Nanda Padayachee: AI Ethics Shouldn't be an Afterthought

Standard Bank Group Head of AI, Automation and APIs Nanda Padayachee shares how his teams are building ethics into the AI development process

Enterprises must take proactive steps to ensure AI is used responsibly, Standard Bank Group Head of AI, Automation and APIs Nanda Padayachee argues in this week's episode of Data Conversations Over Coffee.

"The reality is, AI in one sense is unlike any technology we've seen in the past," he says. "We cant afford a 40, 50 or 60-year lag to figure out how we can do this responsibly."

At present, AI innovation often happens in siloed teams. As a result, ethical questions may not be raised until new AI capabilities are shared with the wider business. To address this challenge, Standard Bank has adopted an ethical framework to help ensure it uses AI for responsible purposes.

"That creates the boundaries for how we deploy any capability," he explains. "We have tried to anchor this to almost an 'AI manifesto' within our organization that says, 'The principles for how we want to use AI are anchored around these points."

However, ensuring AI is used for ethical purposes is not sufficient to ensure AI is being used ethically. Padayachee notes that companies must also ensure AI systems operate in fair ways and do not infringe on consumer rights.

"The main advances in AI over the past decade have been driven by machine learning," he says. "Explainability around that is a real challenge."

He continues: "If you're using [AI to classify] a fraudulent transaction and the customer says 'Why was this deemed fraudulent?' and you can't explain that to the customer, that creates friction."

Padayachee also argues that businesses need to focus more on maintaining AI systems to ensure they continue making ethical decisions over time.

"Because it's adaptive, because it's autonomous and because you're constantly feeding it data, you need to be very deliberate on an ongoing basis," he says. "Is it all behaving within the bounds of what you would expect?"

Key Takeaways

• AI is not a bandwagon to jump on. We must be deliberate about defining the purpose of AI systems to ensure those purposes are ethical
• Diverse teams are ethical teams. Ensuring development teams are diverse reduces the chances of unconscious biases making their way into algorithms
• Take debiasing seriously. Invest in debiasing technologies to help your team spot and address potential issues with their algorithms