How Pokémon Go made you an unpaid employee for years

Pokémon Go creators told users they wanted them to catch all 1,025 characters in a fun, augmented-reality version of their cities. What they actually wanted was an army of bots to take pictures for them all over the world to help develop their product.

Niantic L, the former Google subsidiary that created Pokémon Go, has used gamers to contribute to its mass library of images for nearly a decade in order to hone its artificial intelligence mapping models.

A country essentially has no choice but to participate in espionage in its own streets.

Users contributed to Niantic’s “Scaniverse” by enabling an option in Pokémon Go that would place the adorable monsters in the user’s real world so that when a user looked at their phone, it appeared as if the Pokémon character was standing on their street, in their park, or at the local store.

Niantic said this feature was “completely optional,” and people had to “visit a specific publicly-accessible location and click to scan.”

Despite the feature being optional, the product’s volume of downloads (anywhere between 600 million to one billion) has allowed for more than enough data to extrapolated.

“We have trained more than 50 million neural networks, with more than 150 trillion parameters, enabling operation in over a million locations,” Niantic boasted in a blog post. “We receive about 1 million fresh scans each week, each containing hundreds of discrete images.”

That blog post was edited after a slew of bad press, and added, “Merely walking around playing our games does not train an AI model.”

It’s not clear however, what exactly a “discrete” image means.

The real product

Niantic’s Large Geospatial Model could seemingly be used as a competitor to, or addition to, Google Street View/Google Maps.

Where Google Maps may be able to show users the front of a monument or the entrance to a park, Pokémon Go trained its AI model with thousands of images — from single locations — to precisely determine the terrain, dimensions, and maneuverability of an area.

The true meaning behind Pokémon Go’s scraping of user data does not seem to be to know where players are going or who they’re with, it’s simply that Niantic seemingly wanted to find a way to have tens of millions of people develop its product for them.

The price for Pokémon licensing pales in comparison to having limitless employees around the world, who love their jobs yet at the same time don’t they are working.

“Pokémon Go is just the beginning,” Return’s James Poulos said. “When it comes to the high-tech grey zone of interactive digital overlays, this fast-evolving frontier mixes and blurs military intelligence and law enforcement data with commercial and recreational applications.”

It seems obvious that Niantic is moving toward using this data for augmented-reality glasses and further personalizing user experience in a new form of consumerism. The same way social media platforms or Amazon use a unique advertising ID to suggest that new pair of mittens, Niantic hopes its AI model will not only “help with navigation” but also guide users through the world, answer questions, and provide “personalized recommendations.”

Having a complex understanding of environments will allegedly give people the opportunity to be more “informed and engaged with their surroundings,” the company also claimed.

Why they need you

Niantic explained that without users, it wouldn’t be able to piece together proper measurements and unseen sides of any given object or location.

For example, Niantic revealed that its geospatial model is unable to properly navigate winding streets in an old European town, where elevation fluctuates and objects have unidentical sides.

The company explained, “Appearance changes based on time of day and season ... the shape of many man-made objects follow specific rules of symmetry or other generic types of layouts — often dependent on the geographic region.”

Much of the task, kicked to the consumer, has been finding a way to see what objects or locations look like from different angles. This “spatial understanding” is nearly impossible from satellite images, street view cameras, or AI models, especially for off-the-road locations.

This inability to visualize “missing parts of a scene” was indeed the missing link for Niantic to properly place objects using augmented reality or even send a robot through unmapped terrain.

Eerily, the latter will help autonomous bots navigate these off-road locations.

“The robots are coming,” Poulos explained. “While citizens will doubtlessly seek out boundaries where humanoid machines don’t tread, in the meantime, they need internal guidelines for navigating the physical world, and they need them fast.”

“Slurping up data provided en masse by unsuspecting augmented-reality players is a logical place to turn,” Poulos added.

Much like Mark Zuckerberg’s Meta, Niantic is hoping the future coincides with AR glasses. Along with that comes the need to be able to recognize and place things in the real world.

The company even referred this connection to the real world as the “future operating system.”

What could go wrong?

When asked about a possible military application for its geospatial model, Niantic’s senior vice president of engineering, Brian McClendon, said he could “definitely see it.”

“I think the question is would there be anything that they would do with it that would be outside of what a consumer or a Bellingcat want to do with it,” he added, according to 404 Media.

One obvious application in war time would be using an enemy’s spatial mapping against them. If one country had its every nook and cranny mapped out, that data could be sold to a foe to develop a complex understanding of that country’s terrain.

This would make a grand strategy far easier to implement for nations with this information that those without such information.

However, McClendon said such use would “obviously” be an issue if it was “adding amplitude to war.”

He then noted the project is “months or even years away” from any kind of product but failed to answer whether or not the user data would be sold.

"There will be important questions that arise and we’ll tackle those responsibly and thoughtfully."

Niantic’s team, unlike many in the upper echelon of the tech space, doesn’t seem to have any direct ties to intelligence agencies; CEO John Hanke, art director Dennis Hwang, director Tatsuo Nomura, and the aforementioned McClendon are all former Google employees and Silicon Valley veterans.

While this data seems poised for nefarious use, as it stands, all signs point toward ruthless capitalism.

"Unfortunately, you're going to see more and more of this in the AI era,” said Josh Centers, editor in chief of Unprepared.life.

“Tech companies have been collecting stockpiles of random data for years, often unsure of what to do with it. Now, the answer is obvious: Feed it to a [language model] and see what it comes up with."

This collection of data gives a multinational corporation seemingly more power, in at least one sense, than a government or standing army. A country essentially has no choice but to participate in espionage in its own streets.

Most countries, anyway; Russia didn’t seem to approve of the notion of Pokémon Go from the start, as in 2016 it pushed out its own version of the app focusing on Russian culture and history.

Niantic eventually pulled Pokémon Go from Russia and Belarus in 2022, allegedly in response to the war in Ukraine.

It’s entirely possible, however, that Russia didn’t want it in the first place. Did it know?