Key takeaways from Better Debt: AI and Machine Learning Masterclass

We all know that AI is the buzzword of the last 12 months. But how can you identify real opportunities to apply this technology to your organisation? Our AI and Machine Learning Masterclass tackled this question, and more:

How should you differentiate AI, Machine Learning and data science?
What are real use cases and examples of each, and how does this vary depending on where your business is in its data maturity journey?
How can you align this type of innovation with organisational goals?
What are other practical applications of AI and Machine Learning in teams such as Finance, Compliance or Marketing?

Find out what industry experts Duhita Khadepau (Senior Manager, Enterprise Data - Cuscal), Jo Mikleus (Non-Executive Director AI Strategy - Success2Significance) and Mike Zhou (Chief Data Officer - InDebted) said when they sat down with Better Debt host, Josh Foreman (CEO and Founder - InDebted).

1. The difference between AI and Machine Learning

Despite being used interchangeably, AI and Machine Learning aren’t synonymous. As Duhita explains:

When we talk about AI, it’s artificial, it’s basically not natural. Intelligence is replicating human intelligence, so things like self-driving cars or chatbots. You’d feel that an actual human is doing some of those things, but it’s actually the AI running behind it.

Now if you look at Machine Learning, you have a machine which is learning - but what are those machines? What are they learning? They’re basically a set of rules, algorithms, created by humans, and the machines are learning based on just the data provided. So you give it 3 months of data to train on, then it uses this to predict outcomes.

Looking at the definitions of each, you can start to uncover how their applications differ. For example, AI can be used for tasks requiring human intelligence, such handling inbound customer enquiries. On the other hand, Machine Learning analyses specific data sets. For example, if you wanted an ML model to predict which customer segments are more likely to respond to an SMS instead of an email, you’d feed in recent data which it would use to generate recommendations.

AI is suitable for tasks that require replicating human intelligence…Machine Learning is more appropriate for tasks that involve classifiers and numeric data. — Duhita Khadepau

2. Understand where you are on your Data Maturity journey

Now we’ve aligned on definitions, let’s start looking at how you can find opportunities for AI and Machine Learning in your organisational strategy. To do this, you need to understand where you are on your Data Maturity journey.

Using any form of AI or Machine Learning requires data. Not just any data, but good quality data. This means getting up close and personal with your data to understand:

How is the data collected? What exactly is being collected?
Are there any biases in the data? What steps are taken to mitigate for them?
What are your standardised processes for data management? Do all teams analyse data in the same way using the same tools?

Once you’ve got a good idea of the data available, you can start considering what problems need solving, and how you can use this data to solve them. Duhita says that jumping straight into Machine Learning models isn’t always necessary:

One thing I always tell the team is the answer is often extremely simple, and we don’t need to get complicated with Machine Learning models. Try writing an SQL. Try writing case statements. I think as combinations start to increase, or you can’t code all of those combinations, that’s when you start thinking - can I apply Machine Learning?

3. Adopt a culture of innovation

How do you innovate in highly regulated environments? Jo explains that it starts with mindset adjustment:

Putting risk and innovation in two different categories is an old way of thinking. I’ve found bringing them together to be more productive and progressive.

It’s about an internal cultural shift that gets everyone comfortable with using AI and Machine Learning. Why is this so important? With technology evolving at lightning speed, setting your organisation up for long-term success means adopting a culture of innovation.

Take your entire organisation on the journey. This means supporting teams to find ways they can use AI in their everyday tasks - from Marketing to Finance. Josh unpacks how InDebted is adopting an AI-driven culture and monitoring company-wide uptake:

One of our company goals is to build an AI-driven culture. We monitor the entire company, not just the Data or the Engineering team - who you totally expect to use this technology. We’re tracking usage of tools like ChatGPT across the business, because we want people to see what is possible.

4. De-risk to meet regulatory requirements

Meeting regulatory requirements is always top of the list for Financial Services organisations. As a Board member of the global RegTech Association, Jo broke down how they’re at the forefront of innovation:

Regulators are very focused on AI. They know that the technology is going to be pervasive, and they’re actively working on implementing AI themselves. Our industry benchmark report showed us that 80% of global regulators have implemented innovative solutions

To align with regulatory priorities, demonstrate that you’re minimising harm and ensuring positive outcomes. This means carefully considering how you’re implementing AI and Machine Learning and taking the regulator on the journey. For example, during data exploration, take the time to understand how data is collected, and what specifically is fed into Machine Learning models:

Regulators know that generative AI can benefit consumers directly and indirectly. The concerns are around the data, they want to make sure that there’s a transparent modelling process with solid inputs and no biases. - Jo Mikleus

5. Explore the practical applications of AI and Machine Learning

Use cases for AI and Machine Learning are endless. To get the most out of the technology and any new initiative, Mike recommends starting with your cost to benefit ratio:

The cost to benefit ratio is different for each organisation which comes down to a few things. It’s worth exploring internal data that’s proprietary to your product or service, because that’s unique to your organisation.

Alongside this, get a deep understanding of the processes that you’re trying to improve. How is it currently being carried out and by who? How would improving this process or function benefit your customers? Once you have this foundation, you can then get specific on what data will be the most effective to solve your problem.

When exploring customer benefits, also consider how trust will be built into your AI and Machine Learning. Baking ethical principles into your data management is a strong starting point, and ensuring transparency throughout your customer-facing initiatives. Jo calls out explainability as an optimiser:

One of the core tenements of trust is transparency. The good thing about AI and Machine Learning is you can build in explainability as part of the solution.

AI and Machine Learning: The art of the possible

When it comes down to it, getting the most out of AI and Machine Learning for your organisation is about mastering the art of the possible. As our panel explains, using this technology to solve every problem isn’t the goal. To get the best outcomes, take the microscope to your organisation’s goals and pain points to work out exactly where AI and Machine Learning could offer a better solution. That’s how you’ll bring the ‘possible’ to life.

Join our next event

Expect more from your collections

Our Collect product provides a better solution to traditional third-party unsecured consumer collections services. See 40% increased recovery performance, while using world’s highest rated debt collection solution for customer experience.

Get in touch

See all