Balancing Innovation with Integrity in AI

Artificial Intelligence (AI) is transforming industries worldwide, bringing both excitement and apprehension. In a recent roundtable discussion hosted by Altron and themed “Balancing Innovation with Integrity in AI,” industry leaders and experts gathered at the Saxon Hotel in Sandhurst to explore the complexities surrounding AI, from ethics to integration. The discussion highlighted several critical issues that businesses and policymakers need to consider as they navigate the AI landscape.

Ethics around AI, Gen AI

A key theme from the roundtable was the ethical considerations surrounding AI, particularly generative AI, which has raised new concerns about the unintended consequences of technology. Participants highlighted that people’s apprehensions about AI are not unfounded as generative AI has shown the potential for misuse.

The anxiety stems from uncertainties regarding what could go wrong, such as job displacement, misuse of data, and even AI surpassing human control.

Generative AI comes hand in hand with questions about creativity, ownership, and accountability. Who is responsible when AI-generated content infringes copyrights or misrepresents facts? These issues go beyond technical concerns and tap into deep ethical questions about the role of AI in shaping our future.

The Role of South Africa’s National AI Policy Framework

South Africa’s national AI policy framework has been developed to guide the responsible use of AI in the country. However, attendees cited the rapid pace of AI development, outpacing policy frameworks as an issue. The policy serves as the backbone for ethical AI use, but as AI continues to evolve, policymakers struggle to keep up with the latest advancements and find themselves on the back foot.

The roundtable participants pointed out that merely accumulating policies without regular updates will not resolve the issue. A more agile approach to policy development is needed, allowing businesses and governments to adapt to the changing AI landscape. The issue is particularly pressing in regions like Africa, where participants argued that the continent lacks a strong enough voice in shaping global AI policies (which we’ll cover in more detail further down).

Integration challenges: IT, blue-collar workers, and executive resistance

Several hurdles were identified when it comes to AI integration. One major concern is that AI may take control away from IT departments, creating tensions within entities. Delegates also voiced the opinion that CIOs should not be the sole owners of AI, and its implementation should be a cross-functional effort that includes voices from across the business, from IT and HR to marketing and operations.

The integration of AI is also particularly challenging for blue-collar workers, where ethics and implementation become more complex. For this workforce, AI often represents job insecurity, as automation can lead to roles becoming redundant. Roundtable participants stressed that AI should be a tool for creating value and opportunities within the workforce, but upskilling blue-collar workers is crucial.

For these and other reasons, educating large groups of people is a significant challenge, particularly in industries that traditionally do not rely heavily on technology.

Executive resistance was another barrier discussed, with many executives hesitant to embrace AI due to its complexity and the potential disruption it could cause within their businesses. While AI offers the potential for long-term gains, some executives are impatient to see quick returns on investment (ROI), which can lead to rushed investments, which, in turn, stifle innovation and integration.

Data as the foundation for AI success

Data was highlighted as the starting point for any AI initiative. Without high-quality, diverse, and unbiased data, AI cannot function effectively. However, data sharing comes with its own set of challenges, including the liability of data being lost or misused, particularly considering an increasingly complex cybersecurity threat landscape. POPIA (Protection of Personal Information Act) compliance in South Africa adds another layer of complexity, as businesses must navigate strict regulations on how data is collected, stored, and shared.

AI’s inherent bias, often a reflection of the data it is trained on, also presents a barrier. If AI systems are trained on biased data, they will continue to perpetuate these biases, leading to unfair outcomes. The issue is particularly concerning when AI decisions impact marginalized communities or when they reinforce existing social inequalities.

Malicious intent and geopolitics

No discussion about AI can happen without touching on the dark side of the use of the tools, and the conversation inevitably moved to the potential for AI to be used with malicious intent. From deepfakes to cyberattacks, AI can be weaponized in ways that bring significant risks to businesses, governments, and people alike. This shines a spotlight on the importance of robust legal frameworks and policies to guide AI’s development and prevent its misuse.

Moreover, the geopolitics of AI cannot be ignored. Who controls AI, and whose voice is being prioritized in global AI development? The industry leaders emphasised that Africa still does not have enough influence in the global AI arena, and the lack of representation could hinder the continent’s ability to shape AI to meet its needs. Questions were also raised about whether AI is affordable for African countries and if it is being designed with an African context in mind.

AI education, testing, and upliftment

For AI to be successfully implemented, education and upliftment are essential. Participants agreed that AI must be rigorously tested before deployment in critical areas such as healthcare, finance, and national security. Proper testing ensures that AI systems function as intended and do not produce harmful or unintended outcomes.

Upskilling the workforce is another critical component of AI success, but the scale of education required is daunting. In industries that are slow to adopt new technologies, educating employees on AI’s potential and limitations is a challenge that businesses must address promptly

Asking "why?" Before embarking on an AI journey

The roundtable concluded with a call to action for businesses: before embarking on an AI journey, always ask “why.” AI should not be implemented for its own sake but rather as a tool to create value, solve problems, and improve operations. Diversity of thought, including input from different industries and sectors, is crucial to unlocking AI’s full potential.

Ultimately, AI presents a tremendous opportunity for innovation, but it also comes with significant risks. By taking a thoughtful, ethical approach, businesses can harness AI in a way that benefits society while addressing the potential challenges.