AI voice assistants will cause a flood of fake data

Everyone and their grandma are now working on algorithms to fact-check news. But as manufacturers, marketers and especially as consumers, we are facing a much bigger enemy. It is called artificial intelligence.

This is the present: you run out of milk and go to the store. Or you go online to order. Soon Alexa – Amazon’s AI solution will enter your home, or whatever other system you choose. Everyone seems to be working on one. Like it or not, Voice AI will be the interface of the future. (Incidentally why are all sinister AI systems female? Because of the Adam and Eve story? Note to self: ask anthropologist.)

Now, once you got your new girlfriend up and running and you realize you are out of milk, you can say “Alexa, order milk!” Alexa will perhaps say “which brand?” You could answer “Are there any special offers?” And Alexa will say “at XYZ supermarket, TastesTerrible is having a 2-for-1 sale.” Having never heard of it, you will say “TastesTerrible? Is that a good brand?” And here is where the real kerfuffle begins.

With AI we will rely on automated voice systems to make our purchase decisions, but the AI is based on existing digital information. And that information can be manipulated. How will “TastesTerrible” end up being recommended by Alexa? Either by the number of likes its Facebook page has, or a ‘scientific’ study showing it is the best milk in the world. Or some marketer posting about TastesTerrible in a random foodie forum.

Take this real-life example. Ask Google Home (or simply google on the web) “What is the best motherboard brand?” The answer you will get is this:

“ASUS, MSI and Gigabyte are the 3 best motherboard makers, saying that one or the other are better is personal preference. By its very nature AsRock cannot be the best motherboard brand since its ASUS’s lower end brand, but since it is a derivative of ASUS it does have very good boards, often at very good prices.”

This is a completely arbitrary answer from a September 12, 2010, post on Tom’s Hardware. It is outdated and factually wrong (ASRock is an independent company, and its motherboards often rank better than the other three brands). (Disclosure: ASRock is my client.)

Whether you agree with that statement is beside the point. However you look at it, we are entering the age of marketing through Voice AI. How, as marketers, do we get any product message across, if artificial intelligence systems will repeat any nonsense they find on the net.

Product marketing is becoming be a matter of manipulated data. And no one will get a grip on it initially. Big brands will spend billions in creating data to monopolize the AI systems. Your product, dear startup, not having the clout or money to produce that data, will remain invisible. No amount of Facebook or Google ads will change that. Marketers will ultimately become obsolete. Data manipulators will take their jobs. Hashtag #jobsforthefuture.

Well we are not there yet. But don’t underestimate the impact of artificial intelligence on the marketing business in the long term. Siri & co will soon be everywhere. There are companies working on bringing these AI systems to every corner of your life, for example HomePal. You thought SEO was a hassle? Try doing SEO for Alexa.

The reason why AI is such a problem is that it’s not AI at all. There is nothing artificial about the intelligence Google, IBM, Apple and Amazon are peddling. It’s human sourced intelligence. The misleading post above came from a misinformed human being. Just because Alexa says it doesn’t make it “artificial” or authoritative.

So, when Alexa and her girlfriends recommend a product, they are relying on human sources. Manipulating these sources, in other words “AI-SEO” will be the fastest growing business in the coming years. The crux of the matter is, you can create product comparison studies that prove anything. Just think of clinical studies: the ones that are beneficial to pharma companies are published, the rest is buried. AI based on such skewed data will never be authoritative. Or the organic vs. GMO debate. Who will win the data war – green bloggers or big food brands?

As AI rises, we seem to be destined to drown in fake data. Sooner or later we will come up with fact-checking algorithms for news and data. But that could take some time. And even if we do, nothing in marketing is really objective, as we know. The brands that control the data will control the world.

Enough doom and gloom. Here is the silver lining. Even without fact checking, the more data in the AI system the more objective it will become. Algorithms will improve gradually to weed out the worst offenders. But it will be a painful transition, and for decades to come, brands will fight over data and how to put the right words into Alexa’s mouth.