We’ve seen a lot of what we call “fake AI rebrands” in the last 18 months, and I suspect that as long as AI is a buzzword, we will only see more and more of this. Business leaders are going to have to keep their eyes peeled for these kinds of companies in their midst.
A fake AI rebrand is a company that doesn’t in fact leverage artificial intelligence in any meaningful way, but, nonetheless, rebrands their company around the concept of AI to seem like they’re they’re innovative. In this article, I’ll talk about why companies might do this and which companies do this most frequently in this article. I’ll also talk about exactly how to quickly vet AI companies to parse out the real ones from the fake ones.
A lot of this article is based on our three rules of thumb for cutting through the AI hype, and so readers may want to check out that article in addition to this one.
Why Companies Fake AI Expertise
Lead Generation and Talent Acquisition
A lot of audiences associate artificial intelligence with being “hip,” so to speak. AI is the new buzzword. Most business leaders don’t really know what AI nor the companies that are actually doing it look like. As a result, they’ll see website designs that might look like AI, with pictures of brains or the word “AI” plastered all over the site (like in the header image of this article), and they’ll assume that the company must be legitimate in their claims. Many business leaders might be intrigued by this kind of branding, enough to get on the phone with the company and learn more about them. The branding is a lead generation tactic, but often it’s nothing more.
In addition, job applicants might find the website because they’re looking to work at places that are innovating. Since AI is the new buzzword, they might be looking to work at an AI startup. Again, most applicants, especially the non-technical ones, such as marketers or communications people, might just take the company’s branding at face value and apply on the assumption that the company is legitimate in their claims.
Companies might hype AI because it might get them a little bit of coverage. For example, they might be featured in a list of companies innovating in a certain sector. In reality, all the company did was write “AI” on their homepage, and all of a sudden they’re getting mentioned in Forbes articles by people that are writing about AI who are well-intended but might not know how to vet a company that’s really just pretending to do AI.
Future Plans to Leverage AI
The last reason why companies might lie about using AI is that they don’t feel like they actually are. Oftentimes, companies will rebrand themselves as AI companies only when they have plans to use AI in the future, sometimes the near future. They want to generate excitement about their company on the assumption that they will be leveraging AI soon. The problem is that it takes a lot longer to overhaul your core processes for AI than the company leadership often presumes.
Artificial intelligence talent is remarkably expensive and hard to come by. If a company wants to find people with a robust background in data science and convince them to work for the company, it’s going to take a while. One of the reasons it’s so difficult to procure data science talent is the cultural differences between companies looking to do an AI rebrand and the tech giants, as we’ve outlined in our recent executive guide The AI Advantage of the Tech Giants: Amazon, Facebook, and Google.
Companies looking to rebrand themselves can’t presume that they can hire data scientists and that they will immediately get to work on AI as if it’s that simple. There are two primary factors here at play that prevent something like this from being the case.
(1) A company needs subject-matter experts and people within the company who have at least a conceptual understanding of data science. This is so they can hand problems off to the data scientists in a way that will allow them to get to work. People at the company need to be able to hear feedback from the data scientists and contribute to that conversation if any machine learning model is to be built that can actually solve specific business problems.
Most companies are not investing in this kind of training for their subject-matter experts, and most companies trying to do an AI rebrand have existing employees with no understanding of even the concept that would allow a company to apply artificial intelligence.
In addition, (2) IT personnel at the company also need to be able to understand the concepts of data science. When data scientists build a marketing tool with the help of the subject- matter experts in the marketing department, for example, IT people need to integrate that. They need to manage data infrastructure and the flow of data in entirely new ways.
These two factors require a cultural change to take place before embarking on an AI initiative, let alone an AI rebrand. As a result, when a company hires several PhDs with data science talent thinking they’re going to be able to leverage machine learning right off the bat, they’re probably going to lose those PhDs in the next three or four months. This is because the company won’t know how to articulate problems to these data scientists, and they won’t be able to build anything effective.
They’ll try and find work elsewhere, likely at a company that they know can provide them with meaningful work, such as Google or Amazon.
The fact of the matter is companies need to change their core products before they can call themselves an AI company. The company might have software that helps people keep track of inventory in a warehouse, but to be able to bake artificial intelligence or machine learning into that product is not an easy problem. Executives often underestimate how long that process is going to take and exactly how much cultural change needs to happen for that to occur.
