Data Privacy, Trust, and Ethical AI: Why is This the Need of the Hour?

In his bestseller, I, Robot, sci-fi author Issac Asimov established the three laws of robotics. The first law states that a robot should not harm a human, or by inaction let a human come to harm. The second law states that no instruction from a human shall be disobeyed. The third law states that a robot shall avoid actions or situations that could cause self-harm unless this was done to follow law one or two. Over time Asimov’s laws have become the foundation for all things AI, even in the corporate world. The laws have also become a safety feature for those in the field of Artificial Intelligence (AI) and the base for a trending discussion: How important is ethical AI?

While there has been a great deal of discussion about the use of AI across sectors, very few conscious efforts have been adopted to implement ethical AI practices.  Wondering why ethical practices in AI are important now more than ever? Emeritus recently partnered with Women in AI to conduct a webinar on Data Privacy, Trust, and Ethical AI. Bhuva Subram, Women in AI’s North American regional head was joined by Caroline McCaffery, co-founder & CEO of ClearOPS Inc., and Yang Cheung, co-founder & CPO  of One Creation Corporation. The discussion revolved around the latest trends in artificial intelligence, machine learning, data science, and more importantly, integration with UN Sustainable Development Goals (SDGs) for collective global societal impact.  

What is Ethical AI?

Companies around the world have adopted AI to advance their services and get a competitive advantage. This is happening at a velocious pace. According to experts at Harvard, it is vital to monitor, control, and humanize this growth for the best long-term results. Ethical AI implies adopting AI in a way that is responsible, accountable, and most importantly transparent. It definitely includes abiding by laws, regulations, norms, organizational values, and consumer expectations.

Of late, ethical AI also includes protecting data, especially not letting out biased results. Moreover, every data-based decision must be justified and explainable. 

What is Data Privacy?

Subram: There is an accelerated digital transformation in the post-pandemic new normal. How can we elevate data protection for social participation, especially to focus on human rights?

Cheung: Our lives are entirely digital now. We are either attending meetings on Zoom, sending forwards on Whatsapp, uploading reels on Instagram, or getting cooking instructions from Alexa! And, as a result of this, there’s a lot of data being shared. This has, in turn, resulted in a lot of data breaches as well! Regulation worldwide is slowly tightening and moving more towards self-sovereignty. 

As per a Cisco report, 86% of respondents said they cared about their privacy and did not want data to be wrongly shared. On the other hand, 97% of respondents said they have no idea how their data is being used by companies. There is a challenge, but since individuals do not have a lot of power over their own data, companies should make conscious efforts to protect customer data. Organizations need to establish a data protection privacy culture.

From entry-level employees to top leaders, such a culture must be enforced. Responsible data collection should respect self-sovereignty, which means that we should think about giving the control back to the data owner.

How Can Data Bias be Eliminated?

Subram: What are the techniques with artificial intelligence and machine learning that we can implement to eliminate bias?

McCaffery: You need to think about doing more good than bad and that starts with looking at the data set that you’re using for the models to train the models. Companies need to make sure that it does not have a bias in it which is obviously easier said than done. AI engineers or those working with deep learning technology or even an employee involved with AI technologies, need to think about this. And not just in terms of an awesome data set and what kind of visualization can be done, but how the results of the model are going to affect one single individual.

Eliminating bias is tricky but not impossible. For example, if we have a data set that only contains 20 to 40-year-olds. Then, clearly, that’s not going to work for an age model. We need to have all ages, all the way until 120, the oldest living person or maybe even more. 

Can AI be Humanized? How to Make Ethical AI a Reality?

Subram: How to eliminate some of the discrimination in algorithms, how can we build fairness by design in machine learning and artificial intelligence? Can ethical practices become a norm?

McCafferey: We have to find more data sets if it is not diverse enough. That might be one way to keep humans in the loop and eliminate discrimination. The results will be fair and it will humanize the process. Another way, maybe that we have to build models that clean data sets right at the beginning. Ethics has to be on top of the mind. If the companies aren’t thinking about it, then rest assured, the regulators are. New York has already passed an AI law for bias regulation; it goes into effect in January of 2023. It prevents AI from being used for job applicants and hiring decisions as there is much more to hiring than just an algorithm. This is slowly spreading to Europe too! This is an excellent attempt to enforce fairness in the fast-growing space. 

There are so many open sources, online courses, and pathways to enter the world of AI. Yes, it is important to learn coding languages and work on actual AI-related tasks. However, that is a very singular viewpoint. You need to expand beyond just the core tasks and know the impact of your work. Before working on something ask yourself the following: What is the purpose? What is the impact it can have? Fighting with your internal ethics is the first step. 

How to Enter the World of AI?

Subram: If someone needs to learn and enter this field or career path, be it academic or entrepreneurial pathway, what are a few things you recommend apart from just taking courses online?

Cheung: In addition to technical skills, soft skills are a must. Only that will help make ethical AI a reality. Apart from honing soft skills, it is important to attend events and webinars to keep up with trends. Subscribing to blogs, podcasts, and magazines is also highly recommended. You must know not just the trending technologies but also the regulations and where companies are headed. 

Check out our selection of online AI and ML courses! All courses are designed by premier faculty from the world’s best universities in association with Emeritus.

Watch the entire webinar here:

By Manasa Ramakrishnan

Write to us at content@emeritus.org