Internet users in the European Union are logging on to a quiet revolution on mainstream social networks today: The ability to say ‘no thanks’ to being attention hacked by AI.
Thanks to the bloc’s Digital Services Act (DSA), users of Meta’s Facebook and Instagram, ByteDance’s TikTok and Snap’s Snapchat can easily decline “personalized” content feeds based on “relevance” (i.e. tracking) — and switch to a more humble kind of news feed that’s populated with posts from your friends displayed in chronological order. And this is just the tip of the regulatory iceberg. The changes apply to major platforms in the EU but some are being rolled out globally as tech giants opt to streamline elements of their compliance.
Facebook actually got out ahead of today’s DSA compliance deadline by launching a chronological new Feeds tab last month — doing so globally, seemingly, not just in the EU. But it’s a safe bet Meta wouldn’t have made the move without the bloc passing a law that mandates mainstream platforms give users a choice to see non-personalized content.
Notably the new chronological Facebook news feed does not show any “Suggested For You” posts at all. And that total separation of tracking-based content recommendations from non-personalized content selections is absolutely down to the DSA. If Meta could injection a little AI-powered attention hacking into the humble chronological news feed it surely would. But the bloc’s law requires no crossing of these streams. Respect for user agency demands a space safe from surveilling AIs.
We’ve also recently seen YouTube announce that logged in users with the ‘watch history’ feature turned off won’t be bothered by next video recommendations based on profiling what they’ve watched before. Also, seemingly, a change it’s decided to roll out everywhere, not just in the EU — but again a development that’s clearly been driven by the DSA.
You might ask why does the ability to switch off profiling-based content recommendations matter? Isn’t it a relatively minor detail in the grand scheme of platform power? Well yes and no. The power of platforms to keep users engaged inside their walled gardens derives from a number of factors — one of which is the massive information asymmetry they can wield against our eyeballs by tracking what we click at, engage with, linger on, search for and so on.
Content choices based on this tracking don’t even have to be very sophisticated — and, indeed, the programming can feel terribly crude. Such as how, for the past many, many months, after I happened to watch a cat video on Instagram, my Home feed has been peppered with unavoidable injections of fur. And these suggested cat videos never seem to end. It’s truly been the longest tail…
Instagram feed with cat posts
Instagram feed screengrab: Natasha Lomas/TechCrunch
How this typically went down was after scrolling through the (smaller) stack of Instagram posts from people I do actually follow (still peppered with suggested cat videos) the AI would take over — populating the rest of the feed (apparently bottomless) with what seemed like an infinite selection of cat videos. Cats being cute, cats being acrobatic, cats being funny, cats being memed, cats being rescued from dire conditions… It got to the point where I would dread logging on to Instagram because of what I would be compelled to look at.
Now don’t get me wrong, I love cats. So, naturally, I’m a fan of cute cat videos. But I sure don’t love a firehose of fur being force-injected into my eyeballs just so Mark Zuckerberg can hold me on his platform a bit longer and keep getting richer than Croesus. It’s pure manipulation and boy does that feel ick. So I have actually been counting down the days for DSA compliance to kick in — and usher in a legal end to this unavoidable algorithmic cat parade.
Today on Instagram I can report finding fur-free peace at last!
Of course the AI-selected cat videos haven’t gone very far. The home feed page now offers two choices: “Following” and “For you” — the second of which remains populated with plenty of furry felines. But at least I can now opt to see only posts from accounts I follow and actively avoid the stuff that’s been selected to try to hack my attention.
Instagram’s ‘Explore’ tab appears to default to algorithmic content selections (“For you”) but click on the down arrow next to the label and you’ll also now see a novel option: “Not personalized”. Click on that and the feed of content Meta’s AIs calculated would best grab the user’s eyeballs (in my case that’s cats and climbing videos) is replaced by a grid of images that look culled from a National Geographic-inspired stock photo selection. Frankly it looks a bit boring but I never looked at the Explore tab anyway. And boring is peaceful.
Over on Facebook, switch on the new (though actually retro) chronological news feed and it makes the platform feel — momentarily — like an entirely different product as friends whose posts would typically be buried by the algorithm as too quotidian (i.e. not engaging enough) sudden get their 15 minutes of fame and pop up right there in your eyeline.
The Facebook home page still defaults to an AI-sorted view, including personalized recommendations for Reels and Stories. But if you switch to the chronological news feed it’s a throwback to Facebook circa 2008, before the platform flipped from ranking posts in reverse chronological order to applying a popularity filter (based on engagement). And we all know what happened to the tone of social media discourse after adtech giants’ algorithms started selecting for outrage… So don’t underestimate the power of a humble news feed comprised of friends’ unsorted shower thoughts. This might be just the sort of content revolution our hyper-polarized societies need.
