Lessons from the Facebook Blackout 1:12
(CNN) -
For billions of people around the world, Facebook can be a source of photos of pretty babies, misinformation about vaccines and other things, all of which appear in our
feeds
with the help of algorithms.
Now, hours of testimony and thousands of pages of documents from Facebook whistleblower Frances Haugen reignited scrutiny of the impact Facebook and its algorithms have on teens, democracy and society at large.
The fallout has raised the question of how far Facebook, and perhaps similar platforms, can or should rethink using a series of algorithms to determine what images, videos, and news users watch.
The focus is on the Facebook algorithm that recommends content
How does Facebook's artificial intelligence work?
1:09
Haugen, a former Facebook product manager with a background in "algorithmic product management," focuses his criticism primarily on the company's algorithm designed to show users the content they are most likely to engage with.
According to her, this algorithm is responsible for many of Facebook's problems, such as fostering polarization, misinformation and other toxic content.
In an appearance on the "60 Minutes" show, he said that Facebook understands that by making the algorithm more secure, "people will spend less time on the site, click fewer ads, and make less money."
advertising
Facebook CEO Mark Zuckerberg has rebutted the idea that the company prioritizes profit over the safety and well-being of users.
Facebook's director of global policy management, Monika Bickert, said in an interview with CNN, following Haugen's appearance in the Senate on Tuesday, that it is "not true" that the company's algorithms are designed to promote incendiary content. , and that Facebook does "the opposite" by downgrading the so
-
called
click-bait
.
"I think we don't want computers to decide what we focus on."
At times in his testimony, Haugen appeared to suggest a radical rethinking of how the news feed should work to address the issues it presented through extensive company documentation. "I am a strong advocate of chronological classification, of ordering by time," she said in testimony before a Senate subcommittee last week. "Because I think we don't want computers to decide what we focus on."
But the algorithms that choose what we see are critical not just for Facebook, but for numerous social media platforms that followed in Facebook's footsteps.
TikTok, for example, would be unrecognizable without content recommendation algorithms.
And the bigger the platform, the greater the need for algorithms to filter and classify content.
Algorithms are not going away.
But there are ways Facebook can improve them, algorithm and artificial intelligence experts told CNN Business.
However, this will require something Facebook has so far seemed reluctant to offer (despite arguments from executives): more transparency and control for users.
Facebook is one of the largest and most popular social networks in the world.
What's in an algorithm?
Today's Facebook, with a constant flow of information and advertisements (both selected by algorithms), is a very different social network from the one it started with.
In 2004, when Facebook first launched as a site for college students, it was easier and more tedious to navigate: if you wanted to see what your friends were posting, you had to visit their profiles one at a time.
This began to change significantly in 2006, when Facebook introduced the news feed, which offered users a source of updates from family, friends, and that person they had a couple of bad dates with.
From the beginning, Facebook used algorithms to filter the content that users saw in the news.
In a Time magazine article from 2015, the company's chief product officer, Chris Cox, said curation was necessary even then because there was too much information to show all of it to every user.
Over time, Facebook's algorithms evolved and users got used to algorithms determining how Facebook content would be presented.
What is an algorithm?
An algorithm is a set of mathematical steps or instructions, especially for a computer, that tell it what to do with certain inputs to produce certain results.
It is like a cooking recipe in which the ingredients are the starters and the final dish is the result.
On Facebook and other social networks, however, you and your actions - what you write or the images you post - are the inputs.
What the social network shows you, be it a post from your best friend or an advertisement for camping equipment, is the result.
At best, these algorithms can help personalize
feeds
so that users discover new people and content that match their interests based on their previous activity.
Then, at worst, as Haugen and others point out, they run the risk of directing people into troublesome holes that can expose them to toxic content and misinformation.
And in both cases, they make people stay longer browsing, which can help Facebook make more money by showing users more ads.
Multiple algorithms
Many algorithms work together to create the experience seen on Facebook, Instagram, and other places online.
This can make it even more difficult to figure out what's going on within these systems, especially in a large company like Facebook, where multiple teams build multiple algorithms.
"If some higher power were to go to Facebook and say, 'Fix the algorithm in XY,' it would be really difficult because they have become really complex systems with a lot of inputs, a lot of weights, and it's like multiple systems working together," Hilary said. Ross, director of programs at Harvard University's Berkman Klein Center for Internet & Society and manager of its Institute for Rebooting Social Media.
How a LinkedIn message laid the groundwork for the Facebook complaint
More transparency
However, there are ways to make these processes clearer and give users more of a say in their operation.
Margaret Mitchell, who leads AI ethics for AI model builder Hugging Face and who previously co-led Google's AI ethics team, thinks this could be done by allowing you to see the details of why you see it. seen on a social network, such as in response to posts, advertisements, and other things that are viewed and interacted with.
"You can even imagine that you have something to say about it. You could select preferences for the kinds of things that you want to be optimized for you," he said, such as how often you want to view content from your immediate family, your high school friends. or photos of your baby.
All of these things can change over time.
Why not let the users control them?
Transparency is key, he said, because it encourages good behavior on social media.
Some proposals
According to Sasha Costanza-Chock, director of research and design at the Algorithmic Justice League, another way to push social media towards greater transparency is to increase independent audits of its algorithmic practices.
The idea is that this includes fully independent researchers, investigative journalists, or people within regulatory bodies - not the social media companies themselves or the companies they hire - who have the knowledge, skills, and legal authority to demand access. algorithmic systems in order to ensure that laws are not violated and best practices are followed.
James Mickens, a professor of computer science at Harvard and co-director of the Berkman Klein Center's Institute for Rebooting Social Media, suggests exploring the ways in which elections can be audited without revealing private information about voters (for example, who each person voted for ) to find out how algorithms can be audited and reformed.
He thinks that could give some ideas for building an auditing system that would allow people outside of Facebook to monitor while sensitive data is protected.
Other metrics for success
Instagram and advertisements, how much do you earn for them?
3:08
A big hurdle, experts say, to making significant improvements is social media's current focus on the importance of engagement, or the amount of time users spend scrolling, clicking, and otherwise interacting with posts and ads. of social networks.
Haugen revealed internal Facebook documents showing that the social network is aware that its "core product mechanics, such as virality, recommendations and engagement optimization, are an important part" of the reasons why Facebook's pitch hatred and misinformation "flourish" on your platform.
Changing this is difficult, experts said, although several agreed that it may involve taking into account the feelings of users when they use social networks and not just the amount of time they spend using them.
"The hitch is not synonymous with good mental health," Mickens said.
However, can algorithms really help fix Facebook's problems?
Mickens, at least, is hopeful that the answer is yes.
He believes that they can be optimized more towards the public interest.
"The question is: what will convince these companies to start thinking this way?
In the past, some would have said pressure from advertisers, whose dollars these platforms support, would be needed.
But in his testimony, Haugen seemed to bet on a different response: pressure from Congress.