We caught up with Director of Innovation, James Lowell, to kick off our monthly meet the team series.
In this series, you’ll meet team Geoteric, the faces behind the technology as we explore what it’s like to work at a leading AI seismic interpretation software company, ask experts about industry trends, and share insight into what satisfies each person in their role.
First up, James chats to us about Geoteric’s history in medical image processing, why continuous collaboration between internal teams and wider industry is essential to ensure software evolution, and what motivates him in his role.
Did you think your degree in medical image processing would lead you towards an organisation like Geoteric?
Originally, I thought I would take up a medical profession related to cancer research, a career path most common when completing a medical image processing degree. At the time of my studies, I hadn’t realised that the challenges faced within seismic interpretation, were similar to those found within many medical imaging processing problems. In both domains, you are searching for anomalies in often noisy and highly variable data.
Techniques I developed through medical image processing have been completely transferrable to energy so to answer your question, probably not at the time of my studies but it’s been a very natural progression.
Before you joined the team, did you know much about the energy sector and seismic interpretation?
When I joined Geoteric, the traditional energy sector certainly had, and continues to have a high profile. I believed that working within the oil and gas industry would be both exciting and challenging due to the nature of the problems that required solving. I had a baseline understanding and interest in the energy industry but at the time of joining, didn’t know much about the subtle nuances of seismic data.
What attracted me to Geoteric was its background in medical imaging. Our software was used to analyse medical imagery for features that could indicate disease. Before I joined Geoteric, the company was approached by a large, multinational energy company to see if our technology could be adapted to analyse subsurface data. That’s when the Geoteric as we know it today was born.
What do you find most fascinating about this line of work?
We’re always pushing the boundaries; we’re continuously moving forward.
No day is the same. There’s always a challenge, which I love. As a team and company, we do not stand still. If you take the advancement of Geoteric’s technology and the size of our company, which is still classed as relatively small, and compare it to that of much larger organisations – our technology is considerably more advanced.
You are meeting someone for the first time, and they ask what you do. How do you describe your role?
Put in simple terms, I would say I work in a research team, brainstorming with talented individuals and leveraging artificial intelligence (AI) to help understand the earth’s subsurface.
What does a typical day at work look like for you?
Collaboration with the wider research team is a large part of my day to day. We conduct group brainstorming sessions on individual projects to identify any bottlenecks and how we overcome these. It’s an interactive, fast-paced environment where ideas are shared and solutions born.
Another part of the job is generating initial storyboards to capture new concepts. Once we have proven a technology is effective, we outline how it could be used in a workflow. This storyboard is then handed over to a development team to create the commercial software.
Most of the work we do within the research team is prototyping and proof of concept, just to crosscheck that the idea can work in the field across a range of data sets.
In your opinion, what’s the most important thing that Geoteric can offer its customers?
We’re agile. We can understand an industry problem from the client one day and very quickly define and build a proof of concept solution.
Going from initial concept to proof of concept can be an incredibly short amount of time. Of course, we must go through the full development lifecycle to get the concept into the software, but this speed of turnaround is what sets us apart. It’s what makes us unique and keeps us at the forefront of our customers’ needs.
You’ve been with Geoteric for quite a few years – how has the company's offering evolved?
There’s definitely been a shift in Geoteric’s offering, starting out as medical image processing, transitioning through (non-neural network based) machine learning and now evolving to deep learning. We spend a significant time developing novel deep learning architectures that are purpose-built to help interpreters understand the subsurface – whether that be faults, horizons, or geological features.
As a result of our vast experience in image processing, our software has exceptional processing capabilities and sophisticated techniques that are still at the forefront of the industry. These can be used to clean up data and highlight areas of interest to an interpreter.
How do you anticipate what Geoteric’s customers need?
There are several checkpoints we go through with our customers. Our sales team are in constant communication with them, identifying what they’re struggling with in data sets and what they want to see from our software.
Within Geoteric, I run an industry consortium, comprising of different oil and gas operators. We have open dialogue where we explore ideas, discuss early prototype feedback, and share insight into technology gaps. Ideas from these consortium sessions will often be developed further into proof of concept by our research team.
