All the pieces in society can really feel geared towards optimization – whether or not that’s standardized testing or synthetic intelligence algorithms. We’re taught to know what end result you need to obtain, and discover the trail in direction of getting there.
Kenneth Stanley, a former OpenAI researcher and co-founder of a brand new social media platform known as Maven, has been preaching for years that this methodology of considering is counterproductive, if not outright dangerous. As an alternative of prioritizing aims, Stanley says we must be prioritizing serendipity.
“Generally, as a way to discover these stepping stones that can result in the issues we care about, we’ve to get off the trail of the target and onto the trail of the interesting,” Stanley advised TechCrunch in a video interview. “Serendipity is the other of discovering one thing by aims.”
The concept of looking for novelty for its personal sake began as an algorithmic idea that Stanley research known as open-endedness, a subfield of AI analysis about techniques that “simply preserve producing interesting stuff perpetually.”
“Open-ended techniques are like artificially inventive techniques,” stated Stanley, noting that people, evolution and civilization are all additionally open-ended techniques that proceed to construct on themselves in surprising methods.
This algorithmic perception morphed right into a life philosophy for Stanley. He even wrote a guide about it in 2015 along with his former PhD scholar Joel Lehman known as Why Greatness Can not Be Deliberate. The idea took off, making Stanley one thing of a world point of interest for the brazen concept that, truly, you can simply do issues as a result of they’re interesting, somewhat than as a result of that you must full some acknowledged goal.
However in 2022 whereas main an open-endedness workforce at OpenAI, Stanley stated he was “boiling over with discontent” and “had this epiphany” the place he determined to cease speaking about bringing open-endedness to wider audiences and as an alternative begin doing one thing about it.
What if, he requested himself, he created a “serendipity community,” a system that’s set as much as enhance the likelihood of serendipity, for different folks to take pleasure in?
So he give up his job and set about to create Maven, a social community constructed round an open-ended AI algorithm that evolves to hunt novelty. When signing up, customers choose a collection of subjects to comply with — from neuroscience to parenting — and the algorithm exhibits them posts that align with their pursuits. At this time’s social media algorithms additionally present you belongings you would possibly discover interesting, however the distinction is they’re optimized to maximise consumer engagement, usually by boosting sensationalistic content material, to create extra advert impressions and income. Maven, in contrast, doesn’t simply present you the most well-liked posts on subjects that you simply discover interesting. The algorithm exhibits you posts based mostly on the chance that you simply’d discover them interesting.
Maybe most revolutionary, Maven does away with social media’s present arrange – there aren’t any likes, upvotes, retweets or follows, and there’s no option to amplify content material to the plenty.
As an alternative, when a consumer posts one thing, the algorithm robotically reads the content material and tags it with related pursuits so it exhibits up on these pages. Customers can flip up the serendipity slider to department out past their acknowledged pursuits, and the algorithm working the platform connects customers with associated pursuits. So if, for instance, you’re following conversations about city planning, Maven may also recommend conversations about public transit.
And whereas there’s no option to comply with folks on the platform, you can see and join with different individuals who comply with subjects you’re involved in.
In a number of methods, Maven looks like an antidote to in the present day’s social media, the place the “goal paradox is on full show” as folks fall over themselves to create sensationalist content material that can garner extra consideration and recognition.
“The echo chambers and the toxicity, the narcissism amplification and private branding has gone completely uncontrolled in order that individuals are shedding their soul and turning into manufacturers,” stated Stanley.
The addictive qualities of social media, harm to mental health in adolescents and adults, and skill to polarize nations is effectively documented. These, Stanley says, are the unintended penalties of formidable aims, the end result of constructing recognition a proxy for high quality.
“And then you definitely get all these different issues as a result of upon getting recognition, you’ve perverse incentives,” he stated.
Stanley famous that Maven customers can flag inappropriate content material or misinformation when it pops up, and its AI is actively monitoring for extremely inflammatory, offensive “or worse” content material. He stated Maven can’t repair the nastiness in human nature, however by eliminating the incentives behind sharing such content material, Stanley hopes it may change the “general combination dynamic of how individuals are behaving.”
