Navigating the ethical and legal risks of AI implementation

As Artificial Intelligence (AI) becomes increasingly integrated into various aspects of business operations, it prompts many ethical and legal challenges. Businesses must navigate these complexities carefully to harness AI’s potential while safeguarding the organization from potential risks. Before hastily implementing emerging AI tools and technologies, businesses must explore the ethical and legal risks associated with AI implementation, with a particular focus on their impact on customer and employee experiences, especially in customer service and contact center environments.

Understanding the risks

AI systems, while powerful and transformative, do not come without pitfalls. The primary risks lie in three main areas: legal, ethical, and reputational.

  • Legal risks extend from non-compliance with various AI regulations and legislation.
  • Ethical risks pertain to the broader societal and moral implications of AI use. Ethical risks often extend beyond legal compliance to include fairness, transparency, and the potential for AI to perpetuate or exacerbate existing inequalities.
  • Reputational risk involves potential damage that arises from perceived or actual misuse of AI. Negative public perception can result in loss of customer trust and ultimately impact a company’s bottom line.

Legal risks in AI implementation

Learning and navigating the regulatory landscape should be non-negotiable for any business implementing AI. With AI technology being implemented across every aspect of businesses at an unprecedented rate, the landscape is constantly changing, with significant differences from region to region.

In Europe, the EU Artificial Intelligence Act is poised to build on the already comprehensive data privacy legislation set forth in the GDPR. The EU AI Act categorizes AI models and their use cases by the risk they pose to society. It imposes significant penalties for companies that leverage “high-risk” AI systems and fail to comply with mandatory safety checks like regular self-reporting. It also introduces across-the-board prohibitions, including the use of AI for monitoring employees’ emotions and certain biometric data processing.

In the U.S., a more diverse state-by-state approach is developing. For instance, in New York, Local Law 144 mandates annual audits of AI systems used in hiring to ensure they are free from bias. State-level mandates are directed by the recent Executive Order regarding safe, secure, and trustworthy AI and subsequent Key AI Actions announced by the Biden-Harris Administration. It is imperative for companies to stay up to date on the evolving regulations to avoid hefty fines and legal repercussions.

In customer service, this translates to ensuring that AI systems used for customer interactions comply with data privacy and developing AI laws. For example, AI chatbots must handle customer data responsibly, ensuring it is stored securely and can comply with data subject rights, such as the right to be forgotten in the EU.

Ethical risks and their implications

The ethical risks of AI can be identified by considering two domains of ethical significance: harm and rights. Where AI might cause, compound, or perpetuate harm we must take steps to understand, remedy, or completely avoid those harms.

A key example of this kind of ethical risk is the harm brought to individuals by AI systems that unjustly or erroneously make decisions of great consequence. For example, in 2015, Amazon implemented an AI system to help perform an initial screening of job candidate resumes. Despite attempts to avoid gender discrimination by removing any mention of gender from the documents, the tool unintentionally favored male candidates over female ones due to biases in the training data. As such, female applicants were repeatedly disadvantaged by this process and therefore suffered the harm of indirect discrimination.

Further ethical risks include when AI might infringe on human rights, or when its pervasiveness points to the need for a new category of human rights. For example, in its prohibition of biometric AI processing in the workplace, the EU AI Act seeks to address the ethical risk of having one’s right to privacy undermined by AI.

To mitigate such risks, companies must consider adopting or expanding comprehensive ethical frameworks. These frameworks should include:

  • Bias detection and mitigation: Implement robust methods to detect and mitigate biases in training data and AI algorithms. This can involve regular audits and the inclusion of diverse data sets to train AI systems.
  • Transparency and explainability: Ensure AI systems are transparent to avoid potential deception, with decision-making processes that can be explained. Customers and employees should be able to identify and understand how AI decisions are made and have available avenues to contest or appeal these decisions.
  • Fairness and equity: Implement the necessary measures to ensure the benefits of AI are distributed fairly across all stakeholders. For instance, in customer service, AI should enhance the experience for all customers, regardless of their background or demographics.

Reputational risks and proactive management

Reputational risks are closely linked to both legal and ethical risks. Companies that fail to address these adequately can suffer significant reputational damage, which often leads to tangible, negative impacts on business. For example, a data breach involving AI systems can erode customer trust, lead to public backlash, and ultimately cause a loss in customer loyalty and sales.

To manage reputational risks, Avaya believes businesses should:

  • Engage in responsible AI practices: Adhere to best practices and guidelines for AI implementation. This includes being transparent about how AI is used and ensuring it aligns with ethical standards.
  • Communicate clearly with stakeholders: Keep customers and employees informed about how AI systems are used and the measures in place to protect their interests. This level of transparency builds trust and often mitigates potential backlash.
  • Implement a robust governance framework: Establish an AI governance program to oversee AI implementation and ensure compliance with ethical and legal standards. This program should include representatives from various business units and have clear processes for monitoring regulatory guidelines and evaluating AI projects. To fulfill this role at Avaya, we have established an Artificial Intelligence Enablement Committee, with executive sponsorship.

The ethical and legal risks associated with AI implementation are significant, but manageable with the right strategies and frameworks. By understanding these risks and taking proactive measures, companies can harness the power of AI to enhance customer and employee experiences while safeguarding their business against potential pitfalls.

To learn more about Avaya’s AI capabilities across its solutions portfolio, click here.

© Foundry