48% selected increased competitive advantage arising from the technologies as a potential benefit (Q16).

Respondents’ assessments of potential benefits in specific functional areas also tend to focus around process improvements. In production operations, for example, the top three expected benefits are increased uptime of factory assets, production process innovation, and improved predictive maintenance of plant floor equipment (Q17). And in their supply chain operations, survey takers cited better planning, more predictive insights, and increased agility as their most desired improvements (Q18).

But before they can truly understand the effectiveness of AI and machine learning technologies in any area of their organizations, manufacturers will have to get better at measuring them. Right now, nearly 48% of respondents said they do not have metrics established to measure the impact of the technologies; encouragingly, 38% said they do.

Adoption Challenges Abound 

Among the most significant and provocative challenges attending AI and machine learning are the effects that these technologies may have on the workforce. There is little question that there will indeed be an impact, perhaps even a dramatic one. But people in manufacturing, who have had to cope with skills shortages and the problem of unfilled job for many years, may have a perspective on the issue that is markedly different from those outside the industry who fear a dark future for the human race.

A powerful majority of survey respondents, 65%, does indeed believe that AI adoption will result in workforce headcount level changes by 2025. That number breaks down to 39% saying the impact will translate to a one to five percent replacement or reduction of their current workforces in the next six years. Another 18% expect the impact to range between five and 10% and nearly seven percent see a 10 to 20% impact. Fully one quarter see no impact at all (Q1).

But a noteworthy percentage of respondents, 60%, expect that those displaced will be retrained for other jobs. That number breaks down at about 18% expecting that one to five percent of those displaced will find other jobs, another 18% anticipating five to 10%, , nearly seven percent expecting an offset of 10-20%, and about 16% foreseeing 20% or more being retrained (Q2).

In addition to the workforce issue, there are a number of other significant challenges associated with the adoption of AI and machine learning technologies.

Chief among these are understanding the technologies, at 67% of respondents; understanding the business case for them, at nearly 56%; and data issues, at 53%. The need to upgrade legacy technology systems in order to use AI and machine learning, cited by nearly 49%, is also a substantial challenge for many companies (Q3).

And on the critical question of what impact overall AI and machine learning will have on the manufacturing industry in the future, an interesting but not unusual schism has occurred in the survey data. A majority, 53%, say that AI and machine learning, while significant, will not add up to a force so powerful as to transform what they do. On the other side of the isle, 39% do indeed see AI as not only a game changer for their companies, but also amounting to a new era of technology affecting the business (Q4).

MLC surveys on the impact of Manufacturing 4.0 have revealed a similar dynamic. Several years ago, survey data was pretty much evenly split between those who thought M4.0 was significant but not transformative and those who thought it was truly a game-changer for the industry. But those numbers have slowly shifted over the years toward the more imaginative view as experience and knowledge have developed about the potential of digitization.

Could a similar route be traveled by AI?

The Road Ahead 

The answer to that question will, of course, come with the passage of time, but, in the interim, those manufacturers who are trying to educate themselves about the technology, undertaking research, and even engaging in some pilot projects would be well advised to move ahead deliberately and with a sense of urgency.

Artificial intelligence is a force to be reckoned with. It will come at manufacturing from many directions and affect many functions within the manufacturing enterprise. AI will be part of many different types of application software products, to ERP and supply chain systems, to quality and maintenance systems, and customer-facing systems. It has the potential to be a pervasive influence on those systems, the processes supported by them, and job functions and roles. It could, as Roland Busch of Siemens said, take manufacturing to a new and better level. It could also cause unwanted disruption.

But it is not a force unto itself. People can and should remain in conscious control of deciding when to use it and how much to use of it. Like any technology, and certainly as we have learned with social media, technology can be used wisely or not so wisely.

The decision rests with us. M

Survey development was lead by Executive Editor Paul Tate, with input from the MLC editorial team and the MLC’s Board of Governors.