Prior to joining SAS some 10 years ago, Benoit Van Laethem worked at IBM for just under two decades.

"I could categorize my experience at IBM into two main streams. Initially, I started in the marketing division and worked on various missions at the European level, focusing on European marketing programs. After some time, I progressed to leading a team of marketing professionals, which led me down a management stream. After spending 10 years in marketing, I felt the need for change. That's when I made the move to sales and account management at IBM. The switch allowed me to engage directly with clients, work on tangible projects, and build relationships, which was a different aspect I wanted to explore."

What makes working here that different?

Benoit Van Laethem: "One thing I appreciate is the opportunity to delve deep into a specific topic, as our core focus revolves around data and analytics. In contrast, during my time at IBM, the product portfolio was much broader, encompassing hardware, software, and services. So, while still having a general approach, my current role maintains a strong emphasis on the analytics domain, specifically to support customers from the financial industry, including banks and insurance companies."

What attracts you to the financial sector?

"There are several aspects I find appealing. Firstly, there are ample possibilities within the industry, with substantial budgets often associated with large enterprises. Additionally, the financial sector is a legacy industry, meaning there are numerous legacy systems in place. This presents a continuous demand for transformation projects, which provides great opportunities. Furthermore, the industry generates vast amounts of data, which is not always effectively utilized. This presents further potential for improvement and offers exciting challenges to tackle."

"As an account executive, my role primarily involves listening to the needs of our customers and providing support to them. We are organized to cater to a specific set of customers, ensuring we address their requests while also exploring new initiatives. We collaborate closely with the companies we work with, striving to understand their strategic initiatives and finding ways to assist them accordingly. It requires an engineering mindset, active listening, and empathy to comprehend the unique circumstances of each client."

"Banks and insurance companies often lag in adopting advanced technologies, making it less appealing for younger individuals who are more accustomed to working with cutting-edge tools and systems"

What are some typical issues the financial sector often seeks solutions for?

"There are typically three major types of requests. Firstly, banks and insurance companies often inquire about increasing their revenue or identifying cross-selling opportunities. They aim to optimize their sales and explore ways to generate more demand for their products. Secondly, cost optimization is a common concern. Businesses want to operate more efficiently, reduce expenses, and enhance overall cost management. Lastly, there's a significant focus on regulatory compliance. Despite regulation being even stricter for banks, both banks and insurance companies face stringent regulations such as Solvency II and IFRS 17, which require them to maintain sufficient capital and adhere to anti-money laundering protocols."

What are the key challenges within data science today for insurance companies?

"The first challenge in the insurance industry regarding data science is the scarcity of skilled resources. Finding, training, and retaining data scientists is a struggle for most insurance companies in Belgium. There is a shortage of talented individuals with expertise in data science, and even when they are hired, they tend to move on to other companies relatively quickly."

"The second challenge is the image problem faced by the insurance and banking sector. These industries are not always perceived as attractive career options, despite having the financial means to attract talent. The outdated perception of these sectors hinders their ability to attract and retain young professionals."

"Legacy technology and slow implementation processes also pose a challenge. Banks and insurance companies often lag in adopting advanced technologies, making it less appealing for younger individuals who are more accustomed to working with cutting-edge tools and systems. The absence of the most advanced technology or the delays in implementing them can discourage data scientists from staying long-term in these organizations."

"Another significant challenge is data quality. In the world of analytics, data is crucial, and if the data is of poor quality, the insights derived from it will be unreliable. Ensuring data quality remains a top priority and a challenge for insurance companies."

"Siloed working environments and lack of collaboration between teams hinder the smooth implementation of data models into production"

"A common problem known as 'data model operationalization' is the third challenge. It refers to the difficulty of transitioning data science models from development to production. Many models developed by data scientists do not reach the production stage, where they can generate value for the company. The process often involves passing the models from one team to another, causing delays and inefficiencies. Siloed working environments and lack of collaboration between teams hinder the smooth implementation of data into production. It is essential to adopt principles like DevOps, within data science it's known as ModelOps, where mixed teams work together to ensure seamless operationalization of models. This includes developing models within a framework that enables quick and automated deployment, along with proper governance and oversight."

