Establishing & Maximising an AI Ethics Counsel within your organisation

The adoption and growth of Artificial Intelligence (AI) has been rapid, with the technology becoming increasingly entrenched in various areas of business from sales to marketing and customer support.

While AI capabilities appear to be limitless, it's important for well-meaning organisations to be aware of the risks and potential dangers it may pose. Because of the risks, governments from across the globe are taking proactive measures to regulate AI usage. The EU's forthcoming AI Act, the USA's AI Bill of Rights, and the UK's AI regulation whitepaper are all examples of this.

To help maximise the benefits of AI while operating on safe footing, it is important to establish an AI ethics counsel within your organisation. An AI ethics counsel will help guide your organisation's use of AI, and address concerns such as safeguarding customer and business data, determining when not to use AI, exploring use cases for AI, and adhering to AI ethical standards.

In this article, we will explore what a good AI ethics counsel looks like, and how you can implement one in your organisation to help guide your AI journey.

    

What is an AI Ethics counsel?

First off, let’s explore exactly what an AI ethics counsel is. An AI Ethics counsel is a group of people who meet to discuss the legal and ethical concerns around AI, and how those concerns impact your organisation and its use of AI. In addition, an AI ethics group will also strive to provide broad guidance and thought leadership on the use of AI, and best practice for safe deployment.

An effective AI counsel proactively reviews AI processes, and AI use cases to ensure that they comply with ethical standards. AI ethics counsels can take many forms and vary in size, depending on the organisation and the specific use of AI, but their main goal is largely the same: to act as an unbiased entity that strives to guarantee the safe and secure use of AI.

Why you need an AI ethics counsel

AI has brought about a major shift in the way companies operate and interact with customers. However, the challenge with AI is rooted in its complicated mechanics, and it all comes down to data. AI works by merging vast data sets (inputs) with intuitive processing algorithms to generate outputs. This is accomplished by the algorithm detecting patterns within the data inputs, and then generating an output.

Data and AI tools are not perfect. Even well-meaning organisations can fall into the trap of placing too much trust in the data they use for training sets, as well as the resulting outcomes. This can lead to a range of ethical issues, such as:

• Bias and discrimination: AI systems can amplify harmful bias or discriminate against certain groups. Human-bias is well documented and if data inputs contain these biases, discriminatory issues can easily seep into AI outputs and produce outcomes that amplify social inequalities.
• Security: Data training sets are compiled from vast amounts of information from diverse sources. Because of this, there is a concern about the inadvertent use of personal customer data in AI training sets. If AI training data contains sensitive information such as customer names, ages, medical histories, financial details, etc., it could be considered a privacy violation.
• Exposing business information: Similar to the above, AI systems may inadvertently expose sensitive business data in training sets.
• Misinformation: AI-generated content can be a powerful tool for spreading misinformation and manipulating public opinion, whether intentional or not. The low cost of creating AI-generated content is a big factor in this, as it can be easily designed to mislead people – especially if they are unaware that the content is AI-generated.
• Regulation: Governments all over the world are taking a greater interest in AI and seeking to implement regulations or publish guidelines on safe AI usage. An AI ethics counsel will help your organisation stay on top of the regulatory landscape.
• Intellectual property and fairness: Generative AI and intellectual property implications have also received a lot of attention, particularly in the world of content creation. It's important to acknowledge that generative AI can undermine human authorship since it mimics and replicates patterns based on the input data

What does an AI ethics counsel look like?

If you're looking to establish an AI ethics counsel, it's crucial to consider a range of stakeholders. Rather than solely focusing on AI experts, your counsel should comprise specialists from different fields, including academics, human resources, IT, product development, privacy, compliance, security, consumer affairs, and industry analytics. This ensures that the counsel’s decisions are grounded in diverse perspectives.

Here are some additional considerations for the composition of a well-rounded AI ethics counsel:

• Diversity: Ensure the AI ethics counsel team is diverse 
• Understanding technical AI: Ensure at least a few members of the counsel understand the technical side of artificial intelligence
• Collaborative: An AI ethics counsel should ideally be working with other entities that work on AI ethics, such as competitor AI ethics consoles and AI thought leadership groups and organisations
• Educational: An AI ethics counsel should easily be able to communicate their ideas and educate relevant stakeholders and employees.
• Representative: An AI ethics counsel should have representation across all elements of the business, with members that can speak on behalf of customers, as well as the business itself.