An ‘AI off’ switch could make even bigger splash on TikTok — where the stickiness of its content selection algorithm has been credited with driving major viral trends and powering the platform’s overall popularity. But stepping away from its AI firehose will still require users to exercise their agency — since the regulation only demands that platforms offer a choice which is not based on profiling. So it remains to be seen whether TikTok’s community will engage with the new non-personalized feeds.
They might just be horrified at how banal lots of the stuff posted to the platform can be once they step outside the AI-filtered attention bubble. While a generation of digital native social media influencers will surely flee screaming from the prospect of reduced engagement. But other users who are tired of influencer babble polluting their feeds might just be weeping with relief at the prospect of an easy toggle to remove distracting noise.
The impact of increased empowerment of users on mainstream platforms may not lead to immediate big bang change. But we should celebrate our new ability to quiet quit their algorithms. It’s long overdue.
AI startup Hugging Face has raised $235 million in a Series D funding round, as first reported by The Information, then seemingly verified by Salesforce CEO Marc Benioff on X (formerly known as Twitter). The tranche, which had participation from Google, Amazon, Nvidia, Intel, AMD, Qualcomm, IBM, Salesforce and Sound Ventures, values Hugging Face at $4.5 billion. That’s double the startup’s valuation from May 2022 and reportedly more than 100 times Hugging Face’s annualized revenue, reflecting the enormous appetite for AI and platforms to support its development.
Hugging Face offers a number of data science hosting and development tools, including a GitHub-like hub for AI code repositories, models and datasets, as well as web apps to demo AI-powered applications. It also provides libraries for tasks like dataset processing and evaluating models in addition to an enterprise version of the hub that supports software-as-a-service and on-premises deployments.
The company’s paid functionality includes AutoTrain, which helps to automate the task of training AI models; Inference API, which allows developers to host models without managing the underlying infrastructure; and Infinity, which is designed to increase the speed with which an in-production model processes data.
“AI is the new way of building all software. It’s the most important paradigm shift of the decade and, compared to the software shift, it’s going to be bigger because of new capabilities and faster because software paved the way,” co-founder and CEO Clément Delangue told TechCrunch via email. “Hugging Face intends to be the open platform that empowers this paradigm shift.”
Delangue, a French entrepreneur, launched Brooklyn-based Hugging Face in 2016 alongside Julien Chaumond and Thomas Wolf. The trio originally built a chatbot app targeted at teenagers. But after open sourcing the algorithm behind the app, Hugging Face pivoted to focus on creating a platform for creating, testing and deploying machine learning.
The company has 10,000 customers today, it claims, and more than 50,000 organizations on the platform. And its model hub hosts over 1 million repositories.
Contributing to the growth is the strong, sustained interest in AI from the enterprise. According to a HubSpot poll, 43% of business leaders say they plan to increase their investment in AI and automation tools over the course of 2023, while 31% say AI and automation tools are very important to their overall business strategy.
Much of what Hugging Face delivers falls into MLOps, a category of tools for streamlining the process of taking AI models to production and then maintaining and monitoring them. The MLOps market is substantial in its own right, with one report estimating that it’ll reach $16.61 billion by 2030.
But Hugging Face dabbles in other areas, too.
In 2021, Hugging Face launched BigScience, a volunteer-led project to produce an open source language model as powerful as OpenAI’s GPT-3, but free and open for anyone to use. It culminated in Bloom, a multilingual model that for more than a year has been available to tinker with on Hugging Face’s model hub.
Bloom is but one of several open source models to which Hugging Face has contributed development resources.
Hugging Face collaborated with ServiceNow, the enterprise software company, to release a free code-generating AI model called StarCoder (a follow-up model, SafeCoder, debuted this week). And the startup made available its own free version of ChatGPT, OpenAI’s viral AI-powered chatbot, in partnership with the German nonprofit LAION.
Hugging Face’s team-ups extend to major cloud providers, some of which are strategic investors.
Hugging Face recently worked with Nvidia to expand access to cloud compute via Nvidia’s DGX computing platform. It has a partnership with Amazon to extend its products to AWS customers and leverage Amazon’s custom Trainium chips to train the next generation of Bloom. And Hugging Face collaborated with Microsoft on Hugging Face Endpoints on Azure, a way to turn Hugging Face-developed AI models into scalable production solutions hosted through Azure.
With this latest investment, Delangue says that Hugging Face plans to “double down” on its supportive efforts in many domains, including research, enterprise and startups. It has 170 employees, but plans on recruiting new talent over the coming months.
Hugging Face has raised a total of $395.2 million to date, placing it among the better-funded AI startups in the space. Those ahead of it are OpenAI ($11.3 billion), Anthropic ($1.6 billion), Inflection AI ($1.5 billion), Cohere ($435 million) and Adept ($415 million).
That kind of public interest visibility atop tech giants is also long overdue. And the information asymmetry that adtech giants, especially, have exploited to fatten their bottom lines at our eyeballs’ expense has always been drastically unfair.
It’s past time they gave back. And it’s past time we had simple options to stop their content targeting systems from stealing our free time.
