In 2019, CTDO Next, ATD’s exclusive consortium of talent development leaders shaping the profession’s future, published The Responsibility of TD Professionals in the Ethics of Artificial Intelligence. In this whitepaper, we argued the need for a code of ethics and outlined many of the steps the TD function should take regarding that code. TD professionals are already leveraging AI for administration, learner support, content development, and more.
More than 1,000 technology leaders and researchers recently called for a pause in AI development citing “profound risks to society and humanity,” based largely on the same ethical concerns we raised. Whether such a pause occurs, it seems likely that it will fall to organizations to decide how they will utilize the immense and growing power of AI. Therefore, we renew our call to the TD profession to adopt these seven principles.
1. Fairness. We must see that AI systems treat all employees fairly and never affect similarly situated employees or employee groups in different ways. HR experts should be directly involved in the design and selection process and all deployment decisions. Training should precede the deployment of any AI system.
2. Inclusiveness. AI systems should empower everyone. They must be equally accessible and comprehensible to all employees regardless of disabilities, race, gender, orientation, or cultural differences. And we must ensure the biases of the past are not unintentionally built into the future’s AI.
3. Transparency. People should know where and when AI systems are being used and understand what they do and how they do it. When AI systems are used to (help) make decisions impacting people’s careers and lives, those affected (including those making the decisions) should understand how those decisions are made and exactly how AI influences them.
4. Accountability. Those who design and deploy AI systems must be accountable for how those systems operate. Clear owners should be identified for all AI instantiations, the processes they support, the results they produce, and the impact of those processes and results on employees. Every part of the organization should receive training to help them understand how AI is used, with updates provided as use increases or changes.
5. Privacy. AI systems should respect privacy. We cannot expect employees to share data about themselves or allow it to be gathered unless they are certain their privacy is protected. Anyone with access to this data or the AI systems that collect it should be trained in data privacy.
6. Security. AI systems should be secure. We must balance between the real value of the information we’d like to have and how confident we are in our ability to protect it. The TD function will naturally have access to significant amounts of data we must keep safe. Critical issues are where we create access points to the data, the “hackability” of our systems, and the people we assign to operate those systems.
7. Reliability. AI systems should perform reliably and safely. Education concerning AI must demonstrate that systems are designed to operate within a clear set of parameters and that we can verify they are behaving as intended under actual operating conditions. Only people can see the blind spots and biases in AI systems, so they must be taught how to spot and correct any unintended behaviors that may surface.
TD’s Critical Role
AI will require much of the talent development function. In addition to the training noted earlier, everyone in the organization will need training in prompting. Our organizations will need people trained to evaluate the quality and suitability of the data and models used to train and operate AI. When AI systems are used to make consequential decisions about employees, we’ll need training to adequately explain how the system operates.We must help create a widely shared understanding of when and how an AI system should seek human input during critical situations and build a robust feedback mechanism that all employees understand so they can easily report performance issues they encounter.
We already train in cyber security and threat awareness, but it must now encompass the risks peculiar to the use of AI. If we truly want to manage this transformation and make AI serve us, it must be implemented based on a deep concern for its human impact.
The TD function should actively share best practices for the design and development of AI systems and implementation practices that share predictable and reliable interoperation with employees. The talent function is uniquely positioned to optimize AI’s impact on the organizations we serve. We can lead the organization in ensuring that our use of this technology meets the highest ethical standards.