How Post-Learning AI Could Wipe Out the Advertising Industry
People Want Content
Since the creation of the World Wide Web, the world has seen a surge in the demand for content. There used to only be a few options when it came to entertainment. Richard Greenfield explains in his talk “The Future of Sports Media: The War for Consumers’ Time and Attention” that TV was the original “cure for boredom. When people came home from a long day at work or from school or from doing errands, they sat on the couch and turned on the TV. There may have been a few people who would pick up a magazine or turned to the newspaper or radio for entertainment, but TV was definitely the main attraction. The whole TV experience hasn’t really changed much: sure millennials don’t watch as much TV as the previous generations (many of them have never even owned cable) but the experience of sitting on a couch and staring at a screen hasn’t changed. If you look at the evolution of the TV remote, it really hasn’t changed much at all — the buttons are widely the same as they were 20 years ago. The remote on the left is from 1997, and the one on the right is from 2017. Other than the addition of a Netflix button, not much has changed.
What has changed is the number of options people have now to cure their boredom. I tried to explain what Netflix was to my dad and when he finally understood, he said “oh — it’s for when there’s nothing good on TV”. But Netflix, Amazon video, Hulu, and YouTube are just the beginning of it. The Android App Store has 2.8 billion apps and the Apple App Store right behind it with 2.2 billion (1). Over 25% of Apple app store downloads is games(2). As of 2017, 81% of the US population has at least one social media profile(3). The number of social media platforms has been increasing since the creation of the Internet. Currently, the average American is projected to spend an average of 5 years and 4 months on social media throughout their lifetime (4).
One thing that hasn’t changed is the number of hours in a day. An increasing number media types and “cures for boredom” while the number of hours in a day remains constant means a more competitive landscape for media companies and content creators. They are competing for eyes — and for a good reason too. That’s how they get paid.
People Don’t Want Advertising
Advertising has paid for content creation for a long time. Yes, people pay for newspapers and cable (although decreasingly) — but it is advertising that is the main source of revenue for these media companies.
When traditional media companies like the New York Times made the move to digital in the mid to late 90s, they made a huge mistake: they made their content available for free. No one in media companies know how to use the Internet to generate more revenue, so they just posted their content online for anyone to see. Over the years, consumers became accustom to accessing this content online for free, so they came to expect free content online. This poses a problem for companies who create content. They needed to make money somehow, and consumers didn’t want to pay for their content, so they turned to digital advertising.
It’s no secret that consumers today hate online ads: click through rates are absurdly low (typically less than 1%)(5) and consumers are fed up with advertising that blocks their ability to get to their content. Nothing demonstrates these consumer frustrations more than the rise of ad blocking online. In January 2010 there were 21 million desktop browsers globally that had an adblock installed. That number rose to 236 million desktop browsers by December 2016(6). Mobile browsers have seen an even more rapid increase of adblock; in January of 2015 there were 145 million mobile browsers with an adblock installed. In just under two years (December 2016), that number has jumped to 380 million mobile browsers, surpassing desktop browsers (6).
Consumers want content, but they don’t want advertising. This has caused a huge problem for media companies and advertisers alike; brands don’t want their advertising to leave a negative image of the brand and media companies are struggling to find new revenue models to pay for their content creation. The move to native advertising and branded content has demonstrated that there could be ways around this roadblock, but it has also made it clear that consumers do not want advertising, especially online.
Meanwhile, the Rise in Machine Learning and AI Development
With this increasingly digital world comes a huge amount of data; collection of data online has become commonplace. Things like entering your email to gain access to certain content and data-tracking cookies mean that there is more data available than ever before. Marketers can purchase this data to gain a better understanding of their audience, but this amount of raw data is hard to understand. Enter machine learning and artificial intelligence.
Humanity has been trying to create mimic human intelligence really since the late 1930s. With the exception of a few interesting inventions (like Vaucanson’s Duck in 1737), this craze for artificial intelligence started in America in the late 1930s with two robots who were featured at the World’s Fair in 1939: Elektro and Sparko. Since then, science fiction writers have developed guidelines for developing artificial intelligence like in Isaac Asimov’s I, Robot in 1941 and created characters like C3PO and R2D2 in Star Wars in 1977.