Which Companies Fake AI Expertise?
From our experience, companies that fake AI rebrands are oftentimes those founded sometime in the ’80s or maybe the early 2000s when AI and machine learning were really not that interesting. These are often IT services-oriented companies that are now all of a sudden AI experts despite having dealt with regular software for over a decade. Older technology companies often do the same thing.
These companies are usually somewhere between 30 and 300 people large. This is because it’s harder for larger, more established firms to do this because everyone already knows them for something else. In addition, what they’re doing is already working, so there’s no reason to shock customers by going through a major rebrand in terms of product and services. As such, in general, smaller IT and technology firms are those most likely to undergo the AI rebrand.
They may not have had the kind of dominating success that larger firms had with their services, and so they believe that saying they do AI, even if they’re lying, might make them more appealing to customers or garner them press.
How to Spot Fake AI Rebrands Quickly
Look at Their Talent
One of the primary ways to figure out if a company has undergone a fake AI rebrand is to look at the talent and staff at the company. What often is the case at these companies is they will have maybe one data science person on staff who is formally trained at least to some degree. That’s one person out of around 300. There is a very low ratio of data science talent other talents within the company, and this is a useful signal for determining if a company has undergone an AI rebrand. That said, some companies have ways of fluffing those ratios, and so one can’t stop there.
Companies can fluff their talent ratios by hiring several data scientists, but oftentimes those data scientists are brand new at the company. If a company has six AI personnel out of 100 people at the company, and those six people have been employed at the company for an average of five years, then we can assume the company is in some active way leveraging AI in their product.
Often, the companies that fake AI rebrands will so them when they’ve hired one data scientist and they’ve been at the company for three months. Some data science person’s been there for three months. Three months isn’t enough time to go from a company that doesn’t do AI to one that does, especially when there’s only one data scientist at the company. These companies do the rebrand first and the hiring second.
Also, examine the talent that the company does hire. People can fake their AI expertise on LinkedIn. Although that talent might call themselves a data scientist on LinkedIn, they might have just gotten out of undergrad two years ago. Perhaps they studied English literature and the last company they worked for was Starbucks. That’s an extreme example, but we’ve seen it happen.
Companies will sometimes encourage their new hires to call themselves data scientists or data analysts on LinkedIn to try and fool people who would go and vet them into thinking that they actually employ a robust team of experienced data science and AI personnel, when in fact they don’t.
Similarly, sometimes those who call themselves data scientists at a company will have been regular programmers working with code that isn’t useful for AI. They don’t have evidence of experience working with machine learning at past companies nor evidence of academic backgrounds in AI. What happens is these fake AI rebrand companies just call these people machine learning engineers.
In order to further vet the talent at a purported AI company, look at the leadership. Search for the head of AI or data science at the company and see if they have PhDs in data science, AI, computer science, applied math, or hard sciences like physics. If they don’t have that, check to see if they’ve worked at Google, Amazon, Facebook or another AI powerhouse for three or four years. These are the people who are most likely capable of AI and leading data science teams.
We’ve seen a company take entry-level employees who are akin to marketing analysts and change their titles to “data analyst” all at once. The company then looked like they employed many data scientists, but what they’ve really done is pull people fresh out of undergrad with various degrees and called them data analysts to make themselves seem legitimate.
Look at Their Case Studies
After we’ve taken a look at talent, we also want to take a look at case studies. One thing I really want to make clear here is that the time it takes from an existing company to say they’re doing artificial intelligence to then actually be applying artificial intelligence in their product can be anywhere from 18 – 36 months. It takes a long time for a company to actually bake AI into their existing products. If they have an existing software solution, the likelihood of AI overhauling that product and adding tremendous value to that product six months after the company makes a data science hire is very low.
Oftentimes, when a company blasts out a press release, “We have this AI product. We do amazing things with our AI product,” that means that within 18 – 36 months, there’s a small chance that AI will be part of their product. A lot of business leaders will presume that the press release means that the company is doing AI, but that isn’t the case.
