People have an innate fear of things they don’t understand. Artificial intelligence is just another tool that can have negative implications if not used properly, or is used with inherent biases, believes Alexis Hannah Smith, Founder and CEO, IMGeospatial
How did you conceive IMGeospatial and what was your vision when you set it up?
I conceived IMGeospatial like any good business, out of solving a problem. A water utility company in the UK, Anglian Water, needed a way of understanding the pluvial hazard of its infrastructure and asked me to look at it. In doing this, I realized there was a gap in the market in surveying large areas of the world and understanding change when it had occurred, cost-effectively. That’s where it all began.
My vision at the time, and now, is to enable organizations and enterprises to solve problems via intelligent use of data, across a multitude of sectors, not just water, but insurance, automotive, NGOs and more, with a focus on creating positive societal change.
Tell us a bit about how IMGeospatial came into being, its initial years and current business models.
IMGeospatial came about following the initial project with the water utility company in the U.K., Anglian Water, where we enabled them to understand the pluvial hazard of their infrastructure, cost-effectively.
When we first started out, IMGeospatial was at the forefront of extracting features and has developed and grown from there. Over a few years, the technology and our understanding of what we can achieve have evolved along with the data that goes into flood modelling and all types of data that’s now tied together with geospatial; digesting, distilling and disseminating the data pipeline for automated business intelligence.
Currently, we sell a variety of different data products which are outputted from clever tech to solve problems. For a water utility this might be around leakage, finding loss within a water system; for an insurance company it might be about understanding the underinsurance problem within their customer portfolio; and for NGOs geospatial data can enable them to understand change efficiently and effectively – not change as in a time series of earth observation data, but change on the ground; ensuring there’s no bias, that change with open data is understood and correctly adopted when it needs to be, and if there is malicious change that data is not adopted. Our work with the World Bank in Tanzania, understanding population density over scale, enables planners and people on the ground to do their jobs in a better and faster manner.
What is your leadership style?
My leadership style is empowering, I give people around me the power to make decisions. As Steve Jobs famously said, it doesn’t make sense to hire smart people and tell them what to do. My skill is getting the right people in the room and empowering them, which enables us to collectively achieve what we need to.
My leadership style is also rooted in how I approach solving problems. In some cases data that’s one hundred percent accurate may lead to the best academic solution but not necessarily the best business solution, they’re very different things.
IMGeospatial calls itself an automated business intelligence provider. Business intelligence relies on a slew of emerging technologies like artificial intelligence and machine learning. How do you differentiate between business intelligence and automated business intelligence?
The difference between business intelligence and automated business intelligence is the latter enables data to be processed with zero human intervention. The knock-on effect of this is significant savings in time and cost for our clients and their customers, which we’re hugely excited to be leading the way with.
IMGeospatial’s automated business intelligence is a completely automated data pipeline that can digest, distill and disseminate data without any human intervention.
With AI and machine learning, you often have a lot of people drawing boxes around polygons or doing certain things within the pipeline. These people may be looking at, for instance, earth observation data, sourcing data and manually loading it into a system, the system will then process it for people to manually take that data and do something with it.
Now, people might be scared thinking that the human resources element of our industry will disappear. But that shouldn’t be a concern because the amount of data that will be needed to create a digital twin is overwhelming. If you are trying to digest all the data for the entire world every day, you would need a huge amount of people to carry out quality assurance of that data.
You use data from a wide array of sources. Do you encounter any difficulties while collecting and analyzing these datasets?
We make computers develop an automated data pipeline with datasets, which means structuring data in a way that computers can use it and of course that’s not easy. A good example is the earth observation data. You need Cloud-free data that is normalized in a way that you can interact with certain places. Data needs to be used within a certain time period, it must be the right data in the right resolution and getting a computer to understand that is incredibly hard.
The cleverness in what we do is not in the processing bit. The cleverness is in finding ways to solve problems that generally humans face, like how to decide which parts of images to use and which not to, or how to decide when you need data and when you don’t. How do you get computers to decide all these kinds of things without human intervention? Now that’s clever.
How do you ensure cost effectiveness while using artificial intelligence, machine learning and algorithms at this stage?
Like geospatial, artificial intelligence is a sector and there are so many technologies that we use within geospatial, it’s the same with AI. We do a variety of different things in smarter ways to ensure that we don’t have huge overheads. What we can do is create our own datasets using our technology to help solve problems. It’s important how you structure data or input and how you show data on the output.
So, in the way we think and approach things, we can go up against, or at least stand next to some of the bigger players in our industry without the £75 million ($98.4 million) investment, by just looking at the problem in a slightly different way and tackling it in an automated manner.
Would the use of AI in drones boost accuracy of analytics and increase the role of drones in capturing data?
Absolutely, using AI in drones can definitely boost the accuracy of analytics and increase the role of drones in data capture; there are people already doing things in this area. It’s important to consider how to structure the data that the drone is capturing, ensure it is used with the right control points and then ingested.
When we did a program with the World Bank in Zanzibar, we found the issue was not about what we can do with data, but how the data was captured. There was a problem in the standardization of data, its interoperability and output. In situations likes these, IMGeospatial comes into the picture because we are very good at understanding what is really happening and forecasting what is going to happen.
What is the main deterrent to mass adoption of AI-based models and business intelligence?
People have an inherent fear of machines. When trains were invented, people were scared to ride them. It’s the same with AI, people have an innate fear of things they don’t understand. AI is just another tool, like a train or a plane; it can have negative implications if it’s not used properly or used with inherent biases. If you don’t have a diverse group of people with different gender, background, sexuality and ethnicity involved in the design, building and implementation of AI, then it will be biased. Businesses supporting diversity and collaboration, involving everybody and not being insular, is incredibly important for the long-term and the success of AI in general.
How do you ensure seamless delivery of high precision data analytics in your land use classification tools that are primarily built for the insurance sector?
A lot of hard work goes into ensuring seamless delivery of high precision data analytics across all of our products. We are constantly monitoring our performance, identifying areas for improvement and applying learnings as we move forward. A good example of this was a project where we had to identify leakage within the water system for a company and our automated system appeared to be confusing how the data was being digested. It occurred because we were picking data from irrigation networks that the water company wanted to know anyway, which was a scenario that we learned from.
What are the opportunities and challenges facing the business intelligence sector?
There are huge opportunities within the business intelligence sector. What’s really exciting is people from different sectors being involved which casts a fresh perspective on our outputs. We have a diverse group of people at IMGeospatial with different backgrounds from different geographical locations working in a collaborative way, together and with many other businesses.
I had some very interesting conversations at the Geospatial World Forum 2019 about how we can use our intelligence with a Trimble device or Esri software. Instead of working against each other, if we start working together and trust each other a bit more, we might be able to deliver and solve some of the biggest problems in the world. That is what excites and motivates us at IMGeospatial.
There is a huge mismatch between the pace of profusion of data and the pace at which data analytics capability is getting refined. Then there are soaring customer expectations. What is the way around and the way forward?
We are doing well in our ability to digest, distill, disseminate and use the data; the issue, really, is people not fully understanding how all of this can be used. In the same way, like we have had a digital transformation in the past 20 years wherein people started to use computers and integrate things, they should be doing the same from a geo-transformation point of view; that is, looking to benefit the industry at large, across multiple business sectors, rather than for solving one problem. This would result in a massive increase in business efficiency and lead to huge cost savings. Now, whether it is a standard GIS system or something more disruptive like what we are doing, or perhaps a collaboration of these things, remains to be seen.