Scientists, logicians, and philosophers have worked on creating artificial intelligence and thought about the implications they might have on society. We’ve come a long way from Furby and AIBO robotic pet dog, developing Siri and products like Google Home and Alexa. While scientists like Steven Hawking have expressed their concern for artificial intelligence, the world seems to keep moving forward with machine learning — there are just too many positive implications of this technology, from the convenience of having your house know when to turn on the heat or turn off the lights, to marketers finally being able to make some sense out of the insane amount of data we’ve collected via the internet over these past decades.
We’ve seen some big milestones hit in these past few years, with products like Alexa and Google Home revolutionizing the home, and the Google Pixel2 having features that are more like AI than like a phone. This is just the beginning. For better or for worse, AI development is going to continue to progress, and there are some serious implications of this new tech.
AI Will become Superhuman Post-Learning
To start, let’s define two big buzzwords that are being used somewhat interchangeably right now: artificial intelligence, and machine learning. Artificial intelligence refers to machines that mimic human intelligence. You can think of this as the traditional robot images we’ve come to know like 3CPO or R2D2 from Star Wars. But we have more examples of AI right now that aren’t necessarily robot form. If you have Siri or a “Google Assistant” on your phone, you can ask your smartphone a question and it will answer. It might not answer right, but this progress is still an important step for developing AI. Even Netflix’s recommendations for you are an example of AI.
The term machine learning was coined in 1959 by Arthur Samuel and refers to the process of a machine becoming AI. Imagine if a machine had to go to school: in this case the “school” is exposing it to bunch of data (something we have no shortage of), but the principle is the same. Machine learning is the idea that we should be able to give a machine data and let them learn for themselves.
If you think about AI like a child and machine learning like school, you can better understand how previous AI has progressed and how it may continue to develop.
Through machine learning, AI has started to develop from “smart monkey” to “dumb humans” who could only do one thing or serve one purpose. Now with Google Assistant and progressions in Amazon Alexa, we are starting to move into the “smart human” range of development. This will continue to progressively develop at faster rates, until AI becomes superhuman.
When machines have developed past the state of learning, true superhuman AI will be born, what I call “post-learning AI”. These machines will know their users better than they know themselves and will have a deeper understanding of the world we live in. The timeline is unclear as to when this kind of AI will become a reality, but some experts believe it will be sooner than others think. Experts predict that AI that is capable of performing all human tasks could be only about 40 years away (7).
Despite some people resisting the movement to develop AI, the progression of machine learning isn’t going to stop. In the competitive media landscape people are gladly giving AI a portion of their time. Even kids are interacting with AI and will grow up knowing how to interact with machines similarly to how they interact with people.
AI technology is already pervasive, being in our online accounts, on our phones, and in our home with devices like Alexa. They may not look like the robots in all the sci-fi movies, but AI will be in our lives in the near future.
Purchasing decisions will shift from consumer to post-learning AIs
Once these AIs are post-learning, they will know consumers better than we know ourselves. When that happens, consumer purchasing decisions will shift from the individual to these post-learning AIs. We are already starting to see this shift, especially through Amazon’s dash buttons and Alexa. People don’t want to spend the time to sit down, research a product if they need to, and order it or go to the store to get it. Having an AI know exactly what you need when you need it would be a huge convince to consumers.
These post-learning AIs could take many forms. Right now we’re seeing a lot of voice interfaces with things like Siri and Alexa, but post-learning AIs could be something different. They could be physically modeled after people, like a traditional robot, or integrated into our devices or home like the Star Trek computer. They could even be completely behind-the-scenes and require minimal interaction. It is too soon to tell what form they will take when they get here, but my personal hope is for a Star Trek computer. How awesome would that be?
The End of (Most) Brands
Once the purchasing decisions move from consumers to their post-learning AIs, branding will be irrelevant for one reason: machines don’t care about branding.
