As various forms of artificial intelligence capture the imagination of the broader public, human resources professionals may feel like they are caught in a bit of a whirlwind. While predictive analytics (data-driven approaches to predicting outcomes) have been used in the HR sector for over a decade, we are experiencing rapid development of AI-powered tools for human resources that are changing the entire field at a dizzying rate. While you should be excited about the innovation and development now available at your fingertips, you should also be aware of potential pitfalls in using these tools. Here is a rundown of some of the more popular types of AI-fueled tools available for HR professionals – and some risks to take into account while deploying them.
How is HR Using AI-Powered Tools?
Modern forms of artificial intelligence are particularly useful in human resources contexts when used to do two things in particular: managing and leveraging human analytics, and automating certain processes to increase efficiency and reduce the probability of human error. The specific ways that AI is being used by HR include:
Recruiting
HR can deploy AI tools when applicants are first introduced to the organization. Chatbots can assist job seekers through the application process and assist them in matching their skills with open positions. Larger employers may receive more than 1,000 applicants every day. AI-powered screening programs can take the load of human recruiters by sifting through large numbers of applications and resumes searching for keywords, experience, skills, and other criteria in order to identify qualified candidates that should be given closer consideration. These tools can communicate automatically with candidates about their status in the application process and even direct them towards other job opportunities. AI tools can even help HR leaders identify which candidates are most likely to succeed at an organization using predictive analytics.
Interviewing
Some tools use AI-powered facial scanning during remote interviews to evaluate the applicant by analyzing speech patterns, facial expressions, and eye movements in order to determine whether the candidate would be a good fit for the position. One of the goals in using these tools is to avoid the subjective influences that can impact a hiring decision, including pre-existing biases and prejudices.
Onboarding
A variety of tools are available to ease the process of onboarding new employees – whether being hired individually or as part of a larger scale orientation. You can deploy AI-powered chatbots or other platforms to assist new employees navigate their first days and weeks at your organization, including necessary training exercises.
Performance Management
Employers are using AI-powered analytics to assist with the process of evaluating employees.
- Though performance reviews are intended to appraise employee contributions throughout a particular period of time (e.g., a year), those same reviews should provide insight into both the employee’s current contributions and their potential contributions in the future (for example, evaluating employees for leadership or other advancement potential). As a result, AI-based tools have been developed that can assist companies in capturing all performance related data for each employee and analyzing that data in a manner that provides the company with useful conclusions concerning each employee’s effectiveness and productivity.
- Some other AI tools can assist in analyzing employee engagement on a more continual basis. They can scrape data from employee output, emails, and other digital communications, while also proactively engaging in chats or surveys with employees at various intervals, to gauge employee sentiment.
- AI can aid employers when it comes to learning and development goals by developing personalized learning plans for workers and providing the necessary tools and follow-up to ensure they remain in a continual growth mindset.
- And of course, the entire field of performance analytics is perfectly suited for AI. By reviewing work patterns, supervisor feedback, and a whole host of other available data, some tools can offer the organization insights into an employee’s strengths and areas of improvement.
Again, the goal in utilizing these tools is to remove human subjectivity to the extent possible while simultaneously considering a broader array of performance data in an effort to create a more accurate, meaningful, and fair system of performance management.
Retention
AI tools are also being used to evaluate existing employee engagement in an effort to improve retention (as differentiated from the recruitment-based methods of improving retention described above). These AI applications work in a variety of ways, but more often than not they automate the process of collecting engagement-related survey data from employees, supervisors, and other stakeholders to determine the likelihood or risk that those employees might resign (read more here). Such applications do not necessarily require human monitoring, can be programmed to collect and analyze the data, and can likewise alert HR personnel whenever certain risk factors are identified for certain employees or for employee populations.
These tools can also assist with identifying employees who would be better served by being redeployed on other projects, or who feel as though they are languishing in certain roles and are looking to improve their skillset or reskill entirely (and potentially what skills they would be apt to develop), depending on the circumstances.
Automation
In addition to the AI tools be used to manage, track, and analyze company data, there are also AI tools that have the ability to automate certain HR functions in an effort to improve employee access to the necessary information while simultaneously reducing the number of human resources personnel required to support and assist employees when an HR issue arises.
