Ethics in AI and business – what’s the problem?

No problem at all right? Let the machines run rampant and make our decisions for us, what could possibly go wrong?

We know full well that artificial intelligence (AI) has already revolutionised the way businesses operate through the massive impact it has had on the people who run the businesses.

When automation tools make predictions, automate tasks, and provide insights that help businesses improve accuracy, efficiency, productivity, and profitability, there must be a degree of risk that these outputs “could” be harmful in some way or another to the business or to the people within the business and their livelihoods, if left unchecked.

Of course, we need to talk about ethical issues and concerns in business. AI adopted and adapted businesses must take steps to address these ethical concerns, internally as well as to the face of the customers and markets they serve, to ensure that AI is used in a responsible and ethical way.

This will no doubt have implications on the redesign of business policies and procedures as well as the positioning of regulatory backdrops and Enterprise Content Management.

So, what basic steps can business take to promote ethical AI?

1. Business can create an ethical AI policy, a defined and documented set of guidelines and principles that organisations can adhere to, to ensure that their AI protocols and systems are developed, used, and deployed in a reasonable, responsible, and ethical manner. The policy will typically address items such as fairness, accessibility, transparency, accountability, privacy, and safety.

• Fairness: – AI protocols and systems should be fair and impartial and should not discriminate against any individual or group.
• Accessible: – AI protocols and systems should be accessible to all and should not be a hidden or secretive component of the business operations.
• Transparency: AI protocols and systems should be transparent and explainable, so that users can understand how they work and make informed decisions about how to use them. Moreover, users should easily be able to explain the company’s AI and RPA policy to any external party wanting to know more about the business protocols and impacts.
• Accountability: AI protocols systems should be accounted for, and technologists / managers should be accountable for their actions in deploying and maintaining the AI. There should be clear mechanisms for users to hold organisations responsible for any harm caused by their AI systems to internal as well as external parties.
• Privacy: AI protocols and systems should be deployed in the correct manner, to ensure respect for user privacy and the systems should never collect, use, or store personal data without the user’s consent and without a clear explanation or description as to how the data is being used and stored.
• Safety: AI protocols and systems should be safe and secure and should not pose a risk to users or to society.

2. Establish an ethical AI team: – This team should be responsible for overseeing the company’s ethical AI program. An ethical AI team is one that is committed to developing and using AI in a way that is safe, fair, and beneficial to society. Here are some of the key attributes of an ethical AI team:

• Freedom from bias: An ethical AI team will take steps to ensure that their AI systems are not biased against any group of people. The team will ensure the careful collection and cleaning of data using fair algorithms, and monitoring for bias in the results.
• Freedom from risk: An ethical AI team will develop AI systems that are safe to use and that will not pose a risk to people or the environment. This is achieved through using robust algorithms, thoroughly testing systems, and putting agreed safety measures in place and maintaining them.
• Explainability and trustworthiness: An ethical AI team will develop AI systems that are explainable and trustworthy. This ensures that involved people will be able to understand how the systems work and why they make the decisions they do. It also means that the systems should be reliable, and that people should be able to trust them to do what they are designed to do.
• Trackability: An ethical AI team will develop AI systems that are trackable and reportable. The teams will ensure that it is possible to track the data that is used to train the systems, the decisions that the systems make, and the impact that the systems have on people and the environment.

3. Train employees on ethical AI: – Employees who work with AI should be trained on the ethical issues involved in AI and on how to use AI in a responsible way. They should be made aware of the potential fallout and consequences of irresponsible deployment of AI and related technologies.
4. Be transparent about AI: – Businesses should be transparent about how they are using AI and employees as well as customers should be educated on the potential risks as much as the benefits of using the AI. Ideally, in the larger organisations, the deployment of the AI models should be incorporated into the operational risk register.
5. Get feedback from stakeholders: – Businesses should elicit as much feedback as possible from all stakeholders, such as customers, employees, and investors, on how they are using AI, and whether the AI regime is working successfully or not. Leaving the AI unattended or unchecked for any longer period could result in incorrect machine learning and the operation may end up with more pain that gain.

