When Steve Horton first started work, he used to scan paper files — something he remembers as being a huge duplication of his time.

It’s perhaps why, now, the president of Liberty Specialty Markets could be called something of an artificial intelligence evangelist.

Because for him, AI, which he sees as potentially a fourth industrial revolution, will make his industry not only much more efficient, but used properly to analyse data, it will also help it design new products and write more business.

Efficiency, Mr Horton believes, is one of the key issues that AI can tackle, reducing duplication, streamlining and speeding up systems.

“What does efficiency mean? It means you can hopefully write more business and generate more return,” he said.

“Then with AI, if we have access to much quicker research, theoretically better research, that will open an opportunity for new products,” added Mr Horton.

While efficiency might be an upside of AI, companies still need to focus on the downside — the risks — as well as challenges to the efficacy and adoption of AI, including data.

On risk, Mr Horton said: “If I go to the consumer first, because I think it’s something everyone can relate well to … internet banking today — if I call up a certain bank, they will ask me to say a phrase. I say that phrase and then I pass security.

“If someone gets a soundbite of my voice, they’re able to then recreate that pass phrase that hasn’t changed for the last five years.

“So there’s a challenge from a fraud perspective, without a doubt that should be ringing very loud.

“How do you fight that? Part of it is education and part is an uplevelling of the security protocol.”

Having governance — making sure that people understand what could be detrimental — was also key.

For instance, Mr Horton said it would be easy for someone to put into an AI system something about themselves or the company, that they would not think to post on a website, thinking it was safe, when the opposite could be the case.

At Liberty Mutual, “we have strict governance around this”, he said. “We have our own internal model walled off and we keep our data.”

Then there is the use of AI to take on AI.

“The really cool thing with AI is the speed at which it models and the speed at which it can react and flag things that a human just couldn’t.

“In that respect you’ve got AI as the good guy and possibly as being the bad guy. That’s one of the challenges internally facing the industry.”

Mr Horton added: “I think we’ve got to balance that sort of emerging risk versus what a fantastic opportunity it is.

“If you do it right and AI does all the early work, it starts to support the human in making a decision.”

AI implications for the use of data is a massive issue, but Mr Horton said first of all the data had to be labelled correctly.

“That’s something that I think we are all working on now to label our data much better.

“There’s a lot of manual work that goes on, submissions come in PDF, a Word document, then there’s Excel. This technology can come in and put that all into the right data lake very quickly.”

He added: “I think what’s most exciting though, is being able to use the data. As insurance companies, we’ve got a monumental amount of data.”

The use of AI, he said, meant data could be analysed in a way that was very specific to a client.

“We can actually distil data down to help them from a risk management perspective and also to be able to talk about new products — things that we think are going to be important risks that aren’t products yet.”

Some reports have suggested that the insurance market is going through a paradigm shift as it learns to embrace AI.

How far has that shift gone?

Some reports, he added, showed that adoption of AI had now migrated to the top of the list for insurance businesses.

“I would suggest we are still in the early days, which I think is why it’s so exciting and good talk about now as opposed to ‘oh wow, that happened’.

“The shift at the moment would be around efficiency,” added Mr Horton. “But there’s a tremendous amount being spent on that transformation.”

Mr Horton said: “I’ve spent probably way too many hours on this subject and have undertaken courses, so I’ve adopted a belief in this.

“I don’t think everyone else is the same because there is a scepticism and a trust element about anything new.

“I think there’s a scepticism, a healthy scepticism, towards new technology that will take several years to move forward.”

Has he talked to Liberty Mutual’s clients about AI? And what has the reaction been?

“Intrigued. I think data is the conversation piece at the moment. It’s very much ‘how do we utilise that data’?

“I think the second topic is fraud. Certainly around cyber and making sure we are ready for the new world.”

The introduction and use of AI has raised concerns about job security but Mr Horton does not share those concerns, adding that it will free people up for research, creativity and greater freedom of thought.

“If we are not having to spend majority of our time on things like data input, then we have time to think about the decisions that we’re making. That’s very powerful.

“It doesn’t feel like a scary future to me because there’s so much upside to us as a business.”