Big Thinks is the Digital Magazine of the Global Mastermind Group

Algorithm Auditors: The Auditors of the Future

According to the Brookings Institute:

“Algorithms are harnessing volumes of macro- and micro-data to influence decisions affecting people in a range of tasks, from making movie recommendations to helping banks determine the creditworthiness of individuals.”

Algorithms determine your credir or wheter you get asked to interview

As artificial intelligence (AI) moves from a closed-circuit lab environment to the real world, automated decision-making directly impacts citizens’ lives. AI is no longer a programming project but a multifaceted challenge. It brings up societal biases and requires a thorough understanding of ethical principles to ensure that no one gets hurt or mistreated along the way.

AI utilizes massive data sets to predict outcomes or advise on the best course of action. In theory, the more data it has, the better the outcome should be. With one caveat, data sets need to be free of bias. Unfortunately, big data usually represents the past, not the future. Therefore, the bias embedded in the data is extrapolated and picked up on by AI elements resulting in an unfair outcome. For AI to produce ethical and fair decisions, we need to embed strong values into the algorithms.

Programmers and data scientists develop AI; the heavy focus of the models and statistical equations is targeted to produce a logical outcome that solves a problem. Developers may not realize that the data is not representing reality or that there are blind spots. For example, data could be reliable and align with the facts; however, one might question if that is what we want to see in the future as well. That’s where the algorithm auditors’ role becomes critical – asking questions that will reveal bias in the system or identifying outcomes that do not follow ethical and moral norms.

Algorithm Auditors

When you think of the word “Auditor,” an image of a 3rd party accountant comes to mind who stays on-site with their client executing a playbook of questions and processes to verify that financial statements do not have material misstatements. With companies increasingly relying on artificial intelligence (AI), a new type of audit, fueled by technology, is emerging to verify the fairness and lack of bias in the decision making process – Algorithm audit.

Comparing the two types of audit, traditional, and algorithm, there are more differences than similarities. Both are conducted by a 3rd party and serve as an independent check of the internal workings or the “black box” of an organization. But that sums up the common ground between the two. The complexity of the issues that AI is trying to solve and the constant evolution of the technology itself makes the Algorithm Auditor a mix of data scientist, machine learning engineer, and a socially savvy and ethically equipped critical thinker.

What is included in an algorithm audit?

An algorithm audit strives to diagnose unwanted consequences. An audit represents a complex set of techniques to identify flaws in the system. In an algorithm audit, AI is being reviewed for various scenarios to pinpoint bias or blind spots.

There are two types of algorithmic auditAI: direct and indirect. Direct audit represents a thorough review of the code where clear decision trees and regression models can be traced and evaluated. This is the rather traditional approach and has limitations, as can be applied only to parts of the easily understood algorithms and the impact isolated. An indirect audit consists of feeding test data sets to AI and examining the outcomes to identify bias or unwanted results. This method is especially vital for deep learning systems, where the algorithms are being developed autonomously. 

By nature, audits work best if done regularly. The same rule applies to algorithm audits – as AI systems evolve, unlawful discrimination, new biases, or blind spots may emerge. Having a continuous process of probing AI elements will bring the best results and ensure limited unwanted outcomes

What skills are needed to be an algorithm auditor?

AI systems are complex, and checking on AI to serve the public sector fairly and ethically requires multiple skills from the role of an Algorithm auditor. The algorithm auditor combines technical skills with critical thinking and logic. Most desired technical skills are captured through knowledge of data science, machine learning, modeling, and programming languages. An understanding of how intelligent elements are built will allow auditors to look for problematic patterns.  Critical thinking and logic come into play when determining what kind of scenarios or data needs to be tested and evaluating the outcomes against societal norms and values. Knowledge of regulation and ethics is critical to identifying privacy violations and providing guidance for making the model more ethical and explainable.

As the need for Algorytm Auditors is growing, the path to becoming one is starting with checking the box on multiple technical skills required. This is a continuous learning journey and knowing the latest trends in AI and Machine learning. One thing is clear:  companies utilizing AI for its processes will need to hire Algorithm Auditors to ensure best business practices and compliance with regulation and norms of society.

Big Thinks April 2021 Cover Tracy Levine Forbes Coaches Council and CEO Advantage Talent Inc

Hiring Goes Hybrid: Social Recruiting is Overtaking Traditional Methods of Hiring

The internet has facilitated the most significant impact on how we get hired. For your career to continue to experience growth and expansion, it is crucial that you know where to be seen and heard online. Getting hired in 2021 will require a multi-faceted approach, including targeted resumes and applying for jobs online while capitalizing on the new innovative ways hiring managers are using the internet and social media for hiring.

Read More »
Big Thinks April 2021 Alicia Alfonso

Social Impact: Going Hybrid

The world is no longer split between social impact and non-social impact. We’ve entered a time when every organization, government, and citizen is expected to improve the world around them in one way or another.

Read More »

IT Goes Hybrid

In this era of digital globalization, data sovereignty and privacy issues have become a constant challenge for companies that are running businesses locally or globally. Globalization demands for information flow across the borders. Cloud technology has removed the geographical barriers of data storage, providing companies the ability to store information anywhere globally.

Read More »
Big Thinks April 2021 Cover Karyn Mullins Contributor

Healthcare Delivery: The Future is Hybrid

We expect to order an item online but return it to a physical store and buy groceries online but drive a physical location for quick pickup. This digital transformation has been ongoing in many other industries for years and has now arrived in healthcare.

Read More »