These challenges highlight the need for addressing skills shortages, improving data quality, adopting advanced technologies, and streamlining the operationalization process to fully leverage the potential of data science in the insurance industry. How to tackle this multitude of 'to do's'?

"To address the skills challenge, we have implemented various initiatives. One example is our local program, which has now expanded to a European level. We collaborate with universities to educate and provide free training to students. During the summer, we enhance their skills in data science based on our technology, bringing them to a minimum maturity level. These students are then made available to be hired by partnering companies, creating a young graduate program."

"In terms of technology, we aim to democratize the analytics lifecycle. We believe that data science should not be limited to specialized data scientists but accessible to other profiles as well. We motivate business analysts and individuals with an affinity for data to utilize our platform. By lowering the entry levels and offering easy-to-use interfaces, we ensure that the power of analytics can be harnessed by a broader range of users."

"Furthermore, we have embraced other statistical programming languages like Python and R. While these languages are popular in the market, enterprise-level solutions require proper governance and documentation. Our software enables developers to code in SAS, Python, or R within the same platform, ensuring consistent governance and documentation, which is crucial for regulatory compliance and effective management. Some customers may prefer to make a specific choice, such as developing solely in SAS, R or Python. However, we believe that providing a framework and allowing people to choose what suits them best is crucial in a diverse market. By doing so, we can avoid constantly changing our approach and also support the professional development and retention of young data scientists who want to explore and learn new languages. Attractiveness becomes a key factor in this regard."

"We are highly committed to ensuring the reliability, transparency, and ethical use of artificial intelligence"

"When it comes to addressing the challenge of operationalization, we have integrated the concept of DevOps into the analytics world through ModelOps. ModelOps focuses on developing models with the intention of putting them into production seamlessly. Our platform incorporates the necessary capabilities for developing and deploying models, regardless of the statistical programming language they were built in. We emphasize the importance of maintaining models in production, constantly assessing their performance, retraining them if necessary, and ensuring that the champion model remains the best choice over time. This is achieved through robust model governance and a feedback loop to effectively manage the lifecycle of models, even when dealing with lots and lots of models in production."

"Lastly, the third significant challenge we tackle is trust. Internally, we refer to this as 'Trustworthy AI'. We are highly committed to ensuring the reliability, transparency, and ethical use of artificial intelligence. By incorporating rigorous quality control, robust governance, and adherence to regulatory requirements, we aim to build trust in the analytics cycle and instill confidence in our clients."

Can you elaborate on the trustworthiness of AI?

"We adhere to six key concepts across all our software and capabilities, particularly when it comes to models that have the potential to impact clients. Transparency is crucial, ensuring that models can be explained and reverse-engineered to understand the decision-making process. It is important to provide explanations of why certain decisions were made and which data elements were considered. Models that function as black boxes are not acceptable."

"Another concept is inclusivity, aiming to prevent bias in models. Techniques are available to address bias, although subjective interpretation may arise in certain cases. Establishing governance frameworks within companies helps ensure trustworthy AI and analytics with regards to inclusivity."

"Accountability is another aspect we prioritize, requiring the identification and mitigation of all impacts resulting from modeling efforts. Setting up governance structures, although currently rare among Belgian companies, can enable better control. Robustness is also essential, referring to the consistency of results when the same data is processed on different occasions. Security and privacy are emphasized as well, ensuring compliance with all relevant aspects. These principles are fully integrated into our software products to meet these challenges."

What about data quality?

"That is indeed another challenge, and we address it through technological solutions. While we cannot replace missing customer data, we provide capabilities for data cleaning, matching, and enhancement within our software. Automation is a trend we embrace to streamline data preparation and quality tasks. By automating profiling and providing immediate insights into trends, missing data, outliers, and potentially sensitive or GDPR-related information, we aim to make the work of data scientists and business analysts more efficient. The goal is to shift the focus from spending 80% of their time on data preparation to devoting the majority of their time to relevant work."

