The AI Ethics Tightrope: How Leaders Can Innovate Responsibly

Forbes

5 hours ago

Punnam Raju Manthena, Co-founder & CEO at Tekskills Inc. Partnering with clients across the globe in their digital transformation journeys.
AI has changed the very way we live; no two ways about it. AI helps companies make informed decisions and offer better services, and it can improve their revenue and bottom line. It can also improve products and, importantly, give customers a better experience. With an innate ability to automate repetitive, routine and time-consuming tasks, AI has transformed every facet of our lives, including manufacturing, distribution, transportation, healthcare, education, entertainment and more.
Challenges With AI
However, all is not so great with AI. First, AI is said to cost people their jobs. Also, there are fears that AI may allow personal information to be misused. But an equally big fear relates to cybercriminals using AI for deepfake phishing. In this, the attacker uses AI to fabricate audio, video or images that appear to be genuine and uses them to cheat or even intimidate the victim into taking actions that benefit the attacker.
According to IBM, an attacker spoke to an employee through AI exactly like the company's CFO and made him transfer about $25 million. Much the same way, an attacker scammed a U.K. energy company out of $243,000 in 2019.
There is also the fear of AI bias. An AI with built-in bias can add to existing biases, exaggerate existing or perceived discrimination or reinforce favoritism, preconceptions and stereotyping. Researchers at Carnegie Mellon University have looked at various Google online ads and observed that the platform is more likely to show high-end job openings to male candidates than to female candidates, according to ITRex Group. Similarly, a report from Chapman University in Orange, CA states that you may get biased hiring suggestions if you train your AI using data from male-biased organizations.
Deepfake and biases aside, there is also the issue of data manipulation, where the attacker uses AI not to steal your data but to alter it to make it hard for you to run your business.
So, you can see that AI gives you power, all right—but remember that you need to temper it with a sense of responsibility. When you run a company, you need to walk a tightrope in the sense that you must balance AI technology with ethical and humanistic practices.
To achieve this, you need to be both transparent and accountable. You should be transparent to make your AI system easy to use, understandable and accessible to your users, but also ensure that you or your developers are responsible for the impact it may have. Transparency helps you build your AI system, while accountability ensures that it has a positive effect on people and society.
How To Walk The AI Ethics Tightrope
OK, then. Transparency and accountability are essential to building trust and ensuring ethical AI development and deployment. Fine. But as a company, how do you do that? Rather, as a CEO, how would you ensure transparency and accountability?
To start with, you could document your data sources: how you collect, process and use the data; how you identify biases; and how you work to reduce their impact. Next, clearly define responsibilities: who will collect the data, who will develop the model, who will deploy it and who will supervise—all while ensuring these steps align with your values and goals.
Make sure your people are using AI in an ethical and responsible manner and, importantly, that they are complying with applicable legal and regulatory requirements. While doing all this, ensure that your AI systems do not perpetuate or magnify existing social biases and that they are generally beneficial to society.
Why This Matters
But why do you need to achieve transparency and accountability in your AI system? What do you stand to gain from building them into your AI? If you were to make a business case for ethical AI, looking at your efforts, your costs and the benefits, you might well say that ethical AI is like any ethical marketing effort: It earns you market trust, brand recognition and ROI in the long run.
All of this could lead to greater customer loyalty, stronger employee engagement, reduced regulatory risks and, of course, a significant competitive advantage.
Already, companies have started practicing responsible AI in developing their products and services. A case in point is IBM's watsonx Orchestrate, which automates a majority of workflows and processes and promotes fair play and inclusiveness.
This product has helped NatWest Group achieve a 20% increase in customer loyalty and a 10% decrease in call duration through the use of an AI-powered digital mortgage support tool, according to IBM. It has helped Avid Solutions slash their new customer onboarding time by 25% and reduce their legacy processes-induced errors by 10%.
Similarly, Camping World used a virtual assistant and saw a 40% hike in their customer engagement, a 33% growth in agent efficiency and a reduction in wait times.
Now, how do you govern an AI model? What are the recommended AI governance models, whether for a mid-sized or a large enterprise?
To begin with, you can view AI governance as a structure of policies, processes and rules that guide the development and use of AI to ensure safety, fairness and transparency. It aims to reduce risks related to bias and privacy while promoting the deployment of responsible AI. Governance addresses human errors in AI creation and maintenance, provides a structured approach to mitigating such risks and monitors whether the AI complies with ethical standards and societal expectations to guard against potential negative outcomes.
AI governance is essential for an organization to achieve a higher level of trust and compliance in the development and application of AI technologies.
With AI comes power. With AI power comes responsibility. As a CEO, you must walk a tightrope to balance AI-driven innovation with ethical practices. And finally, don't forget: AI compliance is not a "set it and forget it" task; it's an ongoing commitment.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?