Emergent AI can mend my mistakes on Meta - Zuck

In the final part of my series, I reveal Zuck's vision for an inclusive community, where AI is used to decide what billions should see and know...

Welcome to the final instalment of my deep dive into the brain of Meta CEO Mark Zuckerberg, the guy who wants your kids to follow his bots into the Metaverse.

When this started, I considered him an autocrat and free speech absolutist, but now I have come away with a better understanding.

He emerges as a tech pragmatist, a guy who believes his destiny is to create the tech future, but who knows that despite his genius, the job is just beyond him.

Moving fast and breaking things has always been the engineers’ get out of jail free card. If I get it wrong, don’t sweat it, I’ll fix it next time.

Zuck thought that as well earlier in his career, but he then learned that doesn’t work so well when it undermines democracy, facilitates terrorism or leads to child suicide.

It’s a heavy mantle to bear, but his musings show that he thinks deeply about it.

His manifesto, written in 2017, which I have drawn deeply on, depicts a guy with uncommon power trying to find a right path.

It might be his path, and not yours or mine, but he is at least open and willing to be judged.

We could have had worse people with their hand on the tiller of social over the past 20 years. Just look what Elon Musk has done at X in a far shorter timeframe.

♾️

Before we dive in, join me in welcoming new subs over the past two days from BBC News reaching half a billion readers daily, more joiners from the Dow Jones, from Newsweek in New York, and in London from the Professional Publishers Association, global strategists Human Change in Washington DC, the $160 billion US banking giant Wells Fargo, more from US publisher Hearst, which writes $11 billion in annual revenue, and Australian broadcaster Seven, among others. 👋

And thank you to all the former Meta execs who have worked with Zuckerberg, and who were willing to share their insights with me. 🙏

Let’s get to it.

♾️

My conclusion is that publishers find it hard to pin facts on Zuck because his mind, and his business, move at the speed of thought.

Meta pivots from friend, to frenemy, to enemy, in a synapse, and media companies move slower and are hopelessly outgunned at keeping up.

News is less than one per cent of Meta’s overall traffic, but many global publishers have become almost 100 per cent reliant on Meta for their audience.

When one party’s afterthought is the other’s entire existence, you can see how tensions rise.

Now, in 2024, Zuck’s set his sights on building artificial general intelligence (AGI), but his true North Star remains the Metaverse.

This is his tech-centric vision of a world where everyone lives through, and in his platforms, wrapped in a cocoon powered by his technology.

Social was step one, and AGI is another step towards that. He believes the Metaverse is his legacy.

To achieve it, he needs to bend the news media to his will. He needs to build AI technology that doesn’t exist today, and he must alter global social expression.

If that sounds impossible then remember, he’s already done it once, and listening to his most recent public appearances, he thinks he can do it again.

♾️

Today I’m publishing the final part of Zuck’s manifesto where he reveals his vision for an inclusive community.

He shares how he believes that people need to come together to build a better world, created around community, sharing, and what he calls “collective decision-making”.

And how he intends to use this social melting pot to teach tech to make better decisions around content curation - particularly on polarising issues.

He talks about using data to create the concept of a digital editor that can make the right decisions about what the world sees, on Black Lives Matter and more.

It’s big picture stuff, and something no traditional editor has ever needed to consider, so it’s worth reading.

♾️

Before I hand Zuck the mic, I was asked in New York last week whether he can be trusted.

This series has taught me that there is great danger in publishers taking his word as fact. What he says in the moment, and what Meta does over time, are very different.

But it’s not because he’s lying as his critics would love to believe. That’s wrong. It’s that he’s adapting.

That means publishers who naively take his statements as gospel can only lose. Just ask the Australian publishers who were stunned when he axed their news payouts.

The answer is to ignore what Meta says and focus on what it does. Oh, and to never blink, because it’s moving at the speed of thought, and that won’t change.

One former senior exec who worked with Zuck mailed me last week: “Your series about Zuck is great, but such a mind f*** for me every time I read it. Reliving it is surreal.”

This is