For example, some companies have also explored using AI-based chat bots in order to assist employees when they have basic questions about their employment or benefits. Providing such tools to employees can also increase efficiency, as HR functionaries can focus their efforts on strategic thinking and more impactful pursuits rather than getting bogged down with repetitive and relatively simple tasks. AI chat bots may be sufficiently advanced to handle inquiries such as benefits, pay, paid time off, and how to initiate processes related to leave or disability accommodation. That said, any decision-making should be left to human resources personnel who have experience in managing such matters.
One quick internet search will reveal the growth of Generative AI tools designed to assist HR professionals with some of the more tedious tasks, including creating job descriptions, employment policies and even employee handbooks. While these tools may provide a foundation, HR professionals must be wary of relying on them to produce a final product. As noted, any document produced by a Generative AI tool should be carefully reviewed, analyzed, and edited by an experienced HR professional.
Potential Pitfalls as You Consider Using AI-Powered Tools in Your Workplace
While there is a lot to gain from use of AI-powered tools in HR, employers should review the tools they are considering implementing with a critical eye. Danger can lurk in many possible areas at the intersection of AI and human capital management.
AI tools can be a helpful resource for employers, but they are not infallible. AI-powered systems are built by humans and use a system of judgment that generally reflects human characteristics. For AI that is “trained” on existing datasets in order to predict outcomes or generate content (e.g., ChatGPT), those datasets necessarily reflect past human decision-making. For example, if a company is seeking to hire individuals who reflect the characteristics of the company’s already-successful employees and is trained using those employees’ data, the existing demographics of that company may impact any results provided by an AI-analytics tool.
Possible Bias in the System
Many of these tools may fall prey to known shortcomings in the technology that can result in disparate impacts on certain racial groups. For example, a recent Scientific American article reported that facial recognition software used by police departments can have difficulty distinguishing people of color from one another, resulting in improper arrests of innocent individuals. These same types of facial recognition errors in the employment context could give rise to liability in the HR content as well. For example, if facial recognition software used in the hiring process is less capable of reading facial expressions from individuals with darker skin, these individuals may be adversely impacted on the basis of race or skin color in violation of state or federal law. And just this week, the EEOC secured a settlement in a first-of-its-kind lawsuit involving an employer’s use of an AI recruitment tool that allegedly automatically rejected female candidates over 55 and male candidates over 60. Similar lawsuits are sure to follow.
Unknown Ghosts in the Machine
Employers who utilize third-party AI-based tools provided by vendors will have little insight into what “powers” the tool. For example, if the model is a “learning” model, what data has it been trained on? Which factors are weighed more heavily? Which are weighed less heavily? Does the system have ways to compensate for or mitigate potential impacts caused by age, disability, or other protected characteristics?
Many providers of AI-based tools will consider answers to questions like these valuable trade secrets and may be unwilling to reveal the “secret ingredients” in their products. But this reticence can ultimately put the employer at a disadvantage in the event of allegations of discrimination, as the employer may be unable to explain why certain decisions were made without a full understanding of the technology. Moreover, it may not always be clear if the tools have been tested or validated to ensure that they are in fact analyzing the factors of import and are doing so without creating a disparate impact on the basis of a protected characteristic, further muddying the waters.
Privacy Concerns
Whenever AI-powered tools are utilized in the employment context, potential privacy concerns can arise. Just because the employer has the ability to collect or purchase data (even if available for sale from third parties) does not mean they should. And obviously, employees can be sensitive toward being surveilled – and poorly implemented monitoring processes can lead to public relations and employee relations nightmares.
Employee privacy is protected by a patchwork of state, federal, and international privacy laws, some of which come with significant consequences for noncompliance. In fact, the California Attorney General recently announced a sweeping investigation specifically into employer compliance with the California Consumer Privacy Act (CCPA). As a result, employers need to ensure they comply with the CCPA and other data privacy laws when utilizing AI-powered tools that rely on capture of employee-provided data
Who is the Decision-Maker?
Finally, any time an employer utilizes AI-based technology to assist in its decision-making, the employer must understand that they must be the final decision-maker – not the AI-powered tool. While these tools and related analytics are powerful and can assist with accurate, bias-free assessments, these tools cannot replace independent judgment or common sense.
Conclusion
Put simply, the AI-revolution is here. And tools harnessing the power of AI will only become more and more commonly available and more commonly used in the employment context. Approaching these tools with the right mindset – and the right expectations – allows employers to maximize their efficiency while mitigating the potential risks that can result.