Case studies are a way to figure out if a company is leveraging AI at present. In a case study, we want to see clear evidence of what artificial intelligence is doing and how it’s bringing value to the application. A lot of the time a fake AI rebrand company will say that they’re leveraging AI to help their clients, but when one goes to look at their case studies, all of their them mention the same product they’ve had for the last 10 years. There really isn’t any mention of data, of training algorithms, of applying artificial intelligence. One might see some success stories, which are great, but the company talks about how important AI is to their strategic advantage on their homepage despite the fact that their case studies have nothing to do with artificial intelligence. This indicates a low likelihood that the company is leveraging artificial intelligence.
The fact of the matter is for these companies AI is probably not gonna be part of the product for maybe years to come, and companies like to lie about it and pretend like it’s possible for them today. That’s what marketing departments have to do, but business leaders should not let themselves be fooled by this. They should look at the case studies.
Look at How They Present Their Value Proposition
If a company has artificial intelligence on their homepage and all of the pictures of brains and neural networks, but one can’t figure out on their product page or their LinkedIn page where artificial intelligence is being applied to the problem the company is claiming to solve, then it’s likely they don’t know either.
Companies that are actually using AI know how to say what they’re doing. The marketing space is full of companies claiming to do AI that don’t have the talent to back up their claims. A MarTech company might claim that artificial intelligence can automate your marketing campaigns, but that doesn’t actually explain anything.
What we want to look for is the following:
- What are the software’s inputs?
- What is the software doing with the input?
- What is the software’s intended result?
A marketing AI company that is likely be leveraging AI might hypothetically say:
We work with fashion brands who sell clothing. We take our historical database of related purchases across all the fashion brands that we work with, and we help companies to determine the recommended products that they should suggest next to other products. We help them with email promotion for recommended purchases and checkout promotion for recommended purchases. We do that by finding commonalities in products and helping companies suggest the kinds of products that are most likely to get added to cart and increase that revenue for every transaction, what is called “cart value.” We take the past historical data of related purchases, look at the purchasing behavior of customers, what products were purchased in the same cart with other products, what products were purchased in quick succession with other products, and we use that to determine which products should be suggested to which other products. We take ongoing data from what customers are doing and how they’re responding to our offerings, and we use that to update our model and continuously suggest better, tighter, and more related products per each individual user for both email promotion and for checkout promotions. Our job is to increase cart value.
I’ll run through the list for the above example:
- What are the software’s inputs? Purchase data from other client companies and the user company that is buying the software
- What is the software doing with the input? Determining which products correlate well with which other products for a purchase under which circumstances
- What is the software’s intended result? Improving cart value
Now, I want to make it very clear. If a company articulates a very strong case for its software’s inputs, what it does, and its intended result, it doesn’t mean that the company is doing AI. Refer to the talent on the company’s team for the strongest evidence that a company is or isn’t leveraging AI. That said, business leaders will want to look at these descriptive factors as well. If all the company says are high-level buzzwords, the company is very likely faking their AI expertise because the company themselves likely doesn’t know what their purported AI does in their product.
Concluding Thoughts on Fake AI Rebrands
- Look at the talent.
- Look at the case studies.
- Look at how they describe their value proposition.
If a company has great talent, case studies that clearly show what artificial intelligence is doing in their product, and an explanation of how AI works into their value proposition on their website, that company might actually be doing AI.
I really hope that these tips end up being useful for business leaders in terms of sussing out the fake Ai companies from the real. I want to make clear, however, that we’ve never advised people to work with companies on the basis of them doing AI. It’s misguided to search for an AI company to solve one’s business problems just because they’re doing AI. We don’t want people to make decisions based on those criteria. If the ROI is high, if the evidence of results is higher, that is generally how business leaders should make their decisions.
There are some applications like machine vision for medical diagnostics that necessarily require AI, but we advise against looking at, say, a marketing company and assuming that what they’re doing is better or more powerful than another company just because they say they’re doing AI. Many of these companies have gone through a fake AI rebrand. That said, they might have a great track record of servicing business problems in which case, we can call them out on their claims to leveraging AI, but ultimately, if they have a record of success, they still might be worth buying from. Business leaders should not get the idea that they should be looking solely for AI vendors to solve their business problems; they should work with those companies that can drive them the highest ROI regardless of if those companies use AI or not.
Header Image Credit: Towards Data Science