"AI must deliver tangible improvements and automation to enhance decision-making processes"

Which new technologies are nowadays changing the field of data science?

"One major change in recent years has been the increased availability and ease of acquiring computing power. Unlike in the past, there is no longer a need to compromise on the size of data when performing analyses. Cloud technology has made it possible to analyze large datasets without constraints. For example, you can spin up containers in the cloud for a short period, harnessing a substantial number of cores and achieving rapid results. This availability of power is a significant shift in the data science landscape."

"As for artificial intelligence (AI), it has become a buzzword in recent years. While AI has existed for decades, advancements in compute power have made it more accessible and easier to implement. AI models require extensive training data, distinguishing them from standard models. We enable the use of AI techniques throughout our platform, empowering clients to personalize and deploy AI models. Our focus goes beyond model development; we emphasize extracting value from AI by integrating it into specific applications and processes. Ultimately, AI must deliver tangible improvements and automation to enhance decision-making processes."

AI may have been around for decades, it is talk of the town like never before. Rightly so?

"Regarding the public perception of AI, opinions differ. Some individuals may fear its impact, while others recognize the advantages of automation, which allows for the allocation of human effort to more value-added tasks. We view AI both with excitement and caution. Excitement stems from the increased accessibility of AI techniques, while caution arises from the potential for untrustworthy AI. It is crucial to ensure that AI remains explainable, unbiased, and transparent before widespread adoption. Presently, models like ChatGPT can be seen as black boxes, and their limitations should be considered. For instance, while ChatGPT may provide the most commonly predicted answer for a question, it may not necessarily compute the answer accurately in cases like basic arithmetic."

"Insurance companies, in particular, are investing in data platforms, but there is room for improvement in leveraging them effectively"

Do you see other trends in the rapidly evolving world of technology?

"New technologies are constantly emerging, especially in the field of data science. In terms of recent trends, there are a few notable advancements. Firstly, there is a strong focus on developing trustworthy and responsible AI. The goal is to ensure that AI models can be relied upon for accurate predictions. Secondly, there is an emphasis on data intelligence and augmented analytics. Automation techniques are being applied to streamline the work of data scientists and business analysts. This includes automating data quality processes and incorporating natural language interpretation in reports and analyses. These advancements aim to increase automation and productivity in the field of data science."

"Another significant trend is composite AI, which involves embedding artificial intelligence into various applications. The market is opening up with interfaces and APIs, allowing for the development of AI-driven applications. Additionally, the adoption of cloud technologies is increasing across the board. The cloud offers ease of consumption and operationalization, making it a popular choice for organizations."

Are there significant differences between certain industries?

"In Belgium, there is a mixed picture. While there are efforts being made to incorporate new technologies like AI and robotics, a significant portion of the focus still lies on upgrading and migrating legacy systems. Many companies prioritize building a solid data platform as a foundation before delving into advanced analytics and personalization. The maturity level in Belgium's data analytics space is still a work in progress. Insurance companies, in particular, are investing in data platforms, but there is room for improvement in leveraging them effectively."

Do other countries do better?

"In comparison to other countries, Belgium may lag in terms of AI adoption. Some countries are more advanced in utilizing AI techniques, both in standard analytics and advanced AI modeling. Across the insurance value chain, there are numerous examples of AI being employed, from product design to pricing. Machine learning techniques help optimize insurance contract pricing, ensuring fair and explainable rates. However, regulatory considerations often favor traditional techniques like generalized linear models (GLM). The acceptance and flexibility of AI techniques by regulators are gradually changing."

"In the marketing space, advanced analytics techniques are commonly used to optimize the customer journey and provide personalized offers. By leveraging data and analytics, organizations can propose the best next action or offer for each prospect. Similar to Google's targeted ads, the goal is to deliver the most relevant offerings based on individual profiles."

