Artificial intelligence is still in its toddler stage– now able to walk effectively but somewhat clumsily.
There is uncertainty surrounding future regulation, which presents concern for AI’s innovative users. HR, in particular, has to tiptoe somewhat of a fine line as an industry legally obligated to avoid employment biases and discrimination. Unfortunately, AI bias abounds in hiring applications, and with every AI virtue, it seems there is a potential vice to be aware of.
The first legislation specifically aimed to regulate AI was recently passed in Colorado, and the trend is catching on. So, what does this mean for employers?
Well, it means that AI presents some risks, especially when relied upon for hiring. This is by no means an advisory against the use of AI but an exploration of some AI weak spots that may necessitate formal regulation in the future.
What are the risks?
The first risk is just a gentle word of caution to employers who make initial decisions based on resumes alone.
AI Resume Writing
Job seekers increasingly use AI to structure and compose their resumes. While this is not inherently negative, it can mislead employers. A resume’s tone, format, and content can be completely AI-manufactured without misrepresenting any qualifications. Strategic phrasing and composition can give specific candidates an edge over others, even when artificially produced.
In short, gone are the days when the presentation and composition of a resume can reliably speak to the talent and aptitude of a candidate.
Response to Nuance
AI is generally poor at responding to nuance.
While informed by seemingly endless data stores, AI is not a sentient, reasoning system. It is driven by algorithms and struggles wherever ambiguity exists, incapable of reliable discernment.
When it comes to AI hiring tasks such as resume parsing, poor response to nuance can be an issue. Hiring isn’t entirely suited for algorithm-driven processes; such processes shouldn’t be wholly relied upon without due oversight.
Let’s say there are two near-identical resumes (at least from a qualifications standpoint), the largest difference being an age disparity. Which resume will AI select? Well, analysis has shown that some parsing algorithms have favored younger candidates in such situations, which can be seen as discriminatory. The employer did not necessarily introduce this bias; it existed due to algorithmic inadequacies. Regardless, such bias can harm an organization and compound significantly over time.
AI Misuse
This one is simple. AI is meant to better inform decisions, not make them.
AI can be a fantastic HR help when generating job descriptions, customizing job postings, or automating certain aspects of the candidate search (with due oversight). However, as mentioned above, AI is not a decision-making tool. Humans are the decision-makers when it comes to hiring.
To avoid risk in any AI-driven HR process, we would invite hiring professionals to ask themselves, “What decisions, if any, am I handing over to AI?” and evaluate all answers carefully.
Compliance Gaps
As has been seen with cybersecurity and data privacy, we will almost certainly see more and more regulation of AI as the years wear on. As HR professionals are well aware, regulation and risk go hand-in-hand.
As new legislation emerges, and it’s certainly on the horizon, it will be up to employers to track and account for it insomuch as it affects them and their AI utilization. Thus, diligent process documentation could prove very beneficial in time.
Takeaways
AI can be a fantastic tool for HR professionals. However, it’s important to keep the following basics in mind:
- Candidates are more than their resume
- Better data and careful oversight will help employers avoid perpetuating AI biases
- Use caution when relying on AI for compliance or decision-making
- Hiring is best kept human lead and AI-assisted
Feel free to browse our other HR resources.
