By Demessie Girma (PhD)
1. Overview
The United Nations (UN) achieved a historic milestone on March 21, 2024, by unanimously passing a resolution on Artificial Intelligence (AI) governance. This landmark achievement represents a “global consensus” among its 193 member countries, marking a notable departure from past UN resolutions. What can be inferred from this is that, up to this point, there has never been a UN resolution with a “global consensus”. This accomplishment is truly commendable. Well done, UN!
Why a resolution on AI? Because, as Ambassador Linda Thomas-Greenfield, Representative of the United States to the United Nations, articulated, AI can be “an existential” threat to mankind as much as it can create opportunities to empower mankind. Addressing why the UN sponsored the AI resolution amidst numerous other existential challenges, Thomas-Greenfield emphasized that “the answer is simple, because AI is existential”. She further emphasized that “no one should use AI to undermine peace or repress human rights.” Further added, “the resolution calls on those creating this technology to be responsible when it comes to developing and enlarging the capabilities to root out lies and discrimination in AI systems”. This point is crucial, indeed, as AI can be trained to lie and discriminate.
Beyond the significance of the AI resolution, what resonated during the announcement was the recurring phrase “Digital Divide.” Speakers, including Thomas-Greenfield, emphasized the role of AI in bridging this divide and fostering a more equitable world. Of the six speakers representing cosponsor countries (in order of presentation: USA, Japan, Morocco, Singapore, the Netherlands, and the UK), three directly used phrases such as “bridge digital divide” or “close digital divide,” while the remaining three conveyed a similar sentiment through paraphrasing.
2. The Digital Divide is Already Alive and Kicking!
Many developing countries are already grappling with various forms of digital bias perpetuated by algorithms and business practices. Below are six points illustrating this reality.
2.1 IP Addresses Reinforce Human Biases and Digital Divide
There are numerous manifestations of the digital divide that have emerged, some unintentional, such as IP address assignments by country. Looking back, I wish they were not location-dependent. I express this sentiment because the real world is already biased, and we inadvertently perpetuate bias in the Virtual Digital World. Moreover, just as human algorithms exhibit bias, location-based IP algorithms reinforce this tendency. Allow me to illustrate with a personal experience, although many examples exist.
A few years ago, while on a business trip outside of the UK, I decided to purchase a Christmas gift for my daughter on Amazon (a smartphone worth several hundred pounds for home delivery). Initially, the purchase seemed to proceed without issue, but shortly after, I received a message notifying me that the transaction had been suspended due to suspicion of fraud. This wasn’t a simple decline; it was escalated for further investigation. My educated guess is that a biased algorithm flagged the purchase because it was made from Ethiopia rather than the UK. Despite my inquiries, Amazon provided no convincing reasons for the suspicion other than the standard reassurance of “it is to protect you”. Addressing this unfortunate purchase issue took more than two months to resolve.
Identity authentication, involving a robust password and Multi-Factor Authentication (MFA) through various channels like email, SMS, and voice calls, has long been standard practice. However, the insistence on location-based security raises questions. Why does location matter when other authentication methods are already in place? This insistence seems to stem from biased algorithms, as evidenced by personal experiences like mine. Such biases create barriers rather than eliminating them, highlighting the subtle yet impactful discriminatory nature of these algorithms.
2.2 Developing Countries Miss Out on Vital Resources Due to Discriminatory Access
Inaccessible resources from tech giants like Amazon, Google, and Microsoft pose challenges for education in developing countries, where the lack of internationally recognized bank or credit cards hinders access to free learning and training materials. Addressing this issue requires tech companies to adapt their verification mechanisms to accommodate diverse global contexts.
On the surface, this disparity may not be intentional, but it constitutes a digital divide that significantly disadvantages developing countries. During a visit to a higher education institution in Ethiopia a few years ago, I was dismayed to discover how both lecturers and students were deprived of essential resources crucial for enhancing their teaching and learning experiences simply because they lacked credit cards. It’s important to clarify that in this context, the credit card requirement is solely for identification purposes rather than for making purchases.
The situation I encountered pertains to courses on cloud computing, where tech giants like Amazon, Google, and Microsoft offer extensive free resources for learning advanced software tools and training materials. However, accessing these resources often requires users to sign up, and a common requirement is to provide a credit card. This poses a significant challenge for individuals in developing countries, as many lack internationally recognized credit cards such as MasterCard or Visa. While requiring a credit card offers a reliable verification mechanism to prevent fake accounts, it inadvertently excludes a large portion of potential users who lack access to such cards.
