Our Statistics professor in college began his course by saying, "You can lie with statistics, and by the end of this course, you'll understand how to do it too." This naturally caused an eruption of objections across the class, with horrified and naive students proclaiming it wasn't possible to lie with statistics. "Ah, just wait. Just wait," he said. "You'll understand by the time we're done with this course."
His goal, as we soon learned, wasn't to teach us to lie. His goal was to get us intensely interested in his class, and it worked. We were instantly hooked. It ended up being one of the most useful-in-all-aspects-of-life classes I ever took. He taught statistics, the importance of integrity in statistics (and in life), how to ask questions, how to ask the right questions, and how to think. Not bad for a 3 to 5 credit class.
Then, cementing that university learning, I ended up working with statistics in the field for a bit. (Fascinating stuff!)
So now at the end of 2024, many companies have large tasks such as open enrollment periods for health benefits, and there's a push in the HR field to use AI to assist in these tasks, either to a large or small degree.
And as it so happens, there are also handy dandy statistics supporting AI. But before your company allots funding for AI, you should know more about these statistics first. Because one survey I found as an example has some major flaws.
Let me be clear up front: I'm not bashing AI at all - it can be incredibly useful. I'm just saying you should understand what the data is actually saying. Know before you spend.
Take a look at this survey. It's reasonably large, 851 people. The summary is:
Over 50% of workers trust AI more than human HR professionals
A bold statement, to be sure. But is it accurate? Are HR professionals becoming obsolete?
Upon closer inspection, the answer is no. (My rule of thumb is never trust a percentage. It's too easy to conveniently twist or even lie about information with a percentage.)
Why?
First off, one study or survey doesn't mean a thing. You need a pile of independent surveys before you have any kind of significance. - One massive survey can also pull a lot of weight, but those are few and far between. (The bigger, the better.)
Second of all, with the survey above, we're not told:
- Which industries were polled
- Geographical area(s) polled
- Job levels in industry that were polled
- Age groups (the answers sound like a younger age bracket)
- Level(s) of experience
- Number of years in companies
A lot of important information is missing. And the biggest factor of all is: who funded the study? That's most important because it significantly sways the results.
If you read the survey, (same link as before) you may have noticed that every single question favors AI over actual HR personnel. The way the questions are framed determines how they're answered.
Case in point, read how this question was asked (in favor of AI), and note how it was answered (in favor of AI):
"What are the areas of most trust when adding AI to the employee experience?
Training 39.6%
Performance Reviews 39.1%
Onboarding 38.3%
Salary Negotiation 33.2%
Career Development 26.4%
Employee Support 26.0%
Benefits Selection 23.8%
Guidance for Requests 19.6%
Conflict Resolution 16.2 %
Other 5.1%"
Those numbers are, perhaps, surprisingly high for AI.
However, if you took that question and changed it to favor HR, to something like, "What areas would you want to be handled exclusively by HR (and not assisted by AI)?" Typically, you'll see very different results from above, favoring HR, when you change the way the question is asked.
The way a question is asked makes a huge difference in surveys. Never underestimate the power of questions.
Now, if you choose to integrate AI, especially to help your HR team with open enrollment or other aspects of their job, go right ahead. Just don't let salespeople schmooze you into spending more, by using surveys or studies that are clearly biased in AI's favor.
Use this information so if you're pursuing AI, you can ask more pointed questions, make better judgements, and have a broader perspective on the whole matter. Statistics really do help make a difference.
In the end, you have the potential to save money, allot more to the HR budget for other things, and do so all with clarity and purpose. Regardless of how little or much you integrate AI with HR.
And always, always, always be wary of percentages.
Have a wonderful last month of 2024. |