Here are some of the key ethical issues in AI and business, which the adaptation of the above mentioned pointers seeks to diminish:

• Privacy and Data Security: – Any functional AI system will invariably rely on vast amounts of data, raising obvious concerns about how that data is collected, stored, and used. Businesses must ensure that data privacy and security measures are in place to protect sensitive information and prevent unauthorised access or misuse. This includes considerations such as encryption, secure storage practices, and obtaining informed consent from individuals whose data is being used.
• Bias and Discrimination: – Given the machine learning component of the generative and related AI models, any good AI systems will inherit biases from the data they are trained on, leading to discriminatory outcomes. This can result in unfair treatment or exclusion of certain groups of people based on their race, gender, or other protected characteristics. For example, if a facial recognition system may be primarily trained on data of lighter-skinned individuals, it may struggle to accurately recognise or identify individuals with darker skin tones.
• Algorithmic Transparency: – Many AI algorithms are complex and opaque, making it difficult for users and stakeholders to understand how decisions are being made. Lack of transparency can erode trust and hinder accountability, especially in sectors such as finance, healthcare, and law enforcement. Businesses should strive to develop explainable AI systems that provide clear insights into how decisions are reached and what factors contribute to those decisions.
• Employment Disruption: – The increasing automation of tasks through AI technologies can lead to job displacement and unemployment. Businesses have a responsibility to consider the potential impact on employees and communities and explore strategies for reskilling or redeployment. This may involve providing training opportunities or creating new roles that complement AI systems to ensure a smooth transition for affected workers.
• Accountability and Liability: – As AI systems become more autonomous, determining responsibility and liability for their actions becomes challenging. It raises questions about who is accountable for the consequences of AI decisions, especially in situations where harm occurs. Establishing legal frameworks and ethical guidelines can help clarify the responsibilities of both AI developers and users in ensuring safe and responsible use of AI technologies.
• Fair Competition: – Companies with access to advanced AI technologies may gain a competitive advantage over smaller businesses that lack the same resources. This can create an uneven playing field and hinder fair competition, potentially consolidating power in the hands of a few dominant players. Regulatory measures may be necessary to promote fair competition and prevent monopolistic practices that exclude smaller competitors from benefiting from AI advancements.
• Manipulation and Propaganda: – AI-powered algorithms can be used to manipulate public opinion, spread disinformation, or reinforce existing biases. This poses risks to democratic processes and public discourse. Businesses should employ safeguards to detect and prevent the malicious use of AI, such as implementing fact-checking mechanisms, transparent content moderation policies, and responsible advertising practices.
• Unemployment and Economic Inequality: – The automation of jobs through AI can exacerbate economic inequality if the benefits of AI adoption are concentrated in the hands of a few, while many others face job losses or wage stagnation. To address this, businesses can actively invest in workforce development programs, reskilling initiatives, and support local communities affected by job displacement, aiming to mitigate the negative social and economic impacts of AI-driven automation.
• Dual-Use Dilemma: – AI technologies developed for business purposes can also be repurposed for harmful uses, such as surveillance or autonomous weaponry. Companies need to consider the potential dual-use implications of their AI products and services. Implementing strict ethical guidelines, conducting thorough risk assessments, and engaging in responsible supply chain management can help mitigate the potential for AI technologies to be used in unethical or harmful ways.

Addressing these ethical issues requires proactive measures, such as responsible data practices, algorithmic fairness, transparency, stakeholder engagement, and robust regulatory frameworks to ensure the responsible development and deployment of AI in business contexts.

 

Brendan van Staaden is an interaction automation and customer experience expert and Managing Executive of MoData Interactive

 

Photo by Kevin Ku via Unsplash