How Google’s AI Overview Spreads Misinformation About Travel Nursing

Google has dramatically increased its use of AI Overviews since first launching the feature in May of 2024. This despite the fact that researchers studying Generative AI report hallucination rates as high as 82% on certain topics. Therefore, I thought it would be a good idea to check in on how AI Overviews are performing on the topic of travel nursing. What I found was astonishingly alarming.

In this article, I’ll first share 2 of the horrifyingly misinformative AI Overviews Google served in response to searches about travel nursing. Then, I’ll discuss some general issues related to Generative AI and how they pertain to the topic of travel nursing.

AI Overview Misinforms About Travel Nursing Pay

If you’re familiar with BluePipes, then you know we take the subject of travel nursing pay seriously. I’ve written over 70 articles on the topic since 2012. We’ve even expended significant resources to build the industry’s most powerful travel nursing pay calculator specifically for travel nurses.

So, you could probably imagine my consternation when I saw Google’s AI Overview spout this horse shit at me:

Not only is Google amplifying the single most insidious piece of misinformation about travel nursing pay, but it’s citing me as the source! Ugh….let’s unpack.

Google Says Bill Rate Minus Payrate Equals Profit

First, when I entered the search term “sample travel nursing pay”, Google’s AI Overview barfed out this doozy:

If the bill rate is $90 per hour and the pay rate is $53 per hour, then the agency keeps the difference of $37 per hour as profit.

This statement is so wrong on so many levels. First, there are many additional costs directly attributable to the traveler that come out of the bill rate before we get to the agency’s “profit”. For example, medical benefits, the employer’s portion of FICA, travel stipends, credentialing expenses and non-billable orientation hours are just some of the additional expenses the agency must cover with the bill rate.

Second, Google’s use of the term “profit” here is ambiguous at best and downright stupid at worst. Are they talking about Gross Profit or Net Proft? I ask because, given the way it’s worded, the vast majority of people who read that statement will think they mean Net Profit.

And no, the agency most certainly does not “keep” the difference between the pay rate and the bill rate as profit of any type. In fact, 90% to 98% of that difference would go to cover additional costs of goods sold or operating expenses like recruiter salaries and advertising.

As I mentioned above, I find this to be the single most insidious piece of misinformation about travel nursing pay. It is the reason that everyone thinks agencies are price-gouging.

The reality is that even the largest travel nursing agencies operate with Gross and Net Profit margins that are more in line with your local quickie mart or box retailer than they are with highly profitable tech giants.

That said, travel nursing companies can do quite well at large scale. However, scaling a travel nursing company is very difficult and just because they’re big doesn’t mean they’re gouging.

Google Misrepresents BluePipes

So, where does Google get this nonsense from. Unfortunately, they indicate they got it from BluePipes! This is the most disheartening aspect of all this for me.

Here is what I actually wrote in the article that Google links to as the citation for their misinformation:

When bill rates are high, we routinely receive inquiries and see statements on social media asserting that travel healthcare agencies gouge hospitals and healthcare professionals. In most cases, the evidence is a comparison between the bill rate the agency charges the hospital and the pay rate the agency pays the healthcare professional.

For example, the evidence might be that the bill rate is $90 per hour and the pay rate is $53 per hour. Therefore, the accuser assumes, the agency is keeping the difference of $37 per hour, or 41% of the bill rate, as profit.

This assumption is wholly incorrect. Unfortunately, it is also prevalent and pervasive. It is routinely spread by self-proclaimed experts who have unfortunately become social media influencers.

The truth is that there are many compensation variables and costs that are not reflected in the pay rate or on the pay stub. When we account for these variables, the agencies’ actual profit margins become much more reasonable.

That said, it is still possible for an agency to gouge a traveler or a hospital. We hope that this breakdown of all the variables will help us recognize the difference more easily.

Note how Google used the same exact dollar figures I provided in my example. They just left out the part about this being an example of total inaccuracy!

Contextual Omission on Travel Nursing Taxes

This next example might not look like much. However, it perpetuates another of the most insidious pieces of misinformation in travel nursing. Here is what I got when I searched for “what is the 30-day rule for travel nurses?”:

Let’s set aside the fact that it is not technically accurate to call this an “IRS Guideline” given that it’s actually from a “Revenue Ruling” on a very specific case. Instead, I’m concerned with 2 things here.

First, the vast majority of people reading this would think that they could maintain their tax home by simply paying the bills at home, keeping their personal business at home and returning home for 30 days a year. It leaves out the part about not working in the same metropolitan area for more than 12 months in any rolling 24-month period.

This is the missing piece of information that gets so many travel healthcare professionals in trouble with the IRS. The tax bill for this can be astonishingly high.

So, in this case, the AI Overview omits context that is extremely crucial to getting this issue right.

Contradictory Information on Travel Nursing Taxes

Second, notice how the AI overview lists a question about exceeding the 30-day rule. The answer indicates you would have to pay taxes if you exceeded it.

The opposite is true. The question should ask about falling short of the 30-day rule not exceeding it.

Misinformation or Hallucination

As you can see, AI Overviews have been generating some serious misinformation about travel nursing. However, the AI industry is adamant that we use the term “hallucination” instead of misinformation. Have you ever stopped to ask why?

The reason is buried in the court transcripts for lawsuits against Generative AI companies. You see, the Generative AI companies are trying to straddle reality in ways that humans cannot. Specifically, they are trying to claim that AI is NOT like a human when they are trying to avoid responsibility for outputs while at the same time claiming that AI IS like a human when they are profiting from the exploitation of human-created works for inputs.

For example, when a content creator sues a Generative AI company for copyright infringement, the Generative AI company argues that they’re just like a human reading a bunch of material and storing it for later use. But, when someone sues a Generative AI company for liable or slander, the Generative AI company argues they are nothing like a human, they’re just a bunch of bits, bytes, silicon and aluminum. Therefore, there can be no malice or intent and no liability.

Liability and Probability

So, when one of our frequent readers contacts us and explains how they got misinformation from BluePipes via Google’s AI Overview, which is how I got on this topic in the first place, we have absolutely no recourse. A brand we’ve spent over a decade nurturing is tarnished through no fault of our own.

It’s difficult to determine how often these misinformation responses happen. Research indicates it’s as low as 3% and as high as 82%. The difference depends on the subject matter.

Topics that are fairly straightforward, like math, experience lower rates of misinformation. But topics like law, where judges are able to rule that AI is like a human and also not like a human, experience much higher rates of misinformation.

I’m thinking the topic of travel nursing falls somewhere in the middle to high end of Generative AI’s misinformation probability range. One reason is that, similar to law, there is plenty of ambiguity in the topic of travel nursing. Another reason is that there is a relatively small sample size of accurate information available about travel nursing.

Of course, I, like everyone else, fully expect Generative AI to improve over time. However, for travel nursing, that time may need to come before it reaches the level of Artificial General Intelligence because the incentive for humans to freely share their own knowledge about the topic for AI to ingest and regurgitate has already become a thing of the past.