AI in HR has been a hot topic for a while. What started as pure hype is now a reality with many companies using this technology for recruiting, assessments, onboarding, and employee experience.
While there's a lot to learn, one of the core characteristics of AI is that it gets better over time.
In that way, it's very similar to a consultant that comes into your business with a set of skills, and then learns your business.
The HR consultant lifecycle
A consultant comes in with some baseline knowledge that they've learned from working with other companies.
Let's say you hire someone to help build your employer brand. They've done it dozens of times before and have a lot of general knowledge they can bring to the table.
Over the course of meeting with your team and understanding your recruiting process, they learn the idiosyncrasies of your company. More data = more value.
Their efforts will improve drastically as they learn about your organization, and they'll be an ROI positive influence on your TA spend.
Of course, their value can't be infinite. So, eventually it will plateau. You'll still be getting an ROI by having them around, it just won't be increasing.
How is AI like a consultant?
The key concept behind AI is that it is software that can learn. It may start out really ineffective, but quickly it'll understand how to help your company in more ways.
To see why, let's take the example of a chatbot that is designed to answer common employee questions related to PTO, benefits, etc.
Let's say this chatbot has worked with 100 companies prior, many of which are similar to your business.
Given this background, and a bit of priming, it will start off knowing the answer to very basic employee questions like "what are my benefits?"
This is very straightforward. But, these may only account for 15% of your overall questions. So, the ROI is low to start out with. Just like a new consultant, this AI needs to learn your business!
Over time, it will learn the answer to more intricate questions like "what benefit options do I have in the coming cycle if I'm moving to NY?" And, it will be smart enough to understand this is a similar question to "What are my insurance options if I move to New York?"
As the AI learns, it will increase its ability to answer questions very quickly, and then eventually plateau at the point where it just can't do what a human does.
This learning from one customer kicks off a virtuous cycle where customers gives the AI data, that it derives insights from, which makes it more valuable, and leads to more customers.
For a chatbot that answers employee questions, it may start at 15% of questions answered, get to 30% within a few months, and then plateau at around 70% after 6 months.
70% isn't 100%, but think of all the time you will save (and brain damage - who wants to answer these types of questions?!) with a solution like this.
Given the time savings, this tool is now a very integral part of your business.
This is great from an ROI perspective, however it does mean you should spend a lot of time properly vetting vendors upfront as they will be MUCH harder to replace going forward (just like a consultant that knows your business).
How AI is NOT like a consultant, and won't be for a long time
Where this consultant/AI analogy breaks down is that a consultant isn't infinitely replicable. Meaning, they can only work so many hours in a day/week.
Your AI powered HR bot can work across hundreds of companies at once, and immediately implement learnings into your organization. That is very powerful.
But, humans still win big time when it comes to decision making. AI isn't going to figure out your employee value propositions, build an effective content calendar, or re-design your careers site.
In our HR bot example, a complex employee relations issue isn't going to be solved by a bot any time soon.
However, it is useful to think of AI in this familiar consultant framework to understand the power of a properly trained algorithm. Again, the main caution here is to realize that just like a consultant that knows your business really well, an AI that's been active for a while at your company will be hard to replace!