In the future, post-learning AI will know more about their user than they know about themselves. While still in the learning phase, AI will watch a user’s actions and start to anticipate needs, and consumers are stubborn. They may say they prefer one brand when another brand’s product could work better for their needs. Post-learning AIs will be able to optimize their user’s experience with products by fulfilling their needs in the most efficient and effective way possible, and consumers will be grateful for it.
With this focus on products, companies will be forced to put their energy back into creating superior products instead of trying to convince the consumer that their brand is better than a competitor, and consumers will let this happen. They will see that the purchasing decisions made by the post-learning AIs actually benefit their lifestyle or make their lives easier, regardless of the brand the AI decides to purchase.
We are already seeing a push back against big brands. The rise of organics and the growing demand for natural food products has made waves in the food industry and is an example of this. The swift growth of health expenditures in the U.S. after World War II started a national health craze, only more recently becoming more specifically against large food brands (8).
Another manifestation of this mega-trend is Brandless, an online membership-based store that focuses on cutting out costs that traditional branding cause. They call these expenses “BrandTax” and aim to cut out these unnecessary costs by getting their products straight from the sources. This new startup with $50 million in funding is just one example of the push away from big CPG (consumer packaged goods) brands.
So if it is true that people don’t care nearly as much about the name on the box as they do about what’s in it, then most brands (especially CPGs) are in trouble. There are some exceptions. One of the exceptions is the local, hand-made, artisan brands. Like I mentioned before with the rise of farmers markets and locally sourced products, consumers seem to want to support locals and family-owned businesses. Think of it like shopping on Etsy, a platform for vintage and hand-made ecommerce, instead of shopping on Amazon.
Another exception are brands for a cause. I’m not talking about how Yoplait partners with breast cancer awareness campaigns. I’m talking about brands like TOMS, who gives away one pair of shoes to people in need for every pair purchased. That’s more than some corporate “give back” policy. Or brands like Tito’s Vodka. Tito’s isn’t really a vodka company who helps dogs — they’re dog people who just happen to make vodka. Their cause is genuine and far from a marketing campaign.
The brands that are in trouble are the ones who claim personality without authenticity. Branding for brands’ sake will die. Here’s an example of what the CPG marketplace could look like. The large companies like PepsiCo, P&G, and Coca-Cola would still be around because there will still be a demand for the products they produce, but the market will be far less cluttered with all of the consumer packaged good brands that we see today.
The End of Advertising?
It is estimated that about 75% of advertising dollars in traditional media are spent on brand messaging, while about 80% of online advertising is direct response advertising (9). Direct response messaging is more product-focused and aims to provoke an immediate response from the viewer. If branding is dead, then the money from brand advertising is gone. Let’s do some math:
The global advertising spend on all media in 2016 was about $493 billion (10).
About $178 billion of that was digital, and $315 billion was traditional media (10).
If brand advertising dies, that’s 75% of traditional media ad revenue, about $236.25 billion, and about 20% of digital media ad revenue, about $35.6 billion, gone.
Just like that, the $272.75 billion (noted in blue in the chart) that would have gone to advertising agencies and media companies, gone.
You could argue that we would see a huge increase in direct response advertising after this shift, but would we? If post-learning AIs are making a lot (if not all) of the purchasing decisions for consumers, would direct response advertising even be effective? Maybe, but maybe not, and in the later case, that means all advertising as we know it today would be gone.
Now that may sound like music to your ears if you’re like most people, but for me, a student who is quite literally days away from graduating with a degree in advertising, this is a terrifying thought. So let’s talk about what traditional advertising could turn into.
A New Kind of Advertising
Like I mentioned previously, even direct response advertising may become irrelevant to this future where post-learning AIs make purchasing decisions for consumers. However, I also mentioned that there are brands that could survive this shift in purchasing decisions and those brands will have to have a way to inform the consumer (or the AI) of their products or services. Even if branding as we know it does collapse — which is an intense implication that may never happen, but for the sake of this argument let’s assume is does — there will still be demand for the products those brands create.
