Digital privacy and unbiased data are top priorities, amidst the rise of AI, World Majlis told

DUBAI, 18 November 2021 – Building an inclusive, unbiased and ethically

Creating standards on the ethics of AI, ensuring robust data collection and protecting digital privacy were the most important factors building trust in the increasingly omnipresent technology, according to a global panel of experts taking part in the Expo 2020 Dubai event, as part of Tolerance & Inclusivity Week.

Ray Eitel-Porter, Global Lead for Responsible AI, Accenture – Digital Services Partner of Expo 2020 Dubai – said: “In an ideal world for organisations it would be great if there were one legislative or guideline framework that all countries adopted and for multinational companies and organisations, there was one single set of rules that they had to follow. But I have to say, I think that’s unlikely. And in a way, it’s not surprising because you know, different, different cultures, different countries, etc. have different approaches to regulation.

“But if at least we could agree on a set of standards, and maybe a set of principles that are worded in a similar way, that would help a lot with interoperability … It is going to add cost and time to the development of these solutions. It’s much cheaper and easier to develop a poor solution than it is to develop a good solution. But I think it’s really important that we get leadership from the top of organisations saying this is important enough that we need to invest in it getting it right.”

While there is an increasing awareness in the tech industry of the need for unbiased data, achieving that goal would require a major rethink, Iva Gumnishka, Founder, Humans in the Loop, said: “Instead of having a third person to label [people] on their gender and ethnicity, how do you allow people to self-label and self-identify so that your data does not have mislabeled identities. Is it even good to be labelling identity on these different images?

“So there’s a lot of things on the practical level that have you have to figure out in terms of how do you collect that data. Do you have the consent of absolutely everyone who appears on it? What about copyright? Then there’s the even bigger question of why do we classify people?”

Eva-Marie Muller-Stuler, CTO AI, Chief Data Scientist, IBM Middle East and Africa, said: “Building transparency in AI is a completely different ballgame. You have to make sure that every AI solution you build is unbiased, is fair, is explainable, is transparent, is robust so you can’t tamper with it easily, and protects the privacy of the people. And that is very expensive.”

Because AI is now increasingly used to make high-stakes decisions that could affect people’s lives, such as approving back loans or university admissions, there should be no room for error, Muller-Stuler added: “How good is good enough? How accurate do they have to be? If they are 80 per cent accurate, 90 per cent accurate, they’re wrong for a lot of people.

“We don’t always see the impact of technology. We come up with a good idea, make a lot of money, make everyone happy, but actually, in the long-run, it would have created more problems than it solved.”

The World Majlis was held as part of Tolerance & Inclusivity Week, which runs until 20 November and held in association the UAE Ministry of Tolerance and Coexistence – the fourth of 10 Theme Weeks in Expo 2020’s Programme for People and Planet, offering an exchange of inspiring new perspectives to address the greatest challenges and opportunities of our time.