AI ethicists: Balancing pros and cons of Artificial Intelligence

Once considered a means to deliver tremendous business value for large enterprises, Artificial Intelligence (AI) now affects the daily lives of millions of people. AI-based algorithms are employed in practically every industry, including criminal justice, recruitment, news media, manufacturing, banking, military, and law enforcement. However, the game-changing promise of improving efficiency, lowering costs, and speeding up research and development has been dampened by concerns that these sophisticated, opaque systems may do more societal harm than good.

An overwhelming majority of industry experts agree that the unchecked use of AI systems has led to ethical issues in the past few years. In industries such as banking and insurance, it has led to over-dependence on machine-led decisions without adequate disclosure. In healthcare, it has often resulted in the collection, without consent, of personal patient data.

It's evident that the popularity and utility of AI are growing. By 2024, businesses across the globe are predicted to spend more than US$100 billion on AI annually. However, a survey conducted by FICO among AI-focused business leaders across multiple regions revealed that only 6% of the respondents looked at ethically responsible use of AI. This could erode faith in AI and, as a result, hinder its adoption, limiting AI's positive effects. It may also jeopardize a company's reputation and client trust and violate company principles.

In this situation, the role of AI ethicists has begun to become prominent. 

Who is an AI ethicist?

An ethicist possesses a sound knowledge of ethics and can apply the abstract concepts of ethics to real-life situations. An AI ethicist studies the compliance of AI solutions and combines these with the concepts of ethics. Thus, in addition to being knowledgeable in ethics, an AI ethicist also has to be proficient in AI technology. Without understanding the limitations and capacities of AI technology, they will be unable to advise on the same.

AI ethicists do not program the systems. Instead, they have a firm grasp of concepts such as labeling a dataset, the difference between unsupervised and supervised learning, obtaining user consent, etc. They must also have strong communication skills, an understanding of operations and business management, and current policy developments.

Why is it essential to have an AI ethicist in an organization?

AI developers are often tasked with developing highly evolved AI platforms that can offer tremendous business value. While building an AI platform and augmenting it with multiple capabilities, it is possible that the engineers and developers do not account for inherent biases or ethical quandaries that could creep into the development process. These may not only result in downsides for customers or the larger society but may also result in public backlash and regulatory actions against the organization. 

For instance, in 2018, it was discovered that facial recognition systems developed by tech giants, including IBM, were significantly accurate while identifying men with lighter skin but committed errors when it came to darker-skinned women. This led to IBM scaling back its facial recognition technology offering. Similarly, Amazon had to cull an internal recruitment AI tool that was biased against women. 

AI technology comprises guidelines, codes, and principles at the policy and corporate levels that it strives to adhere to. An AI ethicist uses these principles to determine if the deployment, development, and design of AI systems are right or wrong.

The need for a team of AI ethicists

Usually, issues related to AI ethics are primarily dealt with by regulators and lawyers. Even within organizations, those dealing with AI ethics tend to have backgrounds across business, technology, law, marketing, compliance, etc. We can add disciplines like psychology, sociology and philosophy to this list, considering the subject we are dealing with.  

The lack of clarity about the job description is one cause for the shortage of AI ethicists. Many businesses are still having trouble defining an AI ethics manager’s job description and credentials. Yet, it is ill-advised to hand over the responsibility of ensuring AI ethics to a single individual. Therefore, it is highly recommended to create a team of AI ethicists led by a Chief Trust Officer. A Chief Trust Officer could be a senior professional who is a part of the management or leadership team with considerable decision-making prowess. 

The role of AI ethicists

Such a team of AI ethicists would possess the essential skills to add ethical, social, and political perspectives to the development and deployment of AI systems. They should be able to draft governance frameworks, get involved in public advocacy and possess multi-disciplinary work experience.

They may also have the following responsibilities:

1. Prevent AI from crossing ethical boundaries

Although AI ethicists don’t need to know programming or coding, they must have an in-depth understanding and knowledge of certain technical concepts. These can include supervised and unsupervised learning, how user consent is obtained, the machine learning model being used and several other minute details that impact the design, development and deployment of AI systems. This would enable AI ethicists to implement security measures that inhibit AI from crossing ethical boundaries.

2. Upholding accountability

Another predicament faced by AI ethicists is the lack of accountability over decisions taken across the value chain, which could have a detrimental effect once the technology or solution is released. Due to the complexity of an AI system and the number of individuals involved in developing it, affixing responsibility is extremely difficult and convoluted. This challenge is popularly known as the ‘moral responsibility’ gap. This leads to the AI becoming the scapegoat in case of any major issue. Although there are no legal or compliance regulations that hold the AI ethicists responsible for violations, they would have to shoulder responsibility in case of consequences. 

3. Training and mentoring

AI ethicists would have to work with multiple teams across the organization and educate them about ethical AI practices. They would have to draft guidelines and create relevant training programs. AI ethicists would also have to act as a sounding board and mentor employees. 

For instance, IBM has a group of volunteers drawn from its employees, known as the ‘advocacy network’. This network advocates the fostering of a culture that enables the development of responsible, trustworthy and ethical technology. 

Courage in the face of fear

Interpretation of ethical principles is highly subjective and can be done in different ways. In an organizational eco-system that is largely driven by precision, there is no algorithm that determines how ethics must be applied.

There aren’t any clear metrics or indicators that explain how ethics can improve the profit margins of organizations. In several cases, ethics can be viewed as an impediment to innovation or optimization. Therefore, there might be situations when an AI ethicist might have to singlehandedly face shareholders, management and even employees to uphold ethical values at the cost of profit. Hence, it is clear that an AI ethicist would have to be a courageous individual who can firmly uphold interpretations, take responsibility and stand up to significant opposition.

Conclusion

As AI alters the consumer-organization interaction, it is necessary to understand that this relationship will only be as strong as the ethics that underpin it.

Disclaimer

The contents of this article/infographic/picture/video are meant solely for information purposes. The contents are generic in nature and for informational purposes only. It is not a substitute for specific advice in your own circumstances. The information is subject to updation, completion, revision, verification and amendment and the same may change materially. The information is not intended for distribution or use by any person in any jurisdiction where such distribution or use would be contrary to law or regulation or would subject IDFC FIRST Bank or its affiliates to any licensing or registration requirements. IDFC FIRST Bank shall not be responsible for any direct/indirect loss or liability incurred by the reader for taking any financial decisions based on the contents and information mentioned. Please consult your financial advisor before making any financial decision.