Unlocking the Potential of Artificial Intelligence: What Manufacturers Need to Know

Share:

10/10/2024

Artificial IntelligenceThe potential for AI to help transform the way companies operate and interact with customers is almost too enormous for some people to even grasp. To begin getting a handle, companies obviously need to be aware of the types of AI solutions that are available. At the same time, it’s important to acknowledge that unlocking the potential of AI requires a coherent strategy and clear objective.

“AI is just another technology tool, albeit a very powerful one. The more you can recognize that and break down the barriers and silos of using it, the easier it becomes,” said Taryn Kutches, Vice President of Brand and Business Development at AEM member company Twisthink, a strategic innovation firm that specializes in partnering with B2B organizations to create digital products and solutions.

The Three Types of AI

Kutches co-presented at an AEM Member Education Webinar on AI on Sept. 12. She pointed out that AI first emerged back in 1962, though most of the advancements have taken place over the past decade or so. So, what exactly is AI in the year 2024?

At its core, AI is still the ability of a machine to perform tasks and cognitive functions commonly associated with humans such as perception, reasoning, learning, and decision-making. But AI is not this singular thing today. Rather, AI is a multi-layered discipline.

The top layer is machine learning. “This is all about training datasets,” Kutches said. Datasets are typically smaller and focused on delivering a specific outcome. A good example is a camera on a vehicle or piece of equipment that senses other objects.

The next layer is deep learning. This form of AI is still focused on doing one task, but larger datasets and more complex scenarios are involved. A good example is how Netflix uses a variety of user data to make recommendations. Another example is how equipment telematics data is leveraged to make preventive maintenance recommendations.

The deepest level of AI is known as foundation models, which is where Generative AI fits. This model is trained on broad datasets, enabling it to perform multiple tasks and even generate new information that mimics the patterns of the training dataset. This is where much of the AI hype has been generated over the past couple of years. We’re talking about AI tools such as ChatGPT, Google BERT, and others.

Three Factors Guide the Use of AI

Kutches said companies should examine three areas to begin setting their AI strategy.

Stakeholder. Do you want to use AI to create value for customers within your products, services, or experiences? You could also seek to create internal value through cost reduction, attracting talent, etc.

Outcomes. What are you hoping to accomplish from using AI? Do you want to gain insights from data to improve decision-making? Do you seek to automate certain functions of the equipment you manufacture?

AI type. Which type of AI will best serve your desired outcomes? As Kutches pointed out, Generative AI tools like ChatGPT and MidJourney (image generator) are powerful and exciting to talk about. But more traditional AI (machine learning and deep learning) has enormous potential, too.

“Now that many companies are getting smarter about the data being collected by their machines, on their jobsites, and in their manufacturing operations, traditional AI is still extremely valuable and powerful,” Kutches said.

 


Taryn Kutches

“AI is just another technology tool, albeit a very powerful one. The more you can recognize that and break down the barriers and silos of using it, the easier it becomes. -- Twisthink's Taryn Kutches

 

Three Steps to AI Implementation

Once companies work through the three key facets of AI strategizing explained above, Kutches said there is another framework they should work through: 

  • Define the Need
  • Define the Data
  • Define the AI technology solution and UX (user experience)

According to Kutches, some companies make the mistake of skipping ahead to the third step. But you can’t even begin talking about AI until you determine what you’re trying to accomplish.

With respect to the data that’s needed, Kutches said a company may already have it. In other instances, it may have to seek and acquire data. Sometimes it’s a combination. Regardless, Kutches emphasized the importance of having “clean data” that is accurate and reliable.

At this point a company is finally ready to begin talking about an AI technology tool itself, whether that proves to be Generative AI or a more traditional AI.

A company called Mercator.ai utilized this type of framework when it developed its AI-powered tool in 2020. Mercator.ai transforms real-time construction project data into early opportunities for construction companies. The company knew it can be difficult to find viable construction opportunities early enough in the process to have an influence. So Mercator.ai defined the desired outcome as reducing the lead-gen time from the typical 6-8 weeks using old-school methods, to just minutes using an AI-powered tool.

