You Haul: Complex AI systems are tough to create, hard to integrate with domain expertise, and even harder still to maintain without core-competency
Oui* Haul: With a nod to the Petro.ai Montreal and Houston Development Team who spend every hour of the working day and many weekends updating, building, and adding the new to make a stand-alone AI platform for O&G
How does a complex platform like Petro.ai keep their Artificial Intelligence, or AI, evergreen? The concept of being lasting or evergreen is a statement about remaining integral to business processes. Businesses are always changing and the Petro.ai platform is geared to stay relevant to the O&G industry.
Dr. Derek Ruths, Chief Data Scientist of Petro.ai who oversees the development of the AI capabilities and their incorporation into the Petro.ai products, platform, and product suite explains how he and his team keep the software abreast of cutting-edge machine learning research.
Evergreen Applications
“I think of Petro.ai apps as having a couple of different pieces,” Derek explains. “There’s the principle of the app which is what the app is for.
If you figured out a particular O&G problem, such as well spacing, that app principle or experience doesn’t need to change.
The app itself can remain evergreen because if the problem was picked correctly and that remains a problem, then people keep using the app in the same way.
“What does change is what clients should expect from the performance. Making an AI powered product evergreen is about choosing the problems correctly up front and then you won’t have to pivot later. With the app contextualized appropriately, what you’re doing with subsequent releases on that product is tuning and tweaking and improving as more and/or improved training data becomes available.
New AI Architectures
“Evergreen will always mean exploring new architectures as technologies become available. So, going from convolutional neural networks to transformers to who knows what the next thing is going to be, we will continue to keep trying out these architectures to get the best performance.”
Better Data
“Evergreen is always coming up with better data, data that is more representative of real-world scenarios, and better labels as you understand the use case in greater depth and detail.” Dr. Troy Ruths, CEO of Petro.ai weighs in on this important aspect of evergreen, “I also think data inputs change. Businesses are continually capturing new data. They have a vast number of digital initiatives going on. The most common thing you want to do once you get a new data feed is to use it. And one important way is to send it to an AI and have an app take that in. When you have a new data feed, incorporating it into an AI app will make the data more valuable by improving the prediction accuracy and reducing the error bounds on it.”
Troy continues, “It’s easy to get behind in this never-ending updates, new data feeds, and premier AI technology cycle. To build and maintain AI powered tools requires a blending of domain and machine learning expertise, an understanding of technology direction, and a knowledge of where the high-value options are.”
Forbes emphasizes this point, “We’ll see enterprise customers moving away from building their own machine learning platforms, recognizing that it’s not their core competency. They’ll realize that more value comes from using tools designed with ML applied to business problems versus spending the time to build and maintain the tools themselves.”
Let’s face it. Not everyone understands AI. It’s full of a jargon and methods that are vastly unfamiliar to most. But it’s not going anywhere. AI is here for the long haul. Infoworld reminds us, “Staying abreast of AI trends, technologies, and applications is fundamental to success in modern business, even if you’re not a data scientist or machine learning specialist. Companies that innovate with AI will dominate their industries for decades to come.”
*Oui—In French it means ‘yes’ as in ‘yes we can.’
