AI shapes a lot of the content we see online. Our social media feeds and ads are not random or posted in chronological order, and that’s because AI is curating a lot of it. But who is controlling that? And what does that mean for consumers, creators, and the future of media in general? Today we’re going to talk about the evolution of AI, social media and payment models, and some other fun things like ethical and legal challenges that are happening right now.
But first, here’s a bit about me:
- I’ve worked in the advertising industry (AKA ad tech) in New York City for over a decade. I was the head of marketing at an interactive video and advertising company that made some of the first streaming TV ads. I was also at Yahoo, which is probably like the least cool of all the tech companies at the time. And also at some more obscure tech companies that specialize in online advertising, that use big data and algorithms to deliver what they called the right ad to the right person at the right time. Also known as programmatic advertising.
- More recently, I’ve run some internship programs for college students who wanted experience in UX, online marketing, management consulting, and investment banking.
- I’m, occasionally a VJ at, clubs and events.
- And I’ve also helped some big tech companies interpret the differences between European data laws and policies and how they contrast with American ones.
- Now I run, a streaming media startup based in Lisbon in New York City, focused on AI and decentralized technologies and how they can help generate and distribute media while making sure the system pays creators fairly.
Okay, so do I sound legitimate enough now? Good, let’s do a recap before we go into where all of this is headed. So usually when we talk about the origins of media distribution, people like to start with the printing press in the 1400s, but we’re going to skip that and go ahead to the 20th century and look at the broadcast era: TV, radio, newspapers, and early websites. They pushed out content and it was one to many. You watched, read or listened. And if you had an opinion about it, no one cared!
Then came the Web 2.0 era, the social media era. And at first it was fun and harmless, and it made everyone a creator and a consumer. “Let’s point a bunch of people at each other and see how this turns out.” Sure, nothing’s going to go wrong with that…
The measurements changed to consider things like views, likes, comments, subscribes, and shares, and that became more important than artistic or intellectual expression. Because of this, tracking became the new currency that changed what the priorities were for publishing and publicizing. And now we’re in the age of AI and decentralization, which we will cover in a bit.
Let’s take a look at that time between this early web 2.0 and this new AI and decentralized time period. The shift that happened was that ads stopped becoming hard-coded into the content. The ads became these empty boxes, and companies would fill in these areas through a bidding process to increase the likelihood of clicking on an ad from about 0.1% to 0.2% of ads. That doesn’t sound like a lot, but this kind of targeting made users go from clicking 1 banner ad out of every 1000 times to clicking 2 banner ads out of every 1000 times. And that changed everything. Companies used vast amounts of data to learn more about the users. Sometimes they would look at personal data, sometimes they bought data from online brokers. And most of the time they just collected data by default, looking at things like the user’s geographic location, their declared interests, and the content that they’re posting or watching. This terminology changed to things like “real time bidding” (RTB), “demand side platforms” (DSPs) or “customer data platforms” (CDPs) to describe this. The not-so-nice way to label this is calling it the “surveillance economy”.
The chronological feed was changed because of this. There was too much content to show all the posts from user’s full network. So the social networks started to prioritize content that was considered more engaging. Things like clickbait, ragebait, trolling, misinformation and cat videos. This was one of the first large scale commercial uses of machine learning. It was more simple back then, more like A/ B testing, picking this or picking that, running calculations to pick the best piece of content or ads. There were predefined rules and a lot of human involvement, but the measurement was being recorded and it was possible to know very quickly what was working and what wasn’t.
It’s not like running a TV ad in the 90s and waiting months to. If people bought more of the company’s products, this would happen right away because you’re clicking and buying things and the measurement and the data tracked, and it still works like that today.
Ads are turning into something different as a result of this. And there was a creation of dynamic creative optimization. So instead of a company having hundreds or thousands of ads to show to a consumer and just picking them, they would make a new mix of text, images, videos and sounds to create the perfect ad to show to a customer.
Starting around 2015, this technology got an upgrade. Deep learning models shifted from interpreting the past user data to using AI to predict future behavior from humans and without actually having people involved. It could adjust the content to make sure that generative AI could make new text, images, sounds, etc, based on how people will respond. We’re moving from thousands of options to millions of options. And maybe we’ll get three clicks out of a thousand instead of just one or two!
