Abstract growth curve

Using AI to train junior talent

If you’re familiar with machine learning, you’ll know that we train algorithms to perform tasks. Well, what if trained algorithms could also help train people? We’ve been doing it with games for decades—who hasn’t played chess against their computer? And now, massive advancements in the field allow professionals in Go, probably the oldest and most complex strategy game (we’re talking 10^170 possible board configurations here), to use algorithms like Leela Zero to analyze their games and improve their strategy.

Granted, most real world challenges are more complicated than a game of chess or Go. Many data types come into play: structured, text, image, time series, etc. That’s why we need to combine human with machine, and in some cases, each can facilitate training the other. Let’s take a look at how you might apply this in a facility management setting.

A facility management use case

In a piping and instrumentation diagram (P&ID), related instruments like fluid level gauges and pump controllers are looped together because they work together. If you’re getting ready for Industry 4.0 by creating your ISO 14224 asset hierarchy, and your P&IDs are already digitized and “smart” (essentially, all elements like instrument tags and process flow lines have been converted into a structured format), you’re golden. However, if you’re in the process of extracting and analyzing these unstructured scanned P&IDs, you’re likely using AI to expedite this tedious exercise. Looping related instruments is one task where it can help.

Instrument loops captured in Cenozai's P&ID app

Cenozai's Asset Hierarchy module can automatically detect instrument loops in a P&ID, while allowing engineers the flexibility to make adjustments

Now, the algorithm is probably going to make some mistakes. Since the asset hierarchy is the foundation for your entire maintenance plan moving forward, you need to bring a human into the loop to correct any errors. This person needs to understand the underlying process logic, so a mechanical  or corrosion engineer is right for the job. It doesn’t make sense from a business perspective to ask one of your senior engineers to do it, though. You need them to spend time on more critical and complex tasks. Instead, set freshly hired junior engineers to work. Your senior engineer can allocate some time to guiding them initially, and then allow the remediation process to help cement their understanding. Practice makes perfect, right?

On top of that, they’ll also gain insight into the algorithm’s shortcomings. While it’s not necessary for your engineers to actually know how to design and train machine learning algorithms, they need to understand the possibilities and limitations of this technology. As businesses move into Industry 4.0, domain experts will need to be able to manage and integrate machine learning technology into their systems. Nurture your young team early: Millennials’ affinity for technology may  prove to be a valuable asset for ensuring your digital transformation initiatives succeed.

Preserving scarce resources

In some fields, however, nurturing young talent is not even an option. For instance, there is a shortage of paleontologists—experts are retiring and few young people are pursuing it as a career. Yet, this expertise is vital for oil and gas exploration.

In certain cases, we can also use trained algorithms to help bridge knowledge gaps where expertise is scarce. Machine learning can’t replace a paleontologist, but it can help preserve knowledge and improve the efficiency of existing experts. For example, you can use computer vision algorithms to classify foraminifera. It’s a small piece of a sophisticated job, but valuable when resources are so limited.

Maximizing AI investment

Machine learning has made amazing progress in the last decade, and you can capture more value from it than what you see on the surface. Companies can improve ROI on their AI investments by incorporating upskilling exercises with straightforward use cases, like remediating machine errors to build both domain specialization and a big picture understanding of AI tech. In doing so, you will move tech and talent forward simultaneously.