Artificial intelligence is starting to permeate business systems, including applications used by human resources professionals. AI has the potential to significantly enhance the efficiency and quality of certain HR functions. However, it has a corresponding capacity to harm prospective and existing employees if not adopted and managed properly.
One example of HR use of AI is to summarize and analyze information used to evaluate job applicants, including interview transcripts, writing samples, results from personality tests and skills assessments, publicly available information on the Internet and social media sites, and letters of recommendation, as well as the candidate’s resume and cover letter. Another example of AI use is to aggregate, summarize, and analyze data pertinent to job performance. Unlike the traditional evaluation process, which typically involves supervisors periodically preparing written assessments, AI can be used to aggregate and assimilate an employee’s entire work product, such as all email, customer and co-worker digital interactions, and data created or generated by the employee. HR can then use AI to summarize and analyze the employee’s work effort and outcome, comparing it to set standards and the performance of similarly situated employees across many factors, such as quality, quantity, time to complete, timeliness, etc.
These two illustrations spotlight that HR use of AI has the capacity to harm individuals if not adopted and managed properly. For example, some HR professionals fear that AI will produce unreliable or biased results, or results without rationale. However, an analytics-based AI engine that has access to all pertinent information can produce results that are equally (if not more) reliable and unbiased than a human, who may have hidden biases and be less capable of assimilating all of the pertinent data. Similarly, advanced AI engines are increasingly able to provide rationale for their results, which can be as reliable (if not more reliable) than rationale sometimes offered by humans to justify their decisions.
HR professionals also often fear that AI will produce formulaic outcomes that fail to account for the human and cultural aspects of people and businesses. However, advanced AI tools perform better than might be expected in this regard, particularly as they are trained over time about how HR personnel value these variables. Indeed, HR use of AI must always be managed by an HR professional, and AI outcomes should never be implemented without that human control.
AI for HR (like AI use for all business operations) should be implemented through a planned and structured process and based on certain emerging principles for AI legal compliance. The key aspects of that process and those key principles are as follows.
- Governance: Form an AI governance team for the business. That team will evaluate the business’s existing and prospective uses of AI, and will establish policy and make decisions for the business concerning AI adoption and use.
- Ownership and Control: License AI tools to be used by the business to ensure data inputs remain confidential and cannot be used by the AI developers to train the tools for others. Technologically control which employees are permitted to use AI tools.
- Testing and Auditing: Test and prototype AI tools before deploying them for production to ensure accuracy and reliability of results. After deployment, audit AI results to ensure continuing accuracy and reliability.
- Policy and Training: Implement a policy governing the process for AI adoption and establishing rules for AI use. Train employees about the policy, and technologically limit use of AI for only approved employees and approved job functions.
- Data Integrity: Ensure that only accurate, reliable, comprehensive, and unbiased data is utilized to train and operate AI tools.
- Human Control: Ensure that all AI outcomes are reviewed and questioned by experienced personnel before the results are used for business purposes.
- Transparency and Contracting: Notify customers, employees, consumers, and other constituents how the business uses AI, including when AI is used to generate work product. Incorporate AI use principles into agreements with customer and vendors.
- Assessment and Mitigation: Conduct an assessment to identify the uses of AI that could harm individuals (like HR) or pose dangers to the business or public (like AI use for certain design and engineering functions, and for certain IT and infrastructure systems). Implement measures to mitigate the risks identified in the assessment.
AI is such a transformative technology that businesses failing to integrate it will be competitively disadvantaged. However, that power simultaneously has the capacity to harm if not adopted and managed properly. Businesses need to implement AI through a planned and structured process designed to ensure compliance with the emerging legal principles for AI.