In late October 2021, billionaire CEO Mark Zuckerberg changed the focus of his company from social media to the metaverse with the sudden rebranding of Facebook to Meta. Other tech companies would soon pivot to align with Meta in providing virtual and immersive experiences.

Later at the AI Summit Silicon Valley 2021, a more analytical view emerged. Danny Lange, senior vice president of AI at Unity, said that the re-brand should have looked at industrial applications and not solely gaming and video conferencing.

Meanwhile, Unity has undergone a similar business pivot – shifting from its roots in developing engines that power video games to creating virtual tools and platforms for industrial and wider entertainment settings.

A year on from his remarks, Lange sat down with AI Business to reflect on the broader interest in the metaverse since then. The Unity AI chief also reflected on his time at Uber, as well as imparted his views on the tech industry’s newest obsession – generative AI.

The following is an edited transcript of that conversation. You can listen to the full chat in the latest episode of the AI Business Podcast below, or wherever you get your podcasts.

AI Business: A year on from your comments at the AI Summit, in your view how have things progressed in the metaverse?

Danny Lange: A lot of what I said at that event is coming to fruition. I still think that what Mark Zuckerberg and his team came up with was a bit narrow. They are, of course, a social platform. But I thought they were only scratching the surface here.

The way that we see the metaverse at Unity is a much broader environment where you have a digital twin. You can be in (physical) reality, but in parallel is this digital universe that we call the metaverse, and there is an interplay between them.

It can be in a game for entertainment. But it can also be for industrial applications where you have digital representations of robots or a manufacturing floor. And it allows users to run simulations and experiments and enjoy virtual experiences – it is bridging the gap from entertainment and gaming to advanced industrial applications.

AI Business: In recent years, Unity has built itself out of its traditional gaming focus to work across various sectors – including industrial and entertainment. How do you feel Unity’s past work in gaming prepared it for its wider agnostic work?

Lange: We are coming from a very different perspective than most of the rest of the industry. If you think about my own career, I have been at Microsoft, Amazon and Uber – all of which are very enterprise-oriented.

I joined Unity six years ago, and it was a big change for me because the environment was so different. I still remember one of the first meetings at Unity, where I said, ‘we need to turn art into science,’ which is normal in an enterprise environment. And everybody looked at me and replied, ‘What's wrong with art?’ That difference brings an interesting perspective when you start going into other areas.

If you think about gaming, bringing gaming into a manufacturing context or into a robotics context, you suddenly have avatars of humans in there that you can simulate in your gameplay. This is a perspective that a lot of the industry up to now missed. We at Unity bring that different aspect in.

AI Business: In your view, how does the gaming sector differ from other areas in terms of approaching innovation?

Lange: It is much less algorithmic in a sense, because the whole gaming experience is much more open-ended – it is more social, more human-centric, and it opens up from a different perspective, compared to a more classic industrial or enterprise approach (to innovation). That approach gives gaming companies an advantage when you start looking at applying artificial intelligence and machine learning to areas like the metaverse.

Take autonomous vehicles. I was involved in that space when I was at Uber and we have several customers on the Unity platform in that space as well. And what I have learned is that it is much less of a classic robotic navigation challenge, like having a car driving down the road and not hitting other cars. It is harder to have humans crossing the street, or kids playing on the sidewalk and having a car anticipate that.

You can imagine how both developing and testing a vehicle to understand human gestures and behaviors and anticipate whether there is a risk that a ball will be kicked into the street and a child will run after it.

Predicting that is more akin to gameplay. That is very different from a classic mathematical perspective. To have a vehicle predict that, you have to get into full-fledged gameplay that is simulating real-world behavior and even outlandish behavior that is unlikely to happen.

AI Business: Danny, as someone who joined Unity without a previous interest in gaming, what’s your perspective on innovation in this industry coming into it with fresh eyes?

Lange: I think that, pun intended, the game has completely changed. We have seen a revolution in AI, three-dimensional applications, complex simulations and in gameplay itself. It has been through a complete revolution over the last six years. I left Uber to join Unity because I saw an opportunity. Gaming has become a world of content creation, rather than just pure content consumption, ranging from entertainment all the way to industrial applications.

AI Business: What kinds of companies are partnering with Unity? What use cases are you working one?

Lange: One very interesting customer of ours is DeepMind. Their mission is to create artificial general intelligence. It is a very ambitious goal. They have made some interesting progress since they were founded 10 years ago. They are simulating a lot of everyday situations to train AI systems on vast amounts of data. They run Unity at large scale and generate behavioral data for avatars and agents solving tasks.

Another Unity user is OpenAI, their work is making rounds with GPT-3, ChatGPT and the generative AI, DALL-E.

AI Business: On the subject of generative AI, what’s your take?

Lange: Generative AI is at a transition point, and we are crossing a chasm. This is the first time where the broader population sees the magic happening. I am extremely bullish on this concept. I looked first at generative AI back when I was at Uber, where we were using it to generate street signs that would confuse self-driving cars. The car would see the sign as a stop sign but in reality, it was a picture of a cat. A human would not see a stop sign. We would embed pixels in the image to trigger the computer vision algorithm to stop the car

Generative AI is coming through with the likes of Stable Diffusion, MidJourney, DALL-E, and ChatGPT. It has come together in a short time span of the last six to 12 months. It is an indication of an acceleration in this space. When everybody said AI is out (of favor), it is all about blockchain and crypto. It turned out that AI accelerated, and the others had a meltdown instead. I think that we're going to see very exciting things coming up in the next few years.

AI Business: What use cases do you envision generative AI will be applied to?

Lange: There are many applications for generative AI. It has the ability, as in its name, to generate. It removes some of the bottlenecks in creation; the gaming industry has worked on overcoming that bottleneck. A few years ago, the term Unity developer was on the top 10 list of requested jobs on LinkedIn. The gaming community could create a workforce of skilled creators that can create these games. DeepMind, for example, has a large contingent of employees who came out of game companies. But when you move from gaming into a broader set of applications for enterprise business, you run into this bottleneck due to a lack of creators.

Imagine a city council wants a digital twin of the city to do some planning. It is difficult for that facility to hire skilled developers to sit and code in Unity – that is where generative AI comes in. It empowers creators to be more productive and to create vast amounts of data.

And it is going to help us write code. It is very good at writing code. There is going to be a time when developers are going to spend less time writing code and more time expressing what they need. That increased productivity is going to be transformative for applications of the metaverse outside a narrow space, like gaming.

AI Business: What's your take on legal issues surrounding generative AI? Or are you more interested in the actual advances of what this can do?

Lange: The legal issues are a very serious problem. It is an example of a piece of technology moving ahead of where our concepts and understanding of IP is today. You have a machine that consumes much more data than a human can ever consume. It consumes all of the entire internet and all images ever published.

Then based on that, it will generate new imagery that clearly is influenced by what it has seen. It becomes problematic that you generate content that clearly is inspired by human artists. You have to be concerned here because the rip-off is very efficient. It looks very good, it is very efficient, it only takes milliseconds, and you can produce this at a very large scale − everybody can do it.

It is also a challenge not just on the graphical side but also in terms of the code. A company or an individual may have a patent on a software solution to something and now your generative AI just creates this program that is a complete rip-off with no recognition of any rights. That is a problem that is going to haunt this space, just like privacy haunted the consumer algorithmic space.