Why 2022 could be the year financial services embrace ethical AI

Nearly two years after the pandemic forced banking customers to take their transactions online, most financial institutions have been forced to embrace digital transformation. However, there is still a long way to go. According to Infocorp, in 2021, 73% of the digitalization initiatives of banks in Latin America accelerated compared to the pace they had in 2020, but they have difficulties in achieving innovative results due to challenges related to their technological infrastructure (29% ), an internal culture that hinders progress (25%), and budget constraints (22%). One of the reasons behind this lag is that many banks refuse to use artificial intelligence (AI) and machine learning technologies.

Organizations of all sizes can embrace the AI ​​ethic

Accountable, explainable, and ethical application of AI and machine learning is critical to analyzing and ultimately monetizing multiple customer data, which is a byproduct of effective digital transformation of any institution.

However, the low adoption rates may illustrate the reluctance of some C-Level executives to use AI, which is not entirely unjustified, as there are some studies indicating that AI could be an unreliable technology.

However, ignoring AI is not a viable strategy, as it has already been widely adopted in the business world. Even in emerging economies, AI adoption continues to rise steadily: 57% of those surveyed by McKinsey report adopting AI in at least one business function in 2021, up from 45% in 2020. More importantly, the AI and machine learning present the best possible solution to one of the problems facing many financial institutions: After implementing anytime, anywhere digital access—and collecting the high volume of customer data—organizations commonly find that they are not using that data appropriately to better serve customers.

In a FICO study, published last year, the impact of the poor combination of increased digital access and digital data provided, coupled with unmet customer needs, can be seen. This study found that, although 86% of consumers are satisfied with the services of their banks, 34% have at least one financial account, or participate in one. “additional” activity with a non-bank financial services provider. Furthermore, 70% of respondents indicated that it is “likely” or “very likely” for them to open an account with a competing provider that offers products and services that meet their unmet needs, such as expert advice, automated budgets, savings plans personalized, online investments and electronic money transfers.

The solution, very popular throughout 2021, is for financial institutions to implement AI that is explainable, ethical and responsible, and incorporate interpretable, auditable and humble technologies.

Why ethics by design is the solution

On September 15, 2021, there was a significant step towards the global standard for responsible AI with the launch of the IEEE 7000-2021 Standard. This standard provides companies (including financial service providers) with an ethical framework for implementing artificial intelligence and machine learning with standards in place to:

• The quality of the data used in the AI ​​system;
• The selection processes that feed the AI;
• The design of algorithms;
• The evolution of AI logic, and
• The transparency of AI

In Mexico, this standard has not yet been released, and there is no ethical AI principle, but it is a fact that at some point there will be regulation in this regard. Mexico is a member of the Organization for Economic Cooperation and Development (OECD), which, according to an IEEE document, promotes 5 AI principles: 1) Inclusive growth, sustainable development and well-being; 2) Human-centered values ​​and equity; 3) Transparency and explainability; 4) Robustness, security and protection, and 5) Responsibility.

As a FICO CDAO I have advocated for years for the Ethics by Design as a standard for the development of AI models. We have long awaited the structure established by IEEE 7000. As it becomes widely adopted, I believe that three new complementary branches of AI will gain popularity in 2022:

• The Interpretable AI focuses on machine learning algorithms that specify which machine learning models are interpretable and which are explainable. Explainable AI applies algorithms to machine learning models to infer the behaviors that produced an outcome (usually a score), while Interpretable AI specifies machine learning models that provide irrefutable insight into the latent characteristics that produce the behavior. punctuation. This is a significant differentiation, as interpretable machine learning enables exact explanations (with respect to inferences) and, more importantly, this deep understanding of specific latent characteristics allows us to ensure that the AI ​​model can be tested for ethical treatment.
• The AI auditable it leaves a trail of details of itself, including variables, data, transformations, and model processes, such as algorithm design, machine learning, and model logic, thus simplifying audits (hence its name). As it addresses the transparency requirement of the IEEE 7000 standard, Auditable AI is supported by well-established model development governance structures such as blockchain.
• The Humble AI it is an artificial intelligence that accepts that it does not have the correct answer. Humble AI uses measures of uncertainty, such as numerical decisions, to measure a model’s confidence in its own decisions, ultimately giving researchers more confidence in the decisions produced.

When properly implemented, Interpretable AI, Auditable AI, and Humble AI are symbiotic. Interpretable AI takes the guesswork out of what drives machine learning for explanatory and ethical purposes; Auditable AI records the strengths, weaknesses, and transparency of a model during the development stage and establishes the criteria and uncertainty measures evaluated by Humble AI. Together, the three provide financial services institutions and their clients not only with a greater sense of confidence in the tools that drive digital transformation, but also in the benefits that those tools offer.

By Scott Zoldi, Chief Data & Analytics Officer at FICO

How to apply for the 2022 Innovation Awards?