Hi everyone. My name is Prosper Onyenkwere and it is a pleasure to be speaking today and thanks to AIC for this event. My name is Prosper Onyenkwere and I'll be talking about ethics and responsibility we all share in AI-Driven Digital Identity.
Yeah, as we all know as the technology landscape is evolving, our traditional identity system is no longer sufficient and that underscores the need for a digital identity system to further enable the social interactions and even the transactions we perform online. Now to further simplify comes artificial intelligence into the digital identity system. So AI-Driven Digital Identity System is just talking about the use of artificial intelligence to simplify the process of verification, authentication and even management of user's identity online.
Now a practical example can be during the Know Your Customer processes done in most financial institutions where during the onboarding process the user will be asked to upload probably a government identity card and also take a selfie of him or herself. Now during this process the system which is AI-enabled will verify if the person is actually who he or she claims to be. You know this system comes with so much benefits.
Now we can't continue to talk about the benefits that this system brings without talking about the risk that it can impose especially in the cases of a misuse and that brings us to the ethical implications of artificial intelligence in digital identity system. Of course there is a disclaimer so specific companies mentioned here is actually for informational purposes. I do not mean to criticize or endorse any company okay. Now if we talk about the ethical implications of AI-Driven Digital Identity System we can talk about privacy violation and data breaches as one of the major concerns.
Now you will find out that most times companies that are developing these systems do not prioritize users privacy and oftentimes they tend to violate it. Now we can take the case of Biostar 2 web-based biometric system as a case study. It was noted that this system exposed and even made accessible over a million facial recognition and even fingerprint of its users to unauthorized users and that could have led to identity theft assuming it was not managed well. Another case of ethical implication is lack of informed consent and user autonomy.
Oftentimes when users data are being collected users are not clearly informed about what the data that is being collected is used for, how is it being stored, and in the case of sharing when is it going to be shared. Users are not really informed about that and you find out that most users agree to terms and conditions without knowing the terms included in whatever that is embedded in the system because most of the time these terms comes in lengthy and technical. Now making the user not to freely give consent about this. Now that is one disadvantage of AI driven digital identity system.
Another one we can talk about is lack of transparency and user control. Like I said on the last part, these developers are not really transparent on how these what these data is being collected are used for and in the case of sharing say user will consent once and it will happen that their data will be used for ongoing development which the user might not be aware of and users will not have control over their data. They can't even move their data freely from that application to another application. Now the major concern is bias and discrimination in AI algorithm.
Now we are going to take this case study. It might be a fictional case study but this is what might be obtainable in our system. Imagine you are Eze and you wanted to apply for a business loan to launch your startup. You have zero debt, a strong financial track record and high savings. Yet when you apply for the business loan, it was denied and you apply for this business loan through a credit scoring application called TrustPay and this is an AI enabled system that judges individual credit scores based on their biometric data, their location history and even online behavior.
Now being Eze who is a freelancer, you do not have enough social media activity. Now the system regarded your application as low credit score. You begin to wonder what could have caused this. Nobody was able to give you a clear explanation of what happened because the system was actually AI enabled. Now you felt discriminated. The system discriminated you based on because of probably the area you move in and it did not recognize freelancing as a traditional career path. Now because you have a limited social media activity, it interpreted you as less credible.
You can also not be able to apply for that because nobody was able to tell you why your credit score was low. Now this whole scenario could also cause privacy violation because your biometric data was collected based on your location history and it can create a surveillance on you because your digital credentials was captured. Now all and more of these are implications or ethical implications posed by AI driven digital identity system. So that brings us to the question of how do we incorporate ethical standards or ethical frameworks into this AI driven digital identity system.
Now as a company that integrates artificial intelligence in its system, protecting users' privacy should be your major concern. And how do you protect users' privacy? Now you should make sure that users have rights to their data. Now they should be able to know what data is collected, when it is collected. They should also have the right to give you some limitation over their data. Now how would you enable that? You can enable that by implementing privacy by design approach which is obtainable through the decentralized identity technology.
Now another way to strengthen privacy is by effective data stewardship. Nobody wants their data to be misused. Now in this case you handle individual's data like you would handle yourself, your own data right. Now the important one about strengthening privacy is compliance with privacy regulations. Now laws like GDPR are actually ethical blueprints whereby companies should follow to protect users' privacy. Now users trust your products before they share their data with you. So you should uphold that trust.
Another way you can also incorporate ethical frameworks in your digital identity system is to increase transparency in your system. Now how would you increase transparency in your digital identity system? Now you have to implement user control and consent mechanism as I earlier stated. You also publish policies and methodologies as to how identities are evaluated.
Now especially in the case of identity failures you should be able or your system should be able to tell users why their identity verification failed and provide a recourse mechanism in case of such failure so that users will know exactly what they are doing. Now you have to conduct independent reviews to make sure that you are complying with ethical standard and regulatory guidelines. We have fairness. To promote fairness in your identity system you have to introduce human oversight.
Now if you allow the system to make the judgment, that is the AI system, driven system to make the judgment, you will notice that most people that should be allowed to use access to the products will be unjustly denied because nobody is in the loop to take a look at these things. So it's very important to introduce human oversight to avoid unjust deniers. Now you use data sets that actually reflect the diverse genders, the races, now to make sure that, to train your AI models to make sure that it performs equitably.
You also have to implement human-centered design so as to enable you to see the affected communities and identify exclusion risk early. Regular bias auditing is very important to promote fairness in your AI-driven identity system. If you don't audit bias, you might not, you might, the products will become, you might not know what is going on because bias, people are being discriminated based on what the AI model is being fed on. Now these are many more actually ways of incorporating ethical standard in your AI-driven digital identity system. Now who and who have this responsibility?
Now we could, no, we can now say okay let's leave the responsibility and accountability to the developers or the technology developing these products. No, we can't actually do that. The responsibility take hold from the government, regulators, the tech companies and end users. It is a shared responsibility. The governments are there to publish policies on, just like GDPR and the AI use, on how this AI should be incorporated in your system. Now the regulators are there to make sure that these policies are being enforced.
Tech companies, like I also stated, should make sure that they use the product, they produce their products with fairness, promoting equity, promoting inclusions and also making sure that they protect users' privacy. Now end users are not left out. End users are there to make sure that they advocate for the rightful use of, the rightful development of this product. Now if we keep on being silent about these tech companies, we keep on producing this, we know that some people say that many ethics on these products will actually limit innovation. Now that is not really true.
Now if we advocate for the rightful development of these products, we'll notice that we'll actually make our society better. Now users are also to be held accountable in the case of misuse. You can take a look at this article that users, a user employed an AI avatar in Lego, Appio and George isn't amused. Now this article discusses a case where this individual used this AI replica of a lawyer to impersonate Lego council in court.
This incident, it actually underscores the need for end users to take accountability in AI solution and also particularly in this case of digital identity, you should not use a representation of another person in this whole scenario. Now in conclusion, every stakeholder from the government, policymakers, the AI developers to even users, we all must work together to make sure digital, AI driven digital identity solutions is fair and unbiased and secured and also transparent. And even we have to be held accountable in every decision we make in the usage of these systems. Thank you so much.
Thank you very much. Can you hear me? I can hear you. Yeah. Okay. Is there any questions from the audience? There isn't. Okay. So once again, thanks very much for joining us today remotely and it's great to see you. Thank you.
