For all the would-be celebrities dreaming of an adoring social media following, Botnet—a new social media simulation for iOS—must seem like heaven. Users enter as the only human in a social network filled with bots. The feel of the platform is somewhere between Facebook’s layout, Instagram’s commenting system, and Twitter’s mayhem.

For a fee, users can have bots troll them, or sling dad jokes at their account. Not only that, but human users are treated like royalty on the platform— every post is responded to by a flood of responses. “Botnet is a social network simulator where you’re the only human along with a million bots who are obsessed with you,” the app’s website claims, “You are famous on Botnet.”

Though the shine from having thousands of fake “followers” is sure to wear off for anyone except the most dedicated narcissist, the app is nevertheless an interesting experiment—and one which shines a light on how thoroughly machine learning and Artificial Intelligence are set to shake up the social media landscape.

Indeed, though Botnet’s battalion of phony followers often stumble in their attempts to approximate normal human conversation—one bot replied to the question “What’s the best restaurant in New York?” with “I think it is because they’re so good and easy to navigate”—the app is nevertheless a showcase of how machine learning can enable AI to produce coherent text. Botnet’s “users” were trained via the GPT-2 algorithm—an AI system that was fed no less than eight million text documents from the web, allowing it to generate remarkably rational text.

While Botnet may be a bit over the top, it has certainly raised interest in how artificial intelligence and algorithms could be used to build the Social Media Platforms of the future.

As online users demand safer, more meaningful online connections, the world of social media platforms seems headed for a seismic shift.

Seeking safety and privacy online

While users are edging away from traditional social media platforms over privacy concerns—more than a quarter of American Facebook users deleted the app in 2019 and one out of 10 closed their accounts entirely—a new breed of social media platforms is springing up, harnessing the potential of artificial intelligence to address these privacy issues.

One such platform is Yubo, which has enjoyed an explosive rise in popularity in recent years. Targeting young people aged 13 to 25, the app now boasts more than 25 million users across the world and last year secured a $12.3 million fundraising round led by Iris Capital and Idinvest Partners.

One of Yubo’s main selling points is how it uses a mixture of machine learning and constructive guidance to address child protection concerns.

Since the app offers a live streaming feature, which allows teenage users to stream video from their own bedrooms, Yubo was determined to avoid users appearing nude or in their underwear. To these ends, the French start-up developed algorithms capable of detecting and intervening when users are wearing less-than-desirable layers of clothing.

Just as the GPT-2 algorithm Botnet runs on had to be trained with millions of pages of text, the algorithms Yubo uses to detect inappropriate attire were painstakingly developed. As Yubo COO Marc-Antoine Durand explained, “We spent nearly a year labeling every reported piece of content to create classifiers. Every time we had a report, a dedicated team had to classify the content into multiple categories (ex: underwear, nudity, drug abuse, etc.).

With this, our algorithms can learn and find patterns.” Even after the algorithm was trained, it requires routine tweaking—Yubo has a full-time data scientist who is constantly tinkering with the algorithm, feeding it more information to fine-tune it. Among the tricky things, the AI system had to learn? The difference between a swimsuit and underwear, which it picked up by learning to recognize things like sand or a swimming pool.

Yubo has also partnered with the age-verification algorithm Yoti, allowing moderators to “age-gate” the community between users in the under-eighteens and over-eighteens. Users must submit their ID credentials to Yoti before joining the platform, and the age recognition software managed to remove some 20,000 false profiles in the first few months of activity.

A new way of moderating content

Meanwhile, a host of other innovators are working on algorithms that could detect trolling, hate speech, or other types of harmful content on social media platforms—all while identifying and highlighting positive or helpful information. Amidst controversy over the psychological toll imposed by working as a Facebook content moderator, the social media titan is increasingly relying on AI to detect and remove offensive content. Meanwhile, researchers at the California Institute of Technology (Caltech) are working on an entirely new technique dubbed GloVe to tag keywords related to incidents of harassment.

Giving creativity its just reward

Artificial intelligence isn’t just useful for weeding out inappropriate content on social media.

AI-driven platforms are even popping up to fill niches in the social media landscape. Creatives have long shared their work on traditional platforms like Instagram and Pinterest—excellent ways to gain exposure for their portfolios, but hard to monetize. It’s a lacuna that New Life.AI— a cyberpunk-style app that looks straight out of Blade Runner—is trying to fill. The futuristic platform says it is determined to redistribute global power away from major city centers by using blockchain and other AI tools to ensure that artists and other online creators get the recognition they deserve. The platform’s ambitious strategy includes deploying an AI system to convert users’ appreciation for a particular post—measured via a metric dubbed Aesthetic Quotient— into a cryptocurrency.

Botnet’s hordes of fawning automatons may be a mere flash in the pan, but artificial intelligence is indeed set to usher in a new era in social media. From helping protect Yubo’s young users to sifting through content that violates Facebook’s terms of service, to boosting creative content on NewLife.AI, artificial intelligence and machine learning are likely to drive social media innovation this decade.