Do we really understand why ethics really matter in the context of organisation public or private, large or small using AI? And how could ethics for AI help or undermine that responsible use of AI? Here are some examples of questions raised by Maria Axente, on Wednesday 11 November. As Responsible AI and AI for Good Lead at PwC, Maria seemed to us the best-qualified person to explain to our members the importance of ethics in AI. Therefore, for over an hour, Maria generously gave us some of her time to tell us about her beliefs.
Operationalising ethics in AI is a challenge that requires urgent solutions
Covid-19 definitely increased the significance of AI in today’s world, as we can see by noting that only 5% of companies are not currently using AI, compared to the previous survey, when 47% did not use AI. Given the very recent AI scandals, whether at Amazon or Facebook, the risks of discrimination and other issues are already present in our lives. Thus, “the harms of AI are not in the future”. They already exist. As a consequence, it is essential to act quickly. However, we should all be aware that there is currently no consistent and coordinated approach to addressing ethical issues that arise from AI.
Companies currently utilising AI
Why does it matter?
“AI has impact and will contribute to substantial gains in productivity and consumption but will disrupt life as we know it”. Therefore, we need to find a balanced approach between being overly excited or overly terrified about AI.
Indeed, AI can damage society as a whole, not only by rising inequality and declining social trust, but also by the creation of digital manipulation and algorithm bias. However, it can also bring many positive and valuable advantages to our daily lives.
Moreover, AI is an incredible optimization tool. Even though perfection doesn’t exist, AI has the best prediction and classification we ever witnessed. Not to mention machine learning, and the ability to codify existing models. However, some challenges can arise from this optimization tool, and we need to use our moral compass to navigate through these issues. Furthermore, we also need to have a higher level of awareness and understanding of AI to participate as active citizens, as we are all involved.
What ethics in AI is and is not
We could define it as a “moral code we translate and embed in everything around AI – design, build, acquire, use, own.” Therefore, it involves knowledge, behaviour and skills, and there are “golden rules” to follow to be on the right path. For instance, PwC’s Ethical AI Meta ethical principles are Accountability, Beneficial AI, Data Privacy, Fairness, Human Agency, Interpretability, Lawfulness and compliance, Reliability, robustness and security and Safety.
Maria specifies, moreover, that we cannot assimilate it to law or to a set of fixed rules to follow. The target is to foster good and responsible behaviour by using multiple perspectives, also called ethical lenses (Philosophical, Legislative, Cultural, Organizational).
We can, thanks to Maria, summarize the ethics in AI as doing good work that produces good and responsible AI, maximizing benefits and eliminating harms, understanding the impact and implications of our work, challenge the status quo and bring and include diverse perspectives
To conclude, we should remember that we are still at the “Can we build it” stage when we should focus on “Should we build it” and “What can go wrong”. Enforcing ethics in AI can only work with a collective effort, whether at a personal level or for all organisations. Besides, we should also foster an ethically driven digital upskilling and education to make people realize that building tech has an impact, and eventually create active citizens in the field of AI.
“The target is to foster
good and responsible
behaviour by using
also called ethical
This presentation reminded us all of the role we have to play in the implementation of ethics in AI. Thanks to Maria, everyone now has a much clearer and more precise vision of what is ethics, and how we can all help foster a good and responsible work in the tech industry. We thank her from the bottom of our hearts for her kindness and patience. And above all, let’s not forget: “Artificial Intelligence is humanity’s ultimate mirror.”
Written by Hélène Pignon