As I reflect on my journey in the fast-paced world of analytics and AI, I can’t help but feel a sense of gratitude for the opportunities and challenges that continue to shape my career. From my beginnings as a student pursuing my undergraduate degree in industrial engineering and later my master’s in operations research, to my current role as Principal Researcher at Earnix. Every step of my journey has been a learning experience filled with growth and discovery that has led me to where I am today, leading projects focused on the deployment of AI in financial services.
My passion for data-driven insights ignited during my academic years, where I immersed myself in the study of industrial engineering and operations research. These foundational years laid the groundwork for my subsequent roles in algorithm development and market analysis, where I honed my skills and embraced the dynamic nature of the analytics landscape.
I joined Earnix in 2018 as a Senior Data Scientist, quickly progressing to lead the Algorithmic Research Team, becoming Machine Learning and Algorithmic Research Manager, before being appointed as Principal Researcher at the start of 2024. In my current role, my focus is on exploring the intersection of AI and insurance, a realm brimming with opportunities and challenges.
AI: Outpacing Moore’s Law
Taking a step back, my academic and professional career has run in parallel with rapid advancements in technology, particularly AI. Computer scientists have long measured supercomputing performance, traditionally doubling every 14 months in line with Moore’s Law. However, with the rise of AI and deep learning, computational performance has outpaced this law, doubling every six months over the last decade, and accelerating even more rapidly in the last 18 months. This acceleration is attributed to advancements in algorithms, access to large datasets, and increased computational power, notably through parallel processing.
The period from 2010 marks a significant shift, termed the Deep Learning Era, where AI performance surged, with machines like AlphaGo and AlphaFold showcasing the impact of large-scale models. AI and ML are transforming the landscape of financial services, offering unprecedented growth prospects while presenting notable challenges. Significantly, European businesses are increasingly embracing these technologies, with a projected 32% year-on-year increase in AI deployment by 2024. This surge could potentially contribute €600 billion in gross value added (GVA) to the European economy by 2030, equivalent to the value of the entire European construction industry. As the race for the most powerful AI machines intensifies, the conversation around the ethical deployment of AI in our society grows louder by the day.
Ethical frameworks and best practice
One of the most rewarding aspects of my work is the opportunity to apply ethical frameworks to the deployment of AI in financial services. I am deeply committed to ensuring that our AI models uphold the highest standards of equity and integrity and helping define and shape best practice in deploying AI for the global financial services sector. This means delving into the complexities of predictive modelling to mitigate biases and promote fairness in decision-making processes.
Achieving fairness involves helping algorithms make unbiased decisions, reducing systematic discrimination based on attributes like gender, race, or age. This commitment to fairness extends to mitigating any disadvantages faced by individuals or groups due to automated decision-making processes. Various metrics, including demographic parity, equal opportunity, predictive equality, equalised odds, individual fairness, and calibration, provide a comprehensive framework for assessing fairness, each addressing different aspects of unbiased decision-making.
Earnix has recently been working on an experimental module that serves as a guiding compass for ethical AI development, offering tools to identify discriminatory attributes, select appropriate fairness metrics, assess model fairness, and refine models to address potential disparities and uphold fairness principles.
By incorporating fairness into machine learning, algorithms are empowered to make impartial decisions, promoting equity and social responsibility. By using a systematic approach to address biases that looks at segmentation awareness, metric selection, fairness assessment, and model updates, developers can navigate the complexities of fairness in AI, striving to build models that promote equality and mitigate the potential harms of automated decision-making processes. Ultimately, this experimental solution aims to serve as a crucial component in steering AI development toward ethical and equitable practices.
Explainability is also a critical component amidst the rapid integration of AI and ML, enabling financial service providers to build trust through transparency. Being able to explain rationale behind decisions ultimately fosters customer confidence through ethical deployment of AI. I see collaboration with technology providers and industry experts as crucial to driving innovation and resilience forward in this evolving landscape.
Success in AI adoption hinges on proactive adaptation and strategic engagement with complex questions surrounding fairness and transparency. As I look to the future, I am inspired by the potential of AI to revolutionise financial services. However, I am also mindful of the ethical considerations that accompany this transformation.
By prioritising fairness and transparency in our AI systems, we can create a future where technology serves as a catalyst for positive change in an increasingly complex world. My journey in analytics and AI has been both humbling and empowering. As I continue to navigate the ethical waters of our industry, I am committed to upholding the values of integrity and equity in all that I do. Together, we can build a future where AI not only transforms industries but also enriches lives and fosters a more fair and inclusive society.
About the Author
Luba Orlovsky is Principal Researcher at Earnix, with 20+ years of experience in industrial engineering and operations research. She leads projects focused on the deployment of AI in financial services and has held various analytics-related roles in several companies. Luba holds a Master of Science degree from the Israel Institute of Technology.
