Continuing our series on artificial intelligence in sales, let’s take a look at some of the ways AI is being used today, and how a number of those ways could lead us down the wrong path.
I recently ran across a very interesting fact. An artificial intelligence identity, created by an artificial intelligence services company and labeled as an “AI sales assistant”, has her very own LinkedIn profile.
There are a number of issues with this which I will address separately. First, though, it led me to pose a question that’s kind of scary: how many of our social media followers are actually real people?
Today a single individual can have more of an impact on the world than ever before. Prior to the internet, reaching millions of people was far beyond the capacity of most ordinary citizens. Today you can not only potentially reach millions, but tens or hundreds of millions, even billions. There are certainly celebrities with hundreds of millions of Twitter followers, Katy Perry and Justin Bieber being two of them.
Even the U.S. President, in the old days, couldn’t reach that many people. His speeches were broadcast only on certain television or radio channels and were somewhat filtered. Today, though, the President can communicate directly with the whole world online.
But back to our scary thought: What if some of those people you’re reaching aren’t even real? One person could be a digital identity of some kind, or alternatively one real person could have 20,000 profiles. It then becomes difficult to know how many real people you’re actually reaching.
With companies creating phony AI identities, what is really happening in this world, and where are we going?
The Pigeonholing of Humans
Let’s bring up another direction that AI is going, one that nobody appreciates (at least nobody that I know), that we touched on in an earlier article on AI. I’m talking about those automated answering schemes in which a caller can get totally lost pressing this key for billing, or that one for support, or the other one for customer service, and then you’ve got to punch in your customer number, then press keys to answer more questions…and you still never arrive at a real person. When I encounter such a thing I do what many of you probably do–start punching “0” over and over, or saying and repeating “operator!” or “agent!” until a live human actually answers.
Both the phone-answering AI maze and the phony digital social media personality represent the elimination of communication to humans, replacing it with the communication to AI entities of one sort or another. The only way this would actually work, though, is if potential prospects, customers, callers or followers could successfully be placed into fixed categories–which is what AI programmers have attempted to do. This action comes from the very mistaken belief that human behavior can really be categorized. It isn’t true; every human being, even in the same circumstances, behaves differently.
Such beliefs goes back a long way. Shortly after the turn of the 20th century, Austrian economist Joseph Schumpeter broke ties with the Austrian School of Economics simply because the Austrian School maintained that human action cannot be mathematically predicted. Schumpeter moved into mainstream economics, where then as now theorists relied on a theoretical being called Homo Economicus–a representative person that would react to the same economic stimuli the same way, every time.
Only in very drastic conditions will human behavior be the same or mostly the same. Faced with a raging fire, most people are going to run. If a tiger is in the room, most people would be terrified and would try to get out. In non-life-threatening circumstances, though, people behave very differently from one another.
A very interesting demonstration of this truth can be seen just by examining fashion. Since mass production of apparel began around the turn of the last century, the choices available to the average shopper have not been incredibly broad. Within fashion trends, you’ll often see multiple people wearing the same pair of pants, or the same shirt or top, or shoes. The very interesting thing, though: you’ll never see anyone wearing the exact same outfit!
The AI Sales Assistant…and Reality
Let’s now return to the AI sales assistant I spoke of at the beginning of this article. With prospects and customers (I mean the live ones) being as different as they actually are, with an infinite variety of motivations and goals, and companies having an equally infinite variety of needs and wants, how could this AI sales assistant possibly be programmed with enough options to deal with everyone? Yet the company that sells this service makes incredible promises for “her.” She qualifies leads. She engages every lead, including stale and old leads. She converts and upsells. She even recovers lost accounts.
Anyone who has ever worked in real marketing or sales environments will start laughing when they see a demonstration of what this AI assistant supposedly does. First, they show it responding to a lead with an email asking what prompted the person’s interest. Then there’s a follow-up email asking what the person found interesting about the product, and would they be interested in a demo. (It’s almost entertaining that they’re calling this kind of thing “AI” when in actuality these could just as easily be autoreplies written by marketing–no fancy algorithm needed.)
But the reply from the customer is where a salesperson would have the biggest belly laugh–for they show the prospect responding that, sure, they’d love a demo! When I read this, I reacted pretty much like any salesperson would: “Yeah, right!” If it was this easy to get a positive response to outbound emails, the world would be a very different place.
I honestly think, too, that it wouldn’t take prospects very long to figure out that this “assistant” isn’t a real person. People aren’t that stupid.
Before I leave the “AI assistant” behind, I’ll point out that nowhere on this company’s website do they provide a price. I’m sure that’s because the cost is substantial–the kind and amount of programming that goes into AI is not cheap.
Which leads to this question: Will it really work? And would it be worth the major financial outlay for a company to find out? Just based on the very surface examination I have done of this and a number of other AI offerings, there is no way I’d bet my money on it. Marketing dollars are precious, and not worth gambling. I would add that I do have some expertise in this area, for I myself have been involved with programming high-level pattern recognition for over 15 years for one of my own successful products.
Not Ready to Lead
Even in a somewhat simple B2C sale, AI is definitely not ready for prime time. .
Let us take, for example, the consumer purchase of an SUV. First, there’s the selection from the dozen or so different major brands. Then, within a brand, a particular model. Within that particular model, there’s a selection of engine size, interior materials (such as leather), and options such as sunroof and navigation systems. It’s not only a matter of offering choices, though–it’s reading the customer’s subtle indications and changing offers accordingly, which any skilled salesperson would have to do. Again, with the enormous variety of possible prospect reactions, AI could not possibly be programmed to accurately conduct such a sale.
But moving up to a B2B product such as CRM, we’ve really left possible behind when it comes to AI conducting a sale. Not only is every business (even in the same industry) different, but every process in every business is different, too. A CRM is meant to be modeled on a company’s processes–but how could you possible program an AI algorithm to respond to the infinite variety of possibilities in companies shopping for CRM? On top of that, CRM isn’t standardized like an SUV is–many people call applications “CRM” that, strictly speaking, aren’t, such as ticketing and social media systems. So you’d have to program every possible application that might be viewed as a CRM into AI, too.
In Conclusion…
Contrary to what you might think from reading this article, I am not opposed to AI. I have stated in previous articles, and will be elaborating more in future ones, exactly where I think AI stands today, and where it is going–and I believe there are very positive ways it is being and will be used.
For now, though, I would highly advise companies to stick with their proven marketing methods, and not be taken in by AI “fads” until they are truly demonstrated to be effective.
Pipeliner CRM makes the best use of supportive AI. Get your free trial of Pipeliner CRM now.
Comments