Inworld AI drives toward better AI characters in games with grants and Santa demo

Avatars have been around for a lot of video game history, but for most of that history, players would never mistake those characters for the humans they imitate. Inworld AI wants to change that by infusing AI into virtual characters to make them smarter than ever.

And Inworld AI has demoed its technology in an artificially intelligent Santa Claus to show off what it can do. This holiday season, kids around the world can have conversations with virtual Santa as part of a newly launched Inworld Character Arcade, which also includes Elon Musk and Sigmund Freud.

Inworld uses advanced AI to build generative characters whose personalities, thoughts, memories and behaviors are designed to mimic the deeply social nature of human interaction. This area of the tech industry has exploded this year with generative AI art projects that can churn out painted portraits of people or chat programs such as ChatGPT.

“I’m dead focused on how we build this into a new form of expression, a new way of creation, with these characters,” said Kylan Gibbs, chief product officer and cofounder of Inworld AI, in an interview with GamesBeat. “How do we create a completely new medium for interaction? That’s what I’m excited about.”

Gibbs started working on machine learning while at Bain & Co. He worked at DeepMind, which worked on conversational AI using generative models and natural language processing. He started doing machine learning while working at Bain & Co.

“I’m generally passionate about how this new technology can empower creative use cases,” Gibbs said in an interview with GamesBeat. “We create products that not only extend the creativity of people in terms of new inspiration and pushing that but also address how we reduce the latency between imagination and reality.”

San Francisco-based Inworld AI came out of stealth in August after it raised $50 million in funding from a bunch of investors including Intel Capital and Bikraft Ventures. Since that time, the company has begun releasing AI characters like Santa. Inworld AI went through an accelerator program with Disney, and it has started offering grants to game developers who use AI to create game characters.

Virtual Santa

Santa can be powered to learn about the likes, dislikes, habits and Christmas wishes of anyone who uses it, making it ideal for parents to share with kids. Inworld’s Santa can be easily customized to learn about your kids’ favorite books, movies and games, to ensure they get a fully personalized experience. Kids can speak to St. Nick from the comfort and safety of their homes and ask a myriad of questions in real-time, at no cost and with no time limit.

This virtual Santa is one of many that Inworld AI is creating to populate the metaverse of the future. The avatars can be the world’s number one expert on a sports team or know all your world’s history or lore by heart. They can be funny, endearing, brilliant, kind, evil or utterly ridiculous. With Inworld, metaverse developers can create and add villains, heroes, historical figures or just normal, everyday people into their worlds. 

Chatting with an AI character who will listen to your problems, laugh at your jokes, or answer your pressing questions replicates the social nature many players find appealing on successful metaverse platforms, said Gibbs.

Gibbs believes intelligent AI characters can be even more entertaining. Adding them to your metaverse platform casts your users in the middle of an interactive digital world with any character you can imagine. Or that they can imagine — it’s easy to give a game’s users the ability to create or customize their own AI characters to be their sidekicks, best friends, nemeses, or just their fellow users. 

Inworld characters can also easily be integrated into any metaverse technologies and power any avatar you want to build for them. They are avatar agnostic and have integration with Unity, and Unreal Engine, as well as a Node.js SDK. 

In the grant program, Inworld AI will award up to $1 million in grants to innovators who experiment with new tech and create new experiences with AI characters. The judges include Snow Crash author Neal Stephenson, Oculus cofounder Nate Mitchell, former Lucasfilm movie producer Kiri Hart and Inworld AI creative director John Gaeta on board to judge the applications.

The new program will award grants to innovators who want to broaden their creative palette with AI, be at the forefront of immersive entertainment and bring new stores, characters and experiences to
life.

Inworld AI wants more advanced characters to be able to carry on unscripted, open-ended conversations. And it wants those characters to experience emotions, respond to triggers, remember shared lore or brand knowledge, and pursue their own goals. Some of this sounds like it’s more in the future than today, but the generative AI technologies have come on very strong this year.

“With these characters, you’re now changing the way that games and experiences work, where you’re actually inserting something that allows you to adapt to the player or adapt to the world,” Gibbs said.

The company foresees open worlds populated by generative NPCs, RPG simulations or experiences that push emergent gameplay; characters from movies, books, music or TV, translated into interactive entertainment experiences, trailers or promotions; brand activations or e-commerce experiences with digital ambassadors and sales agents; and educational and corporate training and simulations led by digital facilitators and tutors.

“We’re excited to see Inworld characters unlock new experiences in gaming, entertainment, and
the enterprise” said Gibbs. “The creative possibilities of generative AI are endless, so we welcome applications that push the boundaries of storytelling and interactive entertainment. For us, the more innovative and experimental, the better.”

Accelerating Inworld AI

Over at Disney, the company went through the accelerator and created a project dubbed Star Wars Droid Maker in partnership with ILMxLab, part of the Lucasfilm family.

“One of the biggest things that we took away from it was people are very focused on creating these hyper-intelligent AI,” Gibbs said. “But ultimately, what people want are these hyper-engaging and entertaining characters. If you look at a great Pixar movie, it’s all about how they tell a story. That’s been super useful for us.”

Now the company wants to seed its learnings to those who are applying for grants.

The grants are aimed at jumpstarting the ecosystem so creators use more AI characters with real ability to adapt to a situation to make things more interesting for players. The developers can use these AI characters in place of dumb non-player characters (NPCs) so they are more engaging. You can populate entire worlds with these characters.

“That’s where we’re seeing some folks using it and looking at how to populate these virtual worlds,” Gibbs said.

On the other hand, some users want to work with the AI and write a prequel to a book and tell the entire story through a character interaction. And in games, Gibbs hopes AI can deliver emergent outcomes, where the unscripted conversations can lead to much different gameplay than expected.

“The characters can also help provide backstory. So you could stop and talk with a character in The Lord of the Rings or Game of Thrones or Elden Ring style game, where you’re actually digging in with that character into the lore of the world. And so that both helps you develop a relationship with character, but also kind of deepens your relationship with the world,” Gibbs said. “It can nudge people along the narrative.”

The AI moment

As for the popularity of generative AI in 2022, Gibb said that it reminded him of data science when he was first moving into that part of computer science around 2014. Data science was the big term.

“I remember finding myself in this career and it was definitely blowing up,” he said. “This whole generative AI domain is about to change how things work.”

He added, “Previously, machine learning was an unsupervised thing where you gave it examples and it learned those patterns. But now the models are actually internalizing these patterns. And so, as a person, you just need to give it less and less input to get what you want out of it. And the magic of that is opening it up so you don’t need to be a machine learning practitioner anymore to make use of AI.”

As for whether AI can wipe out some creative jobs, Gibbs believes it will enhance the jobs of creators. He looks at ophthalmologists, who use AI machines to come up with prescriptions for patients. Those doctors are now more productive and they can offer wisdom and services on top of those automated prescriptions.

“Generative AI is in a very interesting early phase where people are just discovering what it’s useful for,” he said. “I think that in the next few years, just the ability to create content, both the visual parts of it, the interactive parts of it, is changing rapidly. We’re seeing an awkward phase of it now.”