How an AI ethics counsel can help with company wide transformation

For organisations that aim to utilise AI, it's critical to establish an AI ethics counsel. The counsel will not only help maintain ethical practices, but also influence the organisation's overall perspective on AI usage from top to bottom. This will naturally filter into the organisation's fabric, influencing how it perceives and approaches AI. This is achieved through counsel duties such as:

• Developing and distributing AI ethical policies and procedures
• Conducting thought leadership sessions on ethical AI
• Reviewing company use cases for AI and analysing their ethical implications
• Ensuring a positive employee relationship with AI through training and upskilling
• Identifying any weaknesses in AI processes and proposing solutions
• Acting as a sounding board for AI purchasing decisions or ideas related to AI
• Ensuring that different populations are represented across AI training data
• Helping to maintain a culture of AI transparency for customers, stakeholders and customers
• Keeping up to date on AI laws and regulations

Key considerations for building a strong AI ethics counsel

Now let’s dive into some practical steps you can consider to help guide the creation of an AI ethics counsel. Keep in mind that every console will differ, based on factors such as industry, company size, and AI use cases.

Here are some factors to consider when building your own AI ethics counsel:

• Responsibility and purpose: It's crucial to establish clear objectives for your AI ethics counsel. Doing so will not only provide your organisation with clear guidelines on how to use the counsel, but it will also ensure the counsel is on the same page as the business. Objectives can range from broad and encompassing to specific, such as overseeing project releases, conducting risk assessments of AI strategy, serving as a point of contact for reporting internal AI issues or even offering commercialisation advice.
• Structure: The structure of the counsel is another factor to consider. We can broadly define the structure of AI ethics counsels as internal or external, both of which have advantages and disadvantages.
  • Internal: An internal counsel, which is part of the company and comprised of employees, will ensure the company has full oversight on the groups activities and the make-up of its members. Internal counsels are generally easier to implement and get up to speed because members will already be familiar with the organisation's structure and goals. However, unless there is strong collaboration, transparency, and trust in place, an internal counsel will likely have a harder time enforcing decisions and speaking out against AI usage if it directly opposes business strategy.
  • External: External counsels can come in various forms, including non-profits, for-profits, contracted knowledgeable individuals, and external stakeholders such as key partners or company advocates. The primary advantages of external counsels are their autonomy and independence. External counsels can prioritise AI ethics and customer privacy over business objectives, and their independence enables them to freely offer suggestions and raise critical ethical issues. An external counsel will have several routes to take when it comes to enforcing ethics standards, ranging from contractual obligations to resigning publicly from the counsel, which can result in a major hit to customer perceptions of the organisation. However, the main disadvantage of an external counsel is the lack of control over the counsel, and the challenge of ensuring the counsel understands the business culture and its inner workings.
  • Membership and size: It's important to also consider how members join the counsel. There are numerous options to choose from, such as direct appointment by the organisation, or letting the council appoint members itself. In situations where the counsel is internal, there may be a voting process to decide who joins the counsel after initial recommendations are made. Regardless of what route is taken to appoint counsel members, it is important to ensure the counsel is diverse, representative of multiple stakeholders, and is thoroughly vetted to ensure the integrity of the members. In terms of size, the number of counsel members should be scaled with the number of AI use cases. While larger counsels will be able to take on more work, it might be harder for them to frequently meet, or reach consensus when issues arise. Conversely, a smaller counsel can offer a more agile service and be easier to manage. However, in situations where counsel members disagree, the limited number of members may lead to a deadlock, making it challenging to reach a decision and move forward.
  • Time commitment and compensation: To ensure that your AI ethics counsel is meeting your organisation's objectives, consider outlining your time commitment expectations. When dealing with an internal counsel, managing workload is more straightforward, where the organisation can define the workload and, for example, request that the counsel meet once per month and publish outcomes the following week. External counsels, on the other hand, often require members to balance their counsel role alongside other obligations or jobs, making it more challenging to define a set schedule. However, members should be expected to meet as often as the AI use cases require, which can be every week or even on an ad-hoc monthly basis during non-busy periods. The more frequently the counsel meet, the better their understanding of the organisation and the deeper their connection. In the same vein, it is also important to consider your compensation structure. Counsels that only require a very light touch could potentially be unpaid, while a full time counsel with high-level qualifications will likely need financial compensation.
  • Education and information sharing: An AI ethics counsel is only as good as the information shared with them. An internal counsel will already have access to relevant information by default, whereas an external counsel relies on the organisation to provide data and information on AI based projects. Here, it is important to ensure that any information shared is represented as factually as possible, as it can be easy to dress-up information so it appears in a better light to counsel members. To help with this, ensure the counsel does not receive information from a single, potentially biased source, rather that they have good working relationships with multiple stakeholders across the business who they can engage with and freely request information.

Conclusion

An AI ethics counsel can be a valuable asset for embedding ethical AI practices within your organisation, and providing an extra added layer of security. Moreover, a counsel can serve as a vital sounding board to discuss all matters related to AI - helping can keep your organisation up-to-date with AI regulations and best practices while ensuring that your organisation prioritises customer privacy and security in all AI use cases.

While the make-up or objectives of an AI ethics counsel may differ from company to company, it is vital to take the time to build a counsel that is suitable for your organisation and reflects your AI use cases, so you can better prepare for your AI journey and remain agile for the future regulatory landscape.

Return to Home