The co-founders of Instagram, who left Facebook in 2018 due to disagreements with its parent firm, have started a new startup to research concepts for cutting-edge social programmes. Artifact is their initial offering; it’s a personalised news feed that utilises machine learning to identify your interests and will soon enable you to debate stories with friends.
Today, Artifact, a platform that combines articles, facts, and artificial intelligence, opens up its waiting list to the general public. According to Systrom, the business intends to quickly admit consumers. The software is accessible for iOS and Android, and you may sign up on your own by doing so here.
The simplest way to think of Artifact is as a text-based TikTok, but you could also describe it as Google Reader resurrected as a mobile application or even as a surprise assault on Twitter. The app opens to a feed of well-liked stories selected from a vetted list of publishers, which ranges from major news outlets like the New York Times to specialised small-scale blogs. As you view videos on TikTok’s For You page, Artifact will eventually give you posts and stories that are similar to those you’ve previously chosen to read.
As of right now, only that centrally sorted feed will be shown to users that join from the waitlist. However, Systrom anticipates two other features, which are presently being tested by Artifact beta users, to become the app’s cornerstones. One of them is a feed that displays posts made by users you’ve selected to follow along with their comments. (At least initially, you won’t be able to publish plain text without a link.) The second is a direct-message inbox where you can have private conversations with friends on the things you read.
Artifact has a retro vibe to it at times. Big social platforms have spent the last several years pursuing short-form video products and the associated ad revenue in response to TikTok’s success.
Meanwhile, Artifact has text firmly in its sights, much like a social network from the late 2000s. However, the inventors are optimistic that more than a decade’s worth of lessons gained and new developments in artificial intelligence can help their software reach a wider audience.
He told me that Systrom and Krieger started talking about the concept for Artifact a few years ago. Systrom claimed that before working at Instagram, he had doubts about the capabilities of machine-learning systems to enhance recommendations.
Every time we used machine learning to enhance the customer experience, he observed, “things became really good really quickly over the years.”
So why return right now? Technically, this isn’t the pair’s first endeavour since Instagram; in 2020, they collaborated to build the website RT.live to monitor the COVID-19 virus’s progress.
However, Systrom informed me that they were waiting for three things before launching a new business. One, there was a significant new wave in consumer technology that he and Krieger might try to ride. Two, a means of tying that wave to social technology, in which he and Krieger still have strong emotional attachments. Third, a potential issue that their solution could answer. Systrom has long thought about technology design in terms of the tasks that it can perform for its users.
The transformer, which Google created in 2017, was the innovation that made Artifact possible. It provides a method for systems to comprehend language with much less inputs than were previously needed.
The release of ChatGPT last year and the ensuing surge of interest in AI were directly influenced by the transformer, which enabled machine learning systems to advance at a far faster rate. (The “T” in ChatGPT stands for transformers.)
Additionally, it gave social networks some fresh options. Social networks initially exposed you to content that your friends found amusing—the Facebook model. Then, using the Twitter model, they started displaying content based on the users you choose to follow, whether or not you were friends.
TikTok pioneered the idea of showing you content based only on algorithmic predictions, regardless of your friends or the people you followed. It quickly surpassed all other apps in terms of downloads.
A text-based attempt to accomplish the same goal is represented by Artifact.
Systrom said, “I observed that shift and I was like, oh, that’s the future of social.” “These disconnected graphs; these learned rather than intentionally constructed graphs. And what made it amusing to me was that I wondered, “Man, why isn’t this happening elsewhere on social media?” as I looked around. Why does Twitter still rely heavily on follows? What’s Facebook for?
It is unclear if tailored recommendations for news stories and blog entries can help Artifact achieve the same level of viral popularity that videos did for TikTok. It’s not a given: a flurry of tailored news applications with names like Zite and Pulse appeared in 2014 but quickly faded away due to their inability to forge lasting habits in users. Additionally, earlier this month, Tokyo-based SmartNews, which employs comparable AI technology to tailor suggestions, cut 40% of its staff in the US and China due to a dwindling user base and a competitive advertising market.
Artifact hasn’t committed to a business model yet, like the majority of businesses at this point. Advertising would be a natural fit, according to Systrom. He is also considering revenue-sharing agreements with publications. If Artifact becomes popular, it might aid readers in discovering new publications and persuade them to subscribe; in this case, it might be reasonable for Artifact to try and take a piece.
Also, Systrom informed me The responsibility of providing readers with excellent news and information is one that Artifact will take seriously. He explained to me that this entails making an effort to only include publications who uphold editorial standards of excellence. Although you can use the app to search for specific outlets, the company is currently not disclosing the names of all the publishers in its database.
Right- and left-leaning publishers were both represented; Fox News is one example. However, Systrom isn’t afraid to say that the business will use its own discretion in determining who belongs and who doesn’t.
The inability of many of these firms to make objective judgements in the name of excellence and human growth, he claims, has been one of the problems with technology recently. “Right? Just take the difficult step.
He adds that specific posts that propagate falsehoods will also be deleted by Artifact. And in order to reward more deeply engaging content, its machine-learning systems will be largely tuned to assess how long you spend reading about various topics as opposed to, say, what gets the most views and comments.
