Mind Readings: Intangibles are Infinite with AI

In this episode, let's talk about intangibles, infinity, AI, and Pepperoni hugging spot. If you missed it, this past week, a uh a synthetic ad made entirely with artificial intelligence, well, almost entirely with artificial intelligence, debuted. A 30-second ad for a fictional pizza chain called Pepperoni Hugging Spot. And in this ad, if you watch it, the text is generated, the voiceover is generated, the images, the video, all that was made by machine. It is very clearly made by machine.

Like when you watch it, you go, this is this is definitely uh machine made. The distortions, the funky looking faces, you name it. Uh, it is definitely not something that you would ever mistake as being a human-led production. But it worked. It worked, and as we've been saying with all things AI, if it's rough now, it'll be less rough in a year, usable in two years, and better than the average human made in three years.

When you look at where GPT 2 was three years ago, its outputs were eh, right? GPT 3, the outputs were usable. It was it was pretty good. GPT 3.5, more than pretty good, and GPT 4, which is what's out today, better than what is average for human beings to create. And seeing this got me thinking, if it's intangible, AI allows it to be infinite.

If it's intangible, AI allows for it to be infinite. Think about this for a second. You have tangible things, right? You know, this this mechanical shaver is a tangible thing. You can hold on to it.

It has a resource cost, a manufacturing cost, a supply chain cost to make this thing. I can't pick up a blog post or five blog posts or a million blog posts in any kind of tangible form. I can't take an image or video or sound. They're intangibles. And what we are seeing in this age of generative AI is that intangibles are something machines can generate.

Now, can they generate them better than humans? Sometimes, I mean, if you take the scale of human capabilities and skills from face rolling on the keyboard to Pulitzer Prize, right? There's definitely a bit of a bell curve there. Um the machines have been able to do better than face rolling for a while. They've been able to do better than the back half of the bell curve, right?

They're now at, I would say beyond the midpoint of the bell curve, beyond that center point, and then nudging towards the front end of the bell curve. Depending on your skills with prompt engineering and stuff, you can get pretty far down that bell curve before you get to a point where no the quality for something that's pure surprise running, still largely the domain of humans. But what machines do that people can't is scale, right? They can scale for anything that is intangible. Words, they can make words like crazy.

They can make images, they can make sounds, they're now making videos. And if it's intangible and it's infinite, that also means that any one given piece of content doesn't have much value by itself, right? Now, there are exceptions, obviously, but take the average of a million blog posts. How many of them are going to generate traffic? How many of them are going to create conversions?

Your stuff individually might be better than average. You might be saying to yourself, well, our content's working. Okay, your content is working. Not everyone's is. And in light of that, if you think about the machines on this bell curve now being able to create content that's better than the midpoint of the bell curve, that me at scale, that means that any economic value from the back half of the bell curve now belongs to the machines.

So when you're thinking about how do we generate value, how do we create value? How do we create something that people be willing to pay for? You have a scale problem with this bell curve, right? The machines allow for an infinite amount of intangible content to be created. And the quality of that is going to be average or below average.

But with each year and each iteration, that quality bar goes up. So how do you make money? How does your company make money? What do you sell? And do you sell something that is tangible?

Do you sell something that is intangible? If you sell something that is intangible, how do you create more value? How do you get people to keep paying for it when machines are creeping up on that area in ways that very few of us predicted, you know, even three years ago. Something like Auto GPT and these other autonomous AI solutions, agent-based AI, was theoretically possible, but we didn't know that it would be that easy to glue a few instances of a like a language model together and just let them do their thing, right? We didn't know that.

We were effectively making autonomous versions of the Sims. Intangibles are infinite. By their very nature, they are infinite, and machines can scale them. So where can you derive value? Value comes in part from scarcity.

There is no scarcity of content anymore. Look on YouTube. Even on YouTube, there's a number of things that are they're still human-led but machine assisted. I was watching some stuff on about Star Trek and the way that people, ordinary people, people with a regular laptop, no need for you know a supercomputer, are generating images and graphics that look so good that 10 years ago that would have been prime time movie or box office uh quality. Now your gaming laptop cranks that out.

You look at what the Unreal Engine can create, and you look at what generative AI can create, we are at a point now where the technology is advancing so quickly, those things that create high quality content, the expense, that exclusivity is going away. Generative AI is making that go away. In some ways, this is good. This democratizes our ability to create high quality content. If you make money on content like movie studios or TV production studios, you should be watching this stuff really, really carefully because it's going to eat into your business model.

Think about this. Suppose you have uh a TV series that was canceled. Looking at you, Netflix and Warrior None. What if you fed the two seasons that existed into a large language model and said extrapolate a season three? And then you fed that to Mid-Journey, and you fed that to 11 Labs, you fed that to all these different tools and said, make a third season.

Regardless of what the IP holder has, make a third season or fourth season or fifth season. Today that would be fairly difficult to do. Not impossible. Pepperoni hugging spot is a good example of that. It's not impossible, but it's more challenging.

In five years' time, that might be just a prompt. Make a third season of Warrior Nun, right? That might just be a prompt. And then the autonomous systems will just glue together all the pieces necessary. And with the way language models are evolving, the quality will probably be as good as what the human beings created.

So think about this. If your value is intangible today, can a machine scale it? If so, what are you doing to create additional value that does not scale? Because value comes from in part from scarcity, and scarcity comes from things that don't scale. What doesn't scale in your business?

Give that some thought. If you like this video, go ahead and hit that subscribe button.