The question that arises is: what about individuals in developing countries who lack access to internationally recognized bank or credit cards? This creates another digital divide. However, tech companies can address this divide by adjusting their requirements to accept bank cards at the country level. While this may incur additional overhead costs, it presents an opportunity for tech companies to demonstrate their social responsibility. Moreover, countries can play a role in facilitating this process by ensuring that customer bank cards are internationally recognized for third-party identity verification, thereby enabling access to free learning and training resources without the barrier of credit card requirements. Drawing from my first-hand experiences in university teaching and cloud computing development, I can attest to the invaluable educational benefits of these resources. They have the potential to empower graduates to pursue self-employment with confidence, bridging the gap between academic curriculum and industry demands.
2.3 Geo-Filtering Content is a Tool for Digital Division
Geo-filtering of social media content, whether intentional or not, contributes to a digital divide by restricting access based on location. Efforts to combat this divide must involve greater transparency and accountability from social media companies to prevent subjective biases from influencing content accessibility.
It is commonplace to come across social media postings, notably from Facebook and YouTube, that are visible in some countries but not accessible in others. This discrepancy extends beyond content suppression due to non-compliance with local laws. Facebook postings, in particular, are notorious for this issue, with various explanations circulating, some even bordering on conspiracy theories. Complaints abound regarding social media platforms manipulating viewer counts to suppress the growth of certain digital channels. Additionally, there are allegations of internal manipulation driven by political motives. Despite these concerns, it remains unclear whether Facebook management is fully aware of such complaints.
The most extreme form of geo-filtering is content blocking, often in violation of social media platforms’ stated policies. This practice underscores the subjective nature of these policies, particularly when employees tasked with enforcement bring their own biases, including political motivations. Addressing these issues is crucial to avoid perpetuating unintended digital divides. Social media companies must take proactive measures to prevent employees from imposing personal biases and ensure fair and equitable access to content across their platforms.
2.4 Unintended Consequences of Tech Giants’ Business Focus
Tech giants’ business decisions, such as app distribution policies that exclude certain countries, further marginalize regions like African countries from the digital economy. These practices highlight the need for a more inclusive approach to app publishing and accessibility to ensure that no one is left behind.
When publishing an app on the Google Play app store, I found out that Ethiopia is notably absent from the list of 176 countries, instead being categorized under the “Rest of the World” as #177. Interestingly, neighboring countries such as Eritrea, Djibouti, Somalia, and Kenya are explicitly listed among the 176. This discrepancy raises questions about the representation of Ethiopia in the app store and highlights the need for equitable treatment of all countries within such platforms.
The reply to my email inquiry to Google dated January 5, 2021, states:
“I understand you want to publish an app in Ethiopia. Unfortunately, Ethiopia is not available for distribution at this time. Until this country is explicitly listed in the Play Console, please select the ‘Rest of the world’ option instead.”
Here is another notable observation with an anecdotal example: Paid-for apps will not appear in the “Rest of the world” category, resulting in the same fate as for four countries subject to US sanctions. This raises concerns about limiting access to paid-for apps for users in these regions. Why deny users the opportunity for “window-shopping” of paid-for apps, which could serve as a source of inspiration for curious minds? This practice seems to create a digital divide based solely on commercial judgment. I suspect this is why Ethiopia, one of the largest consumer markets in Africa, is not explicitly listed, or is it?
Regarding app publishing and accessibility, a similar situation appears to apply with the Apple App Store and Microsoft Store, although I will provide specific details on the latter at a later time.
Segregating apps by country, essentially making them unavailable in certain countries, typically those in the developing world, prevents citizens of countries where the app is available from accessing it for purchase using their own currency if they happen to be located in a region where the app is not available. Following the publication of this article, a British individual commented that he was unable to purchase an app from the Apple Store while in Ethiopia, despite attempting to do so with a UK credit card. This disparity arises solely from being in a specific location, thereby creating a subtle digital divide even for consumers from the developed world. However, fortunately, it may be overcome with the assistance of VPN access.
In my experience, obtaining a development account with Apple in a developing country can be quite challenging. However, I am grateful to Google and Microsoft for their straightforward approach to acquiring a development account.
My personal experience has highlighted the stark contrast between the ease of processes in the UK and the significant challenges faced in Ethiopia. This disparity underscores the daunting obstacles that young entrepreneurs encounter. As one speaker emphasized during the UN resolution announcement on the AI governance, “…no one should be left behind” in accessing the opportunities presented by AI. This sentiment should serve as a reminder for tech giants to take proactive steps in the app market, which fuels the digital economy and serves as a source of inspiration for young learners and entrepreneurs. If left unchecked, this disparity has the potential to exacerbate the digital divide.
