by Jo Ann M Stadtmueller, SR Director, Marketing
Think about a single grain of sand. Think about the places you find sand; the beach, in the ocean, the dessert, all the dunes, even our sidewalks have sand. There are bits of sand almost everywhere. It is estimated that there are sextillion grains of sand found on the face of the earth. Data is like sand and it is everywhere. It was here in the past, here now and will be here in the future.
AI can bring a new modernity with equivalent impact that agriculture and the industrial revolution provided, propelling humanity’s next stage. But AI is also like a sand storm on the horizon. With proper preparation we can handle it and when the dust settles, we can work our way through to sunnier days.
Just how do we navigate this change and the use of data for better insights and outcomes so that the betterment of humanity is prioritized?
The goal of AI Governance is to responsibly navigate to a world that leverages advanced AI. AI Governance puts in place a framework to ensure machine Learning (ML) intentions and outcomes are positive. When an algorithm is supporting human decision making, AI Governance becomes a necessity. Ungoverned AI can undermine trust (i.e. biases negatively affecting the principles of diversity and inclusion). It is critical to understand how and why this might happen and provide prevention. Proper application of AI Governance we can avoid these types of scenarios.
A closer look, example-bias.
Take for example gender bias. There have been many instances of bias where algorithms created searches where women were under-represented. One such instance was a University of Washington study where a search query was run to find the top images for 45 professions and women were under-represented in all of the searches. Another study searched images of CEOs, even with 27% of American CEOs at that time being women, Google still only brought up results that stated just 11% were women. And still another example includes an AI-based study looking at performance reviews at five companies. “At first the study found that men and women were equally likely to meet their work goals. A deeper, AI-based analysis found that when men were reviewing other men, they gave them higher scores then they gave to women – which was leading to women getting promoted less frequently than men. Traditional analytics looked only at the scores, while the AI-based research helped analyze who was giving out the marks” states the Stanford University study.
Through AI Governance, paying close attention to inadequate labeling, biased trained data and other human judgment errors can provide preventive action. AI Governance can also provide corrective action, i.e. deploying AI / analytics to identify ‘fix’ biases.
Ensuring the responsible and ethical use of AI is starting to evolve. AI Governance is the focus of research groups, universities, think tanks, individual countries, governments and other business organizations targeting the responsible use of AI. Currently, business organizations are asked to be self-governed. But, if there is not enough governance, regulation is required (i.e. GDPR). So, self-governance is a necessary start, but as we craft new laws and policies to navigate our way through this new technology revolution, AI Governance is required to get through the storm and to sunnier days.