So, in this future, advertising look more like a database for AIs. Imagine a huge catalog of all the products available to purchase — from groceries and household necessities, to clothes and entertainment. When a company creates a new product, they need to list it in this “catalog” for the AI to access so they can make the most informed decision.
If there isn’t advertising in the traditional sense in this post-learning AI future, media companies will need to find a new payment model for content creation. Native advertising and content marketing could open a door for advertising agencies to work with media companies to help establish a new payment model.
Advertising agencies could also work to identify brands that could survive the shift of purchasing decisions, like brands for a cause, and work with them to maintain their authenticity, or help larger brands develop a more authentic identity.
Blocking Forces
Machine learning will continue to develop smarter and smarter artificial intelligence. This is the road we started to go down since the industrial revolution, and nothing is going to stop it. With that said, there are a few things that can slow it down.
Government opposition to developing high intelligence AIs are going to be, in my opinion, the biggest blocking force. When AI becomes smart enough to replace a lot of human workers, governments are going to have to refigure their economies. When people are without jobs, they can’t earn money. When people can’t earn money, they can’t spend money. This is going to reshape whole economies and will take years, maybe even decades, to sort out. While this is happening, I think there will be a huge push-back against AI development.
Once AIs become post-learning, there will be many implications. One of these implications is the shift of purchase decisions to AIs, but this adoption could be far from universal. Just as there are some people in the US who refuse to use the Internet (not many by now, but some are still resisting), some consumers won’t to give so much control to a machine. I do think, however, that the convenience will eventually outweigh the reservations and once that starts to happen, major shifts will occur.
The Good
It’s easy to go into a doomsday state of mind when thinking about the development of artificial intelligence. It’s a very Western idea — that machines will take over the world and kill us all — and if you’re interested in more about why our society always seems to jump to that conclusion, there’s a great talk by Genevieve Bell from Tech Cocktail. But instead of going down that rabbit hole, I’m going to end this paper by summarizing all of the good that could come from the shift of purchasing decisions to post-learning AIs.
Post-learning AIs making purchasing decisions for consumers gives all of us more free time to spend how we please. No more researching products, no more reading reviews, no more shopping or ordering things online. More time doing what we want.
Some brands can and will survive the switch of purchasing decisions to post-learning AIs, and I don’t know about you, but those are the kinds of brands that I’d like to support anyways: the ones that actually do good things with the money we give them and the ones that give us a more intimate understanding of the product and what went into making it.
For the brands that don’t survive, the company still can. P&G can still produce products, and they’ll have more money to invest into the development of these products because they don’t have to pay for advertising. This will result in serious innovation and better products.
Consumers won’t have to deal with advertising interrupting their media consumption anymore. Huge win in most people’s eyes.
Advertising agencies can reposition their services into consultancies or help media companies find better payment models.
I Never Thought I’d Be Here
I always thought I’d be safe from a machine taking over my job. I felt secure in my decision to study advertising because machines aren’t close to being able to think strategically or be creative about solutions. I found comfort in the fact that, even though the average person hates advertising (or at least bad advertising), it pays for everything, so it couldn’t possibly go anywhere. I never thought I’d be the one to look at my future and wonder if I’ll be obsolete, and I won’t lie. It’s scary. But there is also a lot of good that could come out of kind of future, and that’s what I chose to focus on.
Footnote Sources
(1) Number of Apps Available in Leading App Stores as of March 2017
(2) Most Popular Apple App Store Categories in October 2017
(3) Percentage of U.S. Population with a Social Media Profile From 2008 to 2017
(4) AdWeek: How Much Time will the Average Person Spend on Social Media During Their Life?
(5) US, Europe and Worldwide Display Ad Clickthrough Rates Statistics Summary
(6) PageFair 2017 Adblock Report
(7) Experts Predict When Artificial Intelligence Will Exceed Human Performance
(8) Major Trends in the U.S. Health Economy since 1950
(9) On Branding Versus Direct Response Advertising
(10) MAGNA December 2016 Global Advertising Forecast, Page 3