2.5 Need for e-Commerce Blindness to Customer Location
E-commerce platforms should refrain from location-based filtering during browsing to prevent unnecessary barriers to access. Companies like Amazon should only apply location-based restrictions at the point of purchase, facilitating seamless browsing experiences for users worldwide.
Browsing should be blind to IP addresses, especially considering that users can change their location using VPNs. It’s not helpful for e-commerce companies to inform me that an item can’t be shipped to my current location, potentially showing me different content as a result. This information is only relevant when I decide to make a purchase, at which point my account’s intelligence can handle it appropriately. Furthermore, what if I want to browse while traveling and make a purchase later? In such cases, location-based information becomes unnecessary noise influenced by human bias.
Another related aspect of location-dependent browsing is that browsers should avoid changing their language based on IP locale and instead adhere to the user’s operating system settings. I’ve explored this issue in detail in a separate article titled “Navigating the Language Quagmire: Google’s Expectations with IP Locale“.
2.6 Rethinking Location-Based Sign-In for Enhanced Security
Geo-based login mechanisms, while touted for enhanced security, risk exacerbating the digital divide by imposing unnecessary barriers based on location. Embracing robust identity authentication independent of location is essential for fostering inclusivity in the digital realm.
From personal experience, I’m aware that Microsoft’s Azure Active Directory (a cloud-based identity and access management service), and as of last July 2023 renamed Microsoft Entra ID, promotes geo-based login for enhanced security. Companies are increasingly exploring the use cases of this mechanism. However, the underlying assumption—that a user cannot be in two distant locations within an unreasonably short time frame—can lead to counterproductive outcomes. For instance, if a VPN shifts a user’s location from London to New York within two hours, the system may mistakenly block access, assuming such rapid travel is impossible. This reliance on location for identity authentication is problematic. The virtual world should be free from geographical constraints and blind to location. While opinions may vary on this matter, the drawbacks of location-based authentication are significant, prompting a shift towards robust identity verification methods that are independent of location, even when using VPNs.
3. Conclusion
The UN’s resolution on AI governance signifies a crucial step forward in tackling global challenges. Of particular note is the emphasis on the digital divide, a recurring theme highlighted during the announcement. This focus on bridging the digital gap serves as the inspiration behind this article.
As we navigate the complexities of technological advancement, it is imperative that we remain vigilant against unintentional biases and discriminatory practices that perpetuate inequalities. Bridging the digital divide requires concerted efforts from policymakers, technology companies, and society at large to ensure that the benefits of AI and digital technologies are accessible to all, regardless of location or background.
Therefore, I present my challenge to the UN: to address the Digital Divide comprehensively, not only in the realm of AI governance as outlined in the Resolution but also to tackle the existing disparities, particularly affecting developing nations as highlighted above.
Specifically, I urge the following actions:
- Encourage tech companies to fulfill their role as socially responsible enterprises (SREs) by refraining from employing location-based content modifications, as discussed, which can exacerbate the digital divide.
- Establish guidelines for the acceptance of non-international bank cards for third-party verification, enabling institutions in developing countries to access impactful free courseware materials provided by tech companies, as elaborated in section 2.2.
- Address the daunting issues of tapping into the app market by creating a developer account with tech giants like Amazon, Apple, Google, and Microsoft in the developing world, where many graduates lack access to opportunities in the rapidly evolving technological landscape, including AI, which the UN is endorsing for equitable access, necessitating substantial attention from the UN, albeit impractical to detail all necessary measures here.
- Implement a general framework to address pre-existing biases responsible for digital divides, aiming for an integrated approach to bridging the digital gap akin to the strategies outlined for AI governance.
These measures are essential steps towards creating a more equitable digital landscape, fostering inclusion, and harnessing the full potential of technology for global progress.
The Real World is biased; let’s ensure the Virtual World we’re shaping isn’t!
Editor’s Note: A former senior university lecturer and researcher, Demessie Girma, PhD, CEng, MIET, SMIEEE, now focuses on consultancy in digital transformations. With specialties in Cloud Solutions Architecture, particularly Microsoft Azure Cloud Solutions, and cross-platform Mobile App Development, he is deeply engaged in these fields. Dr. Girma is renowned for his extensive writing on emerging technologies, notably Mobile Cloud Computing and AI, with a special emphasis on their implications for the developing world, earning him recognition as a Tech Evangelist.