“Contrary to perception, AI is not a robot replacing a person; it helps to release the person, to get more of them thanks to the support of the robot”.
National Apprenticeship Week is all about #skillsforlife. With the potentially seismic change to the nature of work upon us due to AI and tech innovation, it’s a timely moment to think about how insurance could evolve.
Over to Stevie…
So Stevie, tell us about your role and how the apprenticeship came about
Stevie said:
"I work in Pricing and the Risk Modelling team. We build the technical rates sitting in the background that eventually become the live prices.
"A lot of the team choose to do the actuarial exams and study, and there is support for that, but it’s not where I wanted to go. I felt that there was a bit of an absence from the machine learning side in terms of studies, as it’s a very new area. There also isn't a formal degree at a university because it's both so new and so quick-changing.
"I spoke with Nigel Carpenter, Chief Data Scientist at RSA, and my manager to decide between the actuarial route or something a bit wider. I came across this apprenticeship, which is Masters Level and covered through the apprenticeship scheme, which helped a lot with funding.
"I’ve been doing it since October and have just handed in the first assignment: a 3,000 word task where I had to build an artefact example, containing dashboards for MI, propose a new IT solution, and then write about the impact and uses of AI within that.
"It’s been a long time since I’ve written assignments, so that was a bit of a shock to the system.
"The online workshops are really great and there is a mixture of people on the course. Some people straight out of university, to an IT director who wants to know how to bring AI into his business and develop an overall AI strategy. Unfortunately, there aren’t many women on the course – I would say 4 or 5 out of 20. Although this is better than when I studied astrophysics at university, where there were only 5 women out of 200. So, there is some progress there."
Did you anticipate yourself taking on this apprenticeship 10 years into your role at RSA?
Stevie said:
"Not really, but I was trying to expand my knowledge. Across the team there’s so much experience with opportunities to expand on our data science, notably since RSA joined Intact. There are over 350 data specialists and scientists in data science at the Intact Lab, so collaboration and knowledge-sharing are present. It felt like a good decision.
"On top of this, I wanted to find something more formal, structured and with an in-depth theoretical element that complemented my work tasks. It brings in this fascinating extra layer, explaining why an IT system has been built in the way it has and why the data appears as it does. I think I'm the guinea pig, but I know other people are keen too."
In what ways does your day job involve machine learning?
Stevie said:
"In risk and pricing, we’ve already built machine-learning models and implemented a few. The data science team already has some learning modules available and it's very much where the insurance market is going.
"In terms of the machine learning that's done in insurance, there is always room for enhancements, notably because of the speed this field is evolving. For quite a few years now we've been able to get elements of machine learning and use it for analysis, but we’ve been limited by the systems.
"AI can be used in a lot of different ways in insurance as a whole. You can use it purely on the risk model trying to predict the likelihood of a claim. You could use it on optimising the actual price, trying to maximise your top-line figures and your premium, retention, looking at fraud behaviour and trying to identify multiple fraud risks."
Had you noticed limitations in your day job relating to data quantity/volume/speed, and what you were needing to do?
Stevie said:
"Certainly on the data side, yeah, that's quite common. It’s really important that the data is complete and ‘clean’ because the pricing community is so reliant on data and systems. RSA as a whole is working to find ways to build in machine learning across the ecosystem. So, it felt like a very good time to start getting to get into it."
What excites you about AI, and equally what scares you about it?
Stevie said:
"What excites me: it opens up a whole new world of possibilities. You still need people to model data and then to produce the risk models. It means you can spend a lot more time reviewing the model and a lot less time physically building it, so it increases the room for applying human expertise on top.
"Contrary to the perception, it’s not about a robot replacing a person, you’re getting more out of the person thanks to the support of the robot. There's an awful lot of general day-to-day bits that could be automated, making the job a lot easier.
"I think I’m also rather excited by the general newness of it as well. The things we can do now that we couldn't before, and some of the things that I've seen people build.
"In terms of underwriting wordings, the potential for AI to make it more consistent also has the potential to do a lot of good and simplification.
"What concerns me about AI in general: there will probably be an instance, and I'm sure there already has been, of somebody putting the data in, taking the model at the end and not thinking about it. The room for mistakes and unintended consequences is high and, without a human watching out for this, it could happen easily.
"I worry that AI could focus on something it shouldn’t or misunderstand subtleties around ethnicity or religion. To this end, Intact has a committee that oversees how the business leverages data and models to create and offer the best solutions for the customer, while treating customers and their data with respect, integrity, and the highest degree of ethics.
"I also worry a little about the regulation side. Europe and the UK are approaching it in different ways to try to mitigate the risks and build in restrictions, but I worry those restrictions could hamstring the efficacy. It could be they almost dial it right back and nobody can use it to full effect, or they'd let it go completely free and people could misuse it. So, I think that's one of the biggest questions around the area at the moment.
"I’m also wary about groupthink and decision makers who might not have the expertise, or the time to dedicate to the nuances or probable outcomes of AI use. For example, there was a piece in the news recently about a company that used AI for a chat bot and the AI wrote a poem for a customer and then swore at them."
AI: excelling in arts and music
Stevie said:
"There are certainly some weird uses of AI outside of the industry. It feels like it should be doing more supportive or clerical work, but it seems to be making art and music now which feels the wrong way around.
"There was someone who asked AI to make an image of a bunny get happier and happier to the point where you see the rabbit transform into a new time and space of ecstasy - an infinite, joyful singularity of rabbit. It's amazing."