stub Wei Xie, COO of ArenaX Labs — Interview Series -
Connect with us


Wei Xie, COO of ArenaX Labs — Interview Series



ArenaX Labs’ first-ever AI-controlled blockchain game, AI Arena, is alive and kicking, and it’s currently making serious moves to broaden the scope of its players’ knowledge by implementing a vast collection of intuitive tools and an “imitation” learning loop. To learn a bit more about it, I decided to reach out to ArenaX Labs COO, Wei Xie.

Thank you for taking the time to speak with us, Wei. Before we delve into the world of ArenaX Labs’ work, please could you introduce yourself to our readers?

Wei: My name is Wei Xie. I’m the co-founder and COO of ArenaX Labs, the company building AI Arena. 

Let’s move on to ArenaX Labs. When was the studio founded, and what inspired you to dip your toes into the world of blockchain gaming?

Wei: ArenaX Labs was founded in 2021. We were inspired to dip our toes into the world of blockchain gaming because we saw the capabilities of what NFTs (non-fungible tokens) could create. To us, NFTs are a general-purpose, primitive technology that allows any type of digital intellectual property to be containerized and made tradable. This enables markets to create value and price discovery for those intellectual properties. This was a game-changer for us. We immediately thought about the ability to tokenize artificial intelligence models, and that's what got us down the path of thinking through how to commercialize tokenized AI through a game.

Let’s talk about your AI-centric game, AI Arena. What exactly is it, and what are some of the core gameplay features that players can expect to discover?

Wei: AI Arena is kind of like a cross between Super Smash Bros and Pokemon. The difference is that the characters in the game are actually artificial intelligence models that are able to learn a wide distribution of potential strategies. It would be like being able to train a Pokemon to fight, but the Pokemon can basically learn anything from you. The core gameplay feature is focused on training, where the process involves a human trainer improving the capabilities of their AI through demonstration, called the imitation learning loop. You teach your AI through demonstration, and over time, you discover more and more effective techniques of training your AI. You're then able to submit your AI into a global competition to compete against other human-trained AIs.

So, how does it work — utilizing AI to compete against other opponents, that is? Please could you walk us through the process of training AI to do your bidding?

Wei: The process is called imitation learning, where your AI learns by imitating you. There are three steps to this process:

  1. Data Collection: This is the demonstration phase. In order for your AI to learn, you have to show it what to do in different situations. You play the game and demonstrate to the AI how you would play, focusing its attention on certain aspects while ignoring others. This creates a dataset for your AI to learn from, and you can create different types of data for specific situations.
  2. Configuration: This step involves tuning and instructing your AI on how to learn the information you just showed it. You can tune it to learn the information more aggressively or in a non-destructive manner, retaining old information while marginally updating based on the new data. It's like adjusting dials and levers in a sports simulation game, where different combinations affect how your AI responds and learns the information.
  3. Inspection: In this environment, you can see into the “brain” of your AI and understand what it has learned, observing changes in behavior based on the latest training. You can see the before-and-after to isolate the impact of the training. This feedback mechanism allows you to understand whether your training is achieving the desired outcomes and strategize on areas for further improvement to make your AI a better fighter.

The process essentially involves demonstrating gameplay to your AI, configuring how it learns that information, and inspecting the results to refine your training approach iteratively.

Are we right in thinking that one of the game’s sole objectives is to broaden gamers’ understanding of AI and how it works in a gaming environment? What are some of the lessons that potential newcomers will find?

Wei: One of the main objectives is to broaden a player's understanding of AI, not necessarily how it works in a gaming environment specifically, but rather to use gaming as an abstraction medium to help people understand something as complex as AI in a more simple and understandable manner. Games serve as an effective abstraction tool or apparatus for this purpose. We use games as a way to help people learn about AI, demystify it, and make the process fun. As for the lessons that newcomers can find, the game loop itself in AI Arena is basically the process that an AI or machine learning engineer or researcher uses every single day in their research and building, tuning, and improving models.

We took the machine learning research process and distilled it into a game. So as you're playing AI Arena, you are learning the fundamentals of how AI works, with a lot of the theory embedded in AI research. You start to internalize and build a strong intuition about the AI research process. The deeper you go in the game and the better you become, the more you'll understand how AIs operate and work. It's really up to the user how far and how deep they want to go down that rabbit hole.

The in-game token, Neuron ($NRN) is also a huge part of the game’s ecosystem. Could you tell us a bit more about some of the NFTs that feature in AI Arena?

Wei: In AI Arena, the NFTs are playable characters, specifically in the competitive league targeting highly competitive players. This cohort of players is playing for $NRN rewards.

It's important to note that not everyone is playing for token rewards. The game is separated into two versions: first, an on-chain version powered by NFTs as playable characters, and second, an off-chain, free-to-play version with no blockchain integration or NFTs, where people can easily play the game through traditional web2 authentication.

Regarding $NRN, it is primarily an in-game utility token. Players are able to stake NRN on their NFTs prior to entering a round of ranked competition. The amount they stake determines the amount of tokens they can earn from the reward pool at the end of the competition. NRN can also be used for in-game purchases and other utilities within the ecosystem. There are also ways to utilize NRN outside of the game's system.

So in summary, the NFTs in AI Arena represent playable characters in the competitive on-chain version, while $NRN acts as the utility token driving the game's economy and reward systems.

So, what’s next for ArenaX Labs? Do you have any plans to evolve AI Arena over the coming months or years? Are there any patch notes or keystone updates en route?

Wei: In the immediate term, the focus is on the $NRN TGE (token generation event), followed shortly after by the mainnet launch of the on-chain version of the game. This will coincide with the launch of the playable NFT character collection.

After that, we are constantly improving the core AI Arena game. Some upcoming updates include releasing new capabilities, expanding the move set, introducing new stage designs, and launching the ability for third parties or individuals to create their own tournaments and play with friends. In the long term, we intend to expand the AI Arena universe into other types of game experiences and derivative games. 

So in summary, the roadmap includes the token launch, mainnet game launch, continuous core gameplay improvements, supporting user-generated content/tournaments, and expanding the AI Arena universe into new game experiences over time. On that last point, while we can’t provide too many details for timeline purposes, something is planned for release later in Q3 this year.

What’s the best way to stay informed about ArenaX Labs’ ongoing efforts to improve AI Arena? Are there any social channels or important newsletters that we can share with our readers?

Wei: Follow us on social media and sign up for our newsletter on the AI Arena website.

Any final words for our readers?

Wei: Join our community at

Thanks for your time, Wei!


For more information on ArenaX Labs’ projects, be sure to check in with the team over on their official social handle here. Alternatively, you can visit the website for additional updates here.

Jord is acting Team Leader at If he isn't blabbering on in his daily listicles, then he's probably out writing fantasy novels or scraping Game Pass of all its slept on indies.