Some people are worried that the Internet’s going to turn into an AI generated garbage dump that lacks a human soul. I don’t have much to say about that… I hope we don’t have that scenario happen.
That can be scary, but her’ are a couple of nicer thoughts. Most of the internet is free and it’s supported by ads, or users who pay for subscriptions. And that does keep the costs down quite a bit. Also, there’s a lot of legal policies that have emerged, like GDPR in the European Union or CCPA in California to protect consumers privacy and support these economic systems. And while there are more ways to spy on you, there are also new rules coming out every couple of years that are designed to protect you…at least for now.
Now let’s talk about the creator economy. Lots of people post content online on these mainstream platforms. These big companies handle the hosting and distribution, which is super nice of them, and they make it easy. They take all the money mostly from these advertisers that they connect you with, and they pay the creators whatever they feel like. It could be just enough to keep the creators from leaving. This is kind of like how Uber changes the prices and payouts, and they can pay drivers less over time.
These companies make the rules about what they want to prioritize and promote. So they want people who post a certain amount of times a week, in a certain format and with certain keywords. They don’t specifically tell you what they’re looking for, and they’re not always saying the most accurate criteria when they do share that information. Often you have to find out through news leaks. The idea is that you have to keep them happy to earn a living from posting content, if that’s your main source of revenue.
So who’s deciding how this technology is being used? Is it a bunch of nerds in California? Executives in New York, maybe authoritarian governments? Are people just throwing their hands up and saying “AI” without thinking deeper about this?”
I’m really trying not to sound so depressing and menacing. So we’re going to shift this into a more encouraging conversation about where the creative economy is heading:
- Right now, creators have more resources than ever before to create higher quality content
- They can make more content than before
- They could do it faster, and
- They can send it to more people
So there’s a shift in payment models to support creators. Services like Patreon, Bandcamp, Substack, even things like Cameo and OnlyFans. People are also migrating to new social platforms based on the communities they like or the policies that the social platforms have, and that’s kind of new. You can own your content and users without giving it all away. And it’s easier to become big on these new platforms that are growing instead of entering a mainstream, crowded marketplace. So I want you to ask your friendly AI chatbot about the Blue Ocean Strategy and the Fediverse to learn more about that.
Also, there’s this whole Web3/blockchain/crypto world that has a lot of opportunity, even if it’s been associated with some bad people and companies. I’ll take a moment to tell you about the way that licensing can take advantage of these new ideas. So here’s an example.
- In the past, people used to buy physical CDs, very retro. They were relatively expensive and it didn’t matter if you only like two songs of the album. You had to pay the like $18 for it.
- Then as the Internet became popular, young people started to use services like Napster and illegally traded songs to each other for free. The record companies were obviously not happy about this and they knew that they had to change.
- And that’s a big reason why we had services like itunes that came around and were able to sell individual song tracks for $.99 each.
- And then streaming arrived. And it pays like $0.003 per song stream. It’s actually a more complicated system that works with record companies and calculates a bunch of things. And generally a lot of people think it doesn’t reward artists enough.
So where does the decentralized Web3 world come into this? There’s this concept of tokenization where physical or digital items are turned into a token that can be more sophisticated. It could do things like break media into many pieces, or running code and payments directly to people. These are terms that we call “fractionalization” or “smart contracts”.
It also allows us to share, remix, and monetize our media in new ways much faster. We’re entering a time where we can easily split one track of audio or video into a bunch of separate layers and add new thoughts and visuals on top of it. We can potentially reinterpret our favorite media and our own fan fictions. And the flow of money can still track that user engagement. But less money could go to these old school gatekeepers, and more of it can flow to the original creators, along with the remixers, curators and promoters, probably people like you.
The utopian ideal is that a large network of people can share their content and support each other. We could have access to more professional looking tools than we’ve ever had before. And it could save a lot of time and money. And we can make a lot more cool things. We still have to consider who controls the technologies and enforces the rules. It could be big companies, telecoms, governments, maybe some lawless, decentralized group nonprofits or a, hippie commune, who knows? But having basic critical thinking skills goes a long way.
Trying out these new systems, even ones that might fail, is a step in the right direction. AI is going to be integrated into most of these processes. So understanding AI’s role and who controls it isn’t optional. We’re at a really important time that will determine who controls it, who benefits, and you’re a part of that right now. Thanks for listening. Keep learning and creating.
Leave a Reply