RDS and Trust Aware Process Mining

Responsible data science (RDS) and trust-aware process mining are proving to be promising solutions to skewed algorithmic results. By visualizing, discovering, analyzing, reviewing, and monitoring processes, organizations can implement rigorous and trustworthy AI policies that earn consumer trust.

This forecast has broad socioeconomic implications because, for businesses, AI is potentially transformative – according to a recent McKinsey study, organizations deploying expected AI-enabled applications will increase cash flow by 120% by 2030.

But implementing AI comes with unique challenges. For consumers, for example, AI can amplify and perpetuate pre-existing biases on a massive scale. One of the leading advocates of AI algorithmic fairness pointed out three detrimental effects of AI on consumers:

Transparency. AI is a black box for many consumers:

Most lack insight into how it works. Ladder. AI often produces misleading results that can be replicated across a broader set of protected groups.
Damages. The misleading results of IA have not yet been countered with a reliable and effective remedy for damages.

In fact, a PEW survey found that 58% of Americans believe AI programs amplify some degree of bias, suggesting a tendency to be skeptical of AI’s reliability. AI fairness concerns relate to facial recognition, criminal justice, hiring practices, and loan approvals, where AI algorithms have been shown to produce negative outcomes , disproportionately affecting people and disadvantaged groups.

But what can be considered fair, since fairness is the foundation of trustworthy AI? For businesses, it’s the million dollar question.

Determining the fairness of AI

The constant evolution of AI highlights the vital importance of balancing its usefulness with the fairness of its results, creating a culture of trustworthy AI.

Intuitively, equity seems like a simple concept:

Fairness is closely related to fair play, in which everyone is treated equally. However, fairness includes a number of aspects, such as trade-offs between algorithmic accuracy and human value, demographic equality versus political outcomes, and fundamental questions. power as who decides what is fair.

There are five challenges related to contextualizing and applying fairness in AI systems:

1. Equity can be influenced by cultural, sociological, economic and legal boundaries.

In other words, what may be considered “right” in one culture may be considered “unfair” in another.

For example, in a legal context, fairness means due process and the rule of law by which disputes are resolved with a certain degree of certainty. Equity, in this context, is not necessarily about the outcomes of decisions, but about the process by which decision-makers achieve those results (and the extent to which that process conforms to standards). accepted legal standards).

However, there are other cases where the application of “corrective equity” is necessary. For example, when dealing with discriminatory practices in lending, housing, education and employment, fairness is not about equal treatment of everyone but affirmative action. As a result, recruiting a team to deploy AI can be a challenge of fairness and diversity.

2. Fairness and equality are not necessarily the same thing.

Equality is considered a fundamental human right: no one should be discriminated against on the basis of race, sex, nationality, disability or sexual orientation. While the law protects against differential treatment – when protected class individuals are intentionally treated differently – AI algorithms can still produce impactful results – when variables at first glance appear to be bias-neutral, causing unintended discrimination.

To illustrate the magnitude of the impact, consider Amazon’s same-day delivery service. It is based on an artificial intelligence algorithm that uses attributes, such as distance to the nearest fulfillment center, local demand in specified postcode areas, and delivery frequency distribution. key staff, to identify locations in favor of free same-day delivery. Amazon’s same-day delivery service was also found to be biased towards people of color, although race was not a factor in the AI algorithm.