Our geoscience team also actively inputs into the development of our software, leveraging decades of expertise to suggest how we can advance the technology to make it even more useful for other geoscientists in the field. Their input is incredibly valuable because they’ve been involved in the transition from manual interpretation to AI over the last three to four years. They have a great understanding of what’s possible and what will be well received by those who are like-minded.
What about customer needs for the future?
AI is here to stay. We’ve proven that Geoteric AI can detect geological features in seismic data that may be invisible to the human eye, however, we believe that seismic interpretation is at its most powerful when human and artificial intelligence are combined.
There needs to be a seismic shift (pardon the pun!) to how interpreters think and work to achieve the best interpretation results. Once they embrace AI and trust it can do a good job, they can focus their knowledge on the more value-adding areas that need their expertise. Together, AI capabilities and human expertise and experience are a winning combination.
As we move forward into a digital era, we’re already seeing this shift in thinking. Operators used to be skeptical but after running pilots, they have incorporated Geoteric’s AI technology into their portfolio and are now using it on every data set. Geoteric’s ability to accurately interpret faster, removing the need for the geoscientist to do this manually, gives the geoscientist more time to concentrate on difficult areas which are less clear.
By freeing up their time, geoscientists can dive deeper into the subsurface and broaden their understanding of what’s beneath. Operators spend millions of pounds acquiring this data, so you really want to gain the best understanding of its geological features as possible.
And what do you think the future will look like for AI interpretation software? I think we’ll be continuously building on where we are now and we’re currently creating the foundation to do this. We have separate networks to define and extract different parts of the subsurface. Whether that be faults, horizons or multiple geological objects such as channels – they can be found in the data and extracted.
These different networks work separately to each other with little awareness of each other. The next step is for these networks to communicate to unlock heightened insights and accuracy.
A hot topic at the moment is ‘explainable AI’ – a concept where the software is able to feedback a narrative as to why a decision has been made. You can train networks to describe their thought process in a written form. The software can share why something has been identified as a fault, for example, because there is a significant breakage in signal on both sides.
This explanation would give additional confidence to the end user. Teams can analyse this written narrative and form their own opinion of it. For example, the operator may decide that the decision isn’t aligned with their thinking, or that they want to zoom in further and show other areas of interest using the same identified characteristics.
In general, what are attitudes like towards machine learning and AI? Is there a degree of scepticism– or acceptance that it’s been game changing for so many industries?
AI is like everything else – you must apply it to really understand it. Until people do that, judgement shouldn’t be passed on its abilities. There are many companies who haven’t yet tried it and could be potentially missing out on remarkable results.
Once you see it and compare it to manual interpretation, trust is gained. As AI becomes the norm in the energy sector and new digital ways of working are used, more companies will perform this test and become comfortable with the technology. Generally the comparison between manual methods and AI assures companies that what they’ve done historically is broadly correct, but they also see there is more useful relevant information captured in the data than they expected, it also gives them the confidence to apply AI to new data sets they haven’t interpreted yet.
We have multiple available networks within Geoteric AI, all of which gain a slightly different perspective. Where all networks agree, you gain an exceptional level of seismic understanding. Where they don’t agree, interpreters can zoom into these areas of dispute and efficiently spend time solving questions – what humans do best.
What makes you proud about the work you do?
The collaboration in our team. Every idea, no matter how crazy, is always explored. Even if the idea doesn’t necessarily progress, it will often spark inspiration for another. We’re all coming together to genuinely drive the technology forward.
In other companies, ideas tend to be held on to by individuals for credit. That’s not how we work. We’re a unit that succeeds together or fails together.
This approach is reflected in our seismic patents too. For a small company, our patent portfolio is extensive. We make sure that we credit and champion internal inventors who nurtured those ideas, capturing their names in the patterns as they are filed.
Tell us something about yourself that would be a surprise to most people.
I’m the undefeated chess champion of Zanzibar! (I think…)
When travelling along the coast of Africa, I stopped off at a lodge in the region. There was a man there who told me he was the undefeated chess champion. I dabble in chess, so we had a game and I won. If he was telling the truth that he was this champion, then that title has passed on to me!