"Data scientists will play a crucial role in leveraging these technologies to drive innovation and improve business outcomes"

How are insurers nowadays using new technologies to improve their products and services?

"There are several areas where we see significant results and benefits from the application of new technologies. One such area is claims handling in the insurance industry. By optimizing the claims process from the initial notice of loss, analytics techniques can be employed to prioritize claims and even automate standard ones. This application of analytics helps streamline the claims handling process and improve efficiency."

"Another area where advanced techniques can be applied is in fraud detection and prevention. While regulators may not yet fully accept advanced techniques, there are still ways to enhance the traditional rule-based systems. By using machine learning techniques, the prioritization of alert queues can be improved. The acceptance of these advanced techniques by regulators may vary, but demonstrating explainability remains crucial."

"When it comes to customer interactions, there is potential for improvement as well. The goal should be to automate mundane tasks through AI, freeing up time for workers to focus on more complex and value added activities. It is important for end clients to see the value in these new technologies, as it creates a win-win situation. Transparent explanations and the ability to showcase the benefits to customers are vital."

How do you see the role of data scientists evolving as financial institutions rely more on new technologies?

"As we continue to advance in the field of AI and data science, data scientists will play a crucial role in leveraging these technologies to drive innovation and improve business outcomes. Let me give you a few key aspects that could shape the evolution of their role - in fact, there are quite a lot of aspects. With the increasing focus on ethical AI and the development of governance frameworks, data scientists will need to incorporate ethical considerations into their work. They will play a vital role in ensuring that AI systems are designed and deployed in a responsible and transparent manner, aligning with the values and regulations of the industry."

"Data scientists will continue to be at the forefront of technological advancements, driving the adoption of AI and data-driven strategies to shape the future of insurance"

"The introduction of regulations specifically addressing artificial intelligence, such as the Artificial Intelligence Act from the European Union, will require data scientists to understand and comply with the legal obligations outlined in these regulations. They will need to ensure that AI systems meet the necessary requirements, such as transparency, explainability, and accountability."

"Data scientists will continue to collaborate closely with industry stakeholders, policymakers, and regulators to contribute their expertise and insights. They will be consulted during the development of regulations and guidelines to ensure practical and effective implementation of AI technologies in the insurance sector."

"Data scientists will also be expected to focus on delivering value through their work. They will play a key role in identifying business opportunities, leveraging data analytics, and developing innovative AI solutions to address industry challenges. Their expertise will be crucial in driving operational efficiency, customer satisfaction, and profitability for insurance companies."

"As automation techniques and augmented analytics continue to advance, data scientists will need to embrace these tools to automate repetitive tasks and enhance their productivity. This shift will enable them to focus on more complex and strategic tasks that require human creativity, critical thinking, and problem-solving abilities."

"Overall, data scientists will continue to be at the forefront of technological advancements, driving the adoption of AI and data-driven strategies to shape the future of insurance."

"Last but not least, it is important for data scientists to find internal allies to make the undervalued potential of the data available to them better known to the executive management within insurance companies."

SAS in short

SAS (short for 'Statistical Analysis System'), is a software suite that can mine, alter, manage, and retrieve data from a variety of sources and perform statistical analysis on it. It is well known for data management, advanced analytics, business intelligence, etc. Benoit Van Laethem: "In terms of our clientele, while we could serve a broader range of companies, our primary focus, based on our history, lies with larger banks and insurance companies. In Belgium, for instance, the top 20 banks and insurance firms represent approximately 90% of the market. However, we remain open to opportunities from smaller entities and collaborate with partners who can provide comprehensive coverage."

Benoit's data

• 12 / The number of years that separate ‘London Calling’ (The Clash) from ‘Nevermind’ (Nirvana), two music albums that struck me.

• 2 / °Celsius, the objective of the Paris Agreement to limit global warming - one of the multiple environmental challenges I’m sensitive about.
• 30 / years ago, I ended my studies… That doesn’t make us younger, hé?!
• 50% / I prefer seeing the half-full glass in terms of its content, not in terms of the void.