Character.AI – What We Can Learn From Scaling Parasocial Relationships to Millions

Noam Shazeer, the CEO and cofounder of Character.ai, has worked for Google for almost 20 years.

If you work for Google, you’ll quickly learn about the challenges of scaling things to billions of users. AI model architecture, distributed algorithms, and quantization — the most important aspect of accelerating AI was to scale it.

Naturally, many Google employees like Shazeer began to wonder: How big can you scale these models? Can you throw $1B, $1T, or even $100T on training AI models and still get performance benefits?

📈 The Scaling Laws by Noam Shazeer (Founder of Character.AI)

"Some of the big unlocks we’re working on are to just train a bigger, smarter model. The scaling laws are going to take us a pretty long way. The model we’re serving now cost us about $2M worth of compute cycles to train last year. We could probably repeat it for $500K now.

We’re going to launch something tens of IQ points smarter, hopefully, by the end of the year. Smarter, and more accessible, meaning multimodal. Maybe you want to hear a voice and see a face and also just be able to interact with multiple people. When a virtual person is in there with all your friends, it’s like you got elected president. You get the earpiece and you get the whole cabinet of friends or advisers. Or, it’s like you walk into Cheers and everyone knows your name, and they’re glad you came.

There’s a lot we can do to make things more usable. Right now, the thing we’re serving is using a context window of a few thousand tokens, which means your lifelong friend remembers what happened for the last half hour. Still, there are a lot of people who are using it for hours a day. That will make things way better, especially if you can just dump in massive amounts of information. It should be able to know like a billion things about you. The HBM bandwidth is there. It just needs to do it." -- (source, highlights by me)

Well, there need to be massive use cases to warrant these kinds of investments, so Noam found one: Character.AI.

Here’s one example chat with one of the more popular AIs “Levi Ackerman” with 109.1m chats at the time of writing:

The entertainment industry is massive, with roughly $2.3 trillion USD (!) a year in revenue. That’s roughly 10x the revenue of Alphabet, one of the largest companies in the world.

source

One of the secrets of the entertainment industry is that there are billions of lonely people out there. These people build what’s called “parasocial relationship”, i.e., relationship with people who don’t know that you exist.

💡 Parasocial relationships are one-sided relationships where one person extends emotional energy, interest, and time, and the other party, the persona, is completely unaware of the other’s existence. (source)

These are TV stars like Brad Pitt, politicians like Donald Trump, or book characters like Harry Potter.

Everybody builds these parasocial relationships — and it’s the first cool use case for AGI.

That’s the idea of Character.AI anyway — and the success in number of users and website traffic proves this point: they have crossed the 20 million users per month mark and north of 500M monthly visits:

source
💕 First, AI Friends. Second, AI Doctors and Self-Driving Cars

"There was the option to go into lots of different applications, and a lot of them have a lot of overhead and requirements. If you want to launch something that’s a doctor, it’s going to be a lot slower because you want to be really, really, really careful about not providing false information. But friends you can do really fast. It’s just entertainment, it makes things up. That’s a feature.

Essentially, it’s this massive unmet need. It’s very important that the thing kind of feels human and is able to talk about anything. That matches up very well with the generality of large language models. One thing that’s not a problem is making stuff up. I want to push this technology ahead fast because it’s ready for an explosion right now, not in 5 years when we solve all the problems."

 -- (source, highlights by me)

Scaling laws have not met their limits, hinting at the immense potential for AI scaling and its implications for the future.

💡 Recommended: AI Scaling Laws – A Short Primer

The moral of the story is that computation isn’t that expensive. Even if you talked to the largest AI models in the world, the costs of that should be significantly lower than the value of your time. And AI Scaling Laws still apply…

In other words, hold your breadth. We will scale things up by orders of magnitude! 🚀

You can watch the full interview here:

Also make sure to check out our prompt engineering courses to help you stay on the right side of change:

Becoming a Langchain Prompt Engineer with Python – and Build Cool Stuff 🦜🔗

This course is all about getting hands-on with the Python Langchain Library, exploring real-world projects, understanding the basics, and playing with some advanced tools to make AI do some pretty neat things.

👉 Check Out the Full Course and Certify Your Prompt Engineering Skills