“From there we started to discuss what the data would need to tell us in order to achieve that outcome,” said Chloe Smith, CEO and co-founder of Mercator.ai, who also presented on the Sept. 12 AEM webinar. “Then we started to develop some metrics around the desired outcome. These are critical steps to take before even talking about what data you need.”

When the time came to discuss the data, Smith said extra care was given to ensure the reliability of the data source.

“For us, that meant we would only pick up (construction project) data that was reported,” Smith said. “So we worked with municipalities, regulatory boards, and private entities. We made a decision to not cross a certain line in order to ensure a high level of accuracy in the data.”

Another consideration is how your data is going to be collected. Who owns it? How often is it updated? Is there an API that will allow you to bring the data into your own system? Furthermore, how is the data going to be housed once you have it? As Smith pointed out, this can prove to be an intensive architecture project that warrants a lot of careful planning upfront.

Options for Implementing AI Tools

Once all of the upfront work is in order, including the identification of the best AI type for your use case, there are different ways to go about implementing AI:

  • Develop your own tool in-house
  • Work with an external developer
  • Utilize an “off the shelf” solution

Building your own AI solution obviously requires an in-house development team. On the plus side, Smith said, that means you maintain full control over what is being built—along with the tool’s adaptability over time. When going the in-house route, Smith said it’s vital that company leadership take ownership. There will likely be some additional hiring and overhead costs associated with this approach. If a company is already skewing more toward becoming a “technology company,” in-house AI development starts to make more sense.

When working with an external developer, the big benefit is that you’re getting expertise and experience. Smith said this can be a good approach when a company has been deliberate in its strategic planning and has a good understanding of its needs.

A third option is to simply purchase an AI tool from an established software vendor. On the positive side, the tool is ready to go and the vendor will typically assist with implementation. On the negative side, a company has no control over how the tool works. Furthermore, if a company’s needs change later on, it may have a tool that no longer meets those needs.

As you can see, there is a lot to consider when seeking to unlock the potential of artificial intelligence. There is also a lot of research and planning that needs to be done. But when a company is deliberate and follows the proper framework, it can turn all of this talk about AI into action, and move past the hype into actual results.

About Member Education Webinars

AEM members have exclusive access to help them stay on top of emerging issues and trends via member education webinars. Experts break down industry issues and pinpoint critical changes in the landscape to help attendees refine their company’s strategy.

For more information on the upcoming series of member education webinars, contact your Account Success Advisor.

Subscribe to the AEM Industry Advisor for more AEM news and updates.

AEM Updates, Technology

For more AEM news and updates, subscribe to the AEM Industry Advisor.

Related Articles

Smart Farming: How Data Analysis Can Help Increase Crop Efficiency

Producing more with less. This is the main objective of the agribusiness industry. The scarcity of resources and skilled labor in the field, in addition to the high costs of...

Unlock Opportunities and Take Your Business to the Next Level at CONEXPO-CON/AGG 2026

By Nicole Hallada, AEM Senior Vice President of Exhibitions and Marketing —Twenty-four hundred exhibitors from 36 countries descended upon Las Vegas in March of last yearto...

AEM’s Celebration of Construction on the National Mall Returns in May 2025

AEM, its members, and construction industry partners will once again take over the National Mall in the heart of Washington, DC, this coming May.The second Celebration of...

U.S. Sales of Ag Tractors and Combines Drop in November 2024

According to new data from the Association of Equipment Manufacturers (AEM), U.S. sales of Ag tractors and combines fell in November 2024 compared to the year before.U.S. sales...

Understanding Generational Differences in the Age of AI

By Gregg Wartgow, Special to AEM —When it comes to different generations’ use of technology, it might be instinctive to assume that older people want little to do with it while...

View all AEM Updates