Some social media firms have tried to fight such incentives prior to now. Instagram in 2019 examined out hiding likes to curb comparisons and harm emotions that include attaching recognition to content material. X, previously Twitter, is preparing to make likes private, as effectively, however for much less healthful causes. In a really Elon Musk-inspired line of considering, X’s aim is to create extra engagement by permitting folks to privately like “edgy” content material that they in any other case wouldn’t to guard their public picture.
Maven is much less involved in connecting customers with audiences, and extra targeted on connecting them with what’s interesting.
The issue of monetization
Stanley and his co-founders – Blas Moros and Jimmy Secretan – soft-launched Maven in late January. The platform publicly debuted in Could alongside a Wired function that Stanley says gave Maven a prime trending spot on Product Hunt and introduced on hundreds of signal ups.
These are nonetheless small numbers in comparison with different new entrants into the social media area. Bluesky, which launched in 2021, has had 5.6 million signal ups. As of January 2024, Mastodon had 1.8 million lively customers. Farcaster, a brand new crypto-based social protocol that just raised $150 million, has counted about 350,000 signups. All of those new networks might want to develop considerably in the event that they’re to be thought-about profitable.
It’s nonetheless an open query over whether or not Maven will even have the ability to develop its consumer base with out the very poisonous qualities we like to hate, however which nonetheless drag us again to the cesspit that’s social media.
Maven raised $2 million in 2023 in a spherical led by Twitter co-founder Ev Williams, Stanley advised TechCrunch. OpenAI CEO Sam Altman additionally participated within the spherical. Stanley stated Williams and Altman invested as a result of, like many people who’ve grow to be endeared by Maven’s nearly too-sweet-for-this-world ethos, they suppose the world and the web wants one thing like this.
And certainly, Maven’s idealistic hope to attach folks to interesting concepts is a breath of contemporary air that smells just like the early 2000s, when the web was a spot of connection and exploration. Sentiments from early customers on the platform are principally constructive and optimistic, as many got here to the platform for real and serendipitous interactions and the promised freedom from toxicity.
However will idealism be sufficient to convey on extra institutional buyers later when Maven desires to develop?
“I believe the problem we face is that going ahead, that turns into a more durable and more durable option to increase cash,” stated Stanley, noting that buyers received’t be throwing down tens of millions except there’s a transparent path to get a return on their funding.
“I simply want to seek out the fitting buyers going ahead and shortly get to a sustainable enterprise mannequin,” he continued, musing over the concept of a subscription mannequin that may permit Maven to maintain its ideology intact.
There are, in fact, different methods for Maven to usher in income. Promoting is one path, however one which appeals much less to Stanley due to how tied up it’s with virality and sensationalism.
Down the road, Maven may additionally doubtlessly promote its knowledge to firms like OpenAI which are coaching their algorithms on reams of knowledge. OpenAI earlier this month signed a deal with Reddit to coach its AI on the social media firm’s knowledge. And Maven’s worth proposition from an AI standpoint isn’t even simply the content material on the platform – it’s the open-ended algorithm working it.
Stanley advised TechCrunch he believes open-endedness is important to synthetic basic intelligence (AGI), a sort of AI that goals to match or surpass human capabilities throughout a variety of cognitive duties. Open-endedness is “such a salient side of being clever,” Stanley stated. “It’s like this inventive and likewise curiosity-driven side of being human.”
“The info is interesting from an AI perspective, as a result of it’s knowledge about what’s interesting,” stated Stanley, noting that present AI fashions are lacking the intuitive understanding of what’s interesting and what’s not, and the way that can change over time. Nevertheless, regardless that the information has potential worth to AI, Stanley stated Maven has no take care of any firm to grant entry to that knowledge.
And whereas he stated he hasn’t dominated that risk out sooner or later, he would suppose very rigorously about what the implications of sharing such knowledge could be.
“That’s not the purpose of this for me,” he stated, noting that he’s not satisfied that it could be a great factor for neural networks to be fully open-ended as a result of which may make any inventive endeavors by people fully pointless.
“I actually needed to create this worldwide serendipitous group,” he stated. “It’s not like I’ve a facet plan that we’re going to make use of Maven to create open-ended AI or one thing. I simply needed to create one thing for folks as a result of I began to really feel like everyone’s gonna be speaking to chatbots an increasing number of and we’re gonna be much less and fewer related with different folks. And I used to be contributing to that being an AI researcher.”
“One thing about this concept of a serendipity community made me really feel morally higher, like I may truly contribute to folks being extra related somewhat than much less.”