The AI Fallacy in Industrial World Applications
In 2019, Oceans.ai embarked on a journey of exploring the world of AI for industrial applications. We specifically began the journey around computer vision and its application for industrial inspections. Now, you would be inclined to ask (and rightly so) “why industrial inspections”?
The answer is a simple one and a real problem in the industrial world. Industrial inspections are the most manual (and hazardous!) and amongst some of the least digitalised process in the industrial world. Several industries must shut down their plants for weeks or even months to proceed with industrial inspections. I was recently talking to a C Level executive who indicated that his plant shuts down for two months in a year for handling inspections. TWO MONTHS!!! This is a waste of time. This is very costly, of course. These losses can vary in % and number from sector to sector, business to business and geography to geography depending on shutdowns and wait periods and erroneous finding leading to sanctions and compliance expenses that follow, for instance.
Industrial inspection is a very complex field. In chemical plants, offshore oil and gas assets, remote energy and renewables infrastructure, inspection is not only about financial cost, but also about human life. These industries are hazardous. They put a very high risk on the human workforce. In many others, it is tedious and expensive but necessary (with mobilisation, weather hazards and other safety and uncertainties putting a huge toll on the process). In few of the cases, it is loss of business and revenues too. The list goes on. But you get the idea. Things must be done better. Industries have recognised this and are embarking on various journeys to robotize. Artificial Intelligence (AI) enables the process.
AI fallacy
Herein lies the reason why this article is called “the AI fallacy”. AI is part of the answer. There is no debate about it. But there is a right way and a wrong way to go about it.
A 2020 MIT Sloan/BCG survey found that about 70% of companies in all sectors had piloted or deployed AI solutions. Of these firms, only about 10 % reported significant gains from AI projects. This is disappointing. It also went on to say that “organization need a multidimensional, complex relationship with AI—one that involves several methods of learning and different modes of interaction”. We believe one of the key reasons for this abysmal rate of failure is the approach being taken to AI learning and more importantly the way data is processed and by whom.
Industrial inspections data is nowhere like having data scientists identify a car or a human or a bird. It requires years of expertise and knowledge and experience of the industry, the asset, the environment, the science, and the insight to be able to discern the defects. This comes from years of knowledge and experience. Unless industries learn to capture this experience in AI, the success of an AI implementation is going to be limited.
Just like human intelligence, AI is not a magic wand, where there is ready AI kit that can deliver the AI solution overnight. Just like human intelligence, AI is something that grows over time. There is a context and a continuum to the learning process. The gathered data and intelligence happens over time with the right guidance and corrections and inputs from experts. If industries need to see the benefit of AI, they need to deploy the right engines and the platform that will power the learning process and will empower the right experts to improve the learning process.
The workforce is the greatest asset
In our view, one of the key factors that defines a successful AI effort is data science. Feel lucky! The biggest advantage the industry has is it already has the required data scientists for the AI JOURNEY! This is his workforce ! Industrial workforce has been working day-in-and-out on industrial assets. They know the ins and out of the assets, the environments, and the risks. They know the why, the how and when of every event and very often they are right here in the midst of the industrial asset, putting themselves at risk and fully aware of the unfolding problem.
But the workforce is busy doing their day jobs. They can’t be bothered to be taking time to do data science. That’s where industry can build in the infrastructure where the workforce is empowered with tools to and solutions that will help them do their existing task better while using AI data science process. This is more efficient. This is time-saving.
In AI, we often hear the phrase “Human in the loop” process of AI learning. This is not how we see things at Oceans.ai. We have flipped this to call it “AI in the loop” process. This is where the AI slips into the existing business process and delivers powerful data science for the AI learning process so that the continuous AI learning improves over time. AI does not replace human experience and knowledge; it is here to facilitate work life and enhance results.
Inspections and plant surveys are some off key areas of big impact for AI. Industries do inspections either manually or in some cases using drones/robots. Inspections are a classic case of how the workforce and experts from across the globe (inspection companies, standard organisations etc.) sitting within the business process can power the AI learning process for the industry. We are now working to add this technological brick at Oceans.ai.
It is imperative that industry thinks of AI not as a tool, but an integral part of their business process transformation. It is not without reason that AI is being called the fourth revolution of industry (Industry 4.0). This transformation is not going to happen overnight. Industry will need to put in place systems, processes, and workflows in place to make sure that the business, the industry and most importantly the workforce adapts and adjusts itself gradually to the transformation that is most certainly coming. AI explain ability and acceptability are key for long-term success.
Safe travels,
Vinod
Vinod Govindan, Founder and CEO of Oceans.ai.
About Oceans.ai
Oceans.ai is a transformative digital solutions company that is building digital solutions for the industrial world, enabling them to systematically and methodically adopt solutions and technologies to power their AI transformation journey. Visit oceans.ai at www.oceansai.tech or write to us at sales@oceansai.tech to learn more about their solutions and how their offerings can help.
Leave a reply