The future of social apps — AI friends versus IRL friends
There’s been a social network for every computing platform and era of the internet. Initially it was email, IRC and Usenet at the dawn of the internet, followed by AOL Instant Messenger and Geocities in the 1990s. Years later, Friendster, MySpace, and Facebook, followed by a few generations of mobile-first social apps like Instagram, Signal, and Whatsapp.
In the new era of generative AI, what’s next? How will social networks change?
Some ideas:
– creative expression, tools, and new platforms
– a new kind of status competition
– a race towards authenticity
– interactive experiences, not just content
– AI friends plus IRL friends
– vtubing versus AI celebrities/dating/etc
Let’s start with creative expression. It’s obvious that social products have been on a long climb towards more and more expressiveness over time. You could see it with smileys on text 🙂 followed by photos and blink tags on MySpace, Facebook, and early social networks. More recently this is led to an explosion of photos and then eventually the short-form video craze.
It’s easy to imagine that generative AI tools will ultimately support every new kind of content creation in social apps. This could mean that blogging and texting our assisted by AI — perhaps to make us sound more clever — and it also means that most photos/videos we see online might be generated from scratch. That all feels straightforward and obvious as a first order effect of generative AI. The weird thing will be that it will be much harder to tell the difference between fiction and reality. Although we have many folks that are speculating about the future of fake news and misinformation in relation to AI, I think there will be just as important of a question to ask — is my friend’s cute travel photo real, or was it created after the fact?
Social networks are often sort of massively multiplayer online games where people compete and earn status competitions based on where they are traveling, who they are wearing, and what they are eating. You score points with likes, comments, and followers, and you can score direct hits on others via quote retweets and downvotes/flags. (Eugene Wei’s essay “Status as Service” claiming that “humans are status-seeking monkeys” is a good place to read more about this POV).
What happens to social networks in a world of supposed AI-driven abundance, when it’s commonplace to see AI-fabricated photos/videos on social media? I think we’ll see a bifurcation: First, we will see a push towards authenticity — with people finding solace in costly signals of reality. Perhaps streaming will get even more popular, since it’s one of the few ways you can show “you’re really there” — or perhaps apps will self-constrain, as Snapchat does in pushing you towards the camera and not your photo library. Second, more weirdly, there will be some that push to explore the fictional element of this creative expression, creating versions of themselves that blend fiction and non-fiction. In the future, we’ll all post mutual fantasies together, in the same way that we read fictional novels (which are, by the way, a lot more popular than non-fiction). And the status-competition of social apps will start to be built on who can tell the more cohesive and creative fiction.
In the same way there is Virtual Reality, emphasizing an experience that takes over your reality, versus Augmented Reality, which only adds to your current experience — I think there will be two kinds of social networks as well. Some will tell stories in the way that celebrity gossip and drama between influencers already does. The others will be real life, with costly signals to prove that it’s real, and focus on the small group of reality around you.
It won’t stop at photos and videos of course. While text-based emojis have evolved into photos, then videos and avatars, what consumers can create will just richer and deeper. I theorize that in the coming decades, with the popularity of gaming and 3D environments more generally, consumers will want to use AI to create shared virtual environments to spend tie with other people. Today, this already happens among GenZ and the gamer generation, where they hang out with their friends in products like Roblox or Fortnite. I think the future of user-generated tools will allow people to not only create and publish photos and videos, but also to create and publish fully immersive 3D environments. Generative AI will underly these platforms, letting people create full-scale worlds, immersive sounds/music, characters, and everything else— and these will be social environments that people can hang out.
While this might sound wild, let me ask you — if the trend of computing has been to make more and more sophisticated forms of content easier to create/publish, then what is the next stage after video? We can easily create, edit, publish text, audio, photos, and video. That can’t be all? What’s next, I think, are either complex pieces of software, of which I count gaming and virtual worlds as a sub-category. I think it’s inevitable that we expand the leverage we give to consumers, and they use it to build new creative and social environments.
In addition to all of these forms of creative expression, it’s interesting to ask what types of humans and AI will you meet on social networks. It seems inevitable that you will spend a lot of your time on social apps interacting with very smart, interactive bots. The reason for this is pretty simple — in a world where content creation costs are nearly zero from generative AI, what bots will have to say to us will be much more interesting (and more responsive) than other humans. This is already happening today, quite broadly, in gaming, and also in various AI companion startups that have emerged.
Today we make a very clear distinction between other human players and NPCs (non-player characters) who are usually powered by simple AIs. What will happen when the cognitive gaps between humans and NPCs disappear? Or when NPCs are actually more interesting, informative, and compelling than other humans?
I think what will happen here is that we will stop making a distinction between human friends and AI friends. People will end up finding friendship and even relationships their AI companions. And I say companions, plural, because perhaps instead of one-on-one interactions you may be have an entire friendship circle that is mostly online and mostly AI. I say this with great confidence, because humans already have a number of parasocial relationships with celebrities and social media influencers. People already deeply bond with streamers and Onlyfans creators even though they are unlikely to ever meet them in person. AI companions are likely to be the same, but even more responsive and with more time, so that you feel like you’re having a 1:1 relationship — or a bunch of them, in different forms of friendship.
Just a decade ago, people were embarrassed to admit that they admit their significant other in an online dating app. It was stigmatized that you would meet somebody off the Internet, because it implied that you simply could not meet anyone in person. I think the same will happen for AI dating. If one of your friends told you today that they had an AI girlfriend, it would be similarly stigmatizing. I predict in a number of years it will be an acceptable social construct. Perhaps it will be a phase of life thing, as peoples’ lifestyles and friendships change so much between high school, college, and the real world, or maybe it will be a ubiquitous phenomenon. But it will definitely be there, and much more mainstream than we might think.
Returning to my earlier point, perhaps we will think of our future social lives as Augmented versus Virtual — one where technology/AI will help us have better and stronger relationships. The other, where AI will help us replace our relationships with companion ones. And perhaps people will jump between them, as we jump between movies/TV and reality, or fiction and non-fiction. And just as fictional novels and films don’t confuse people as to what’s real and not real, perhaps we will be able to handle this dichotomy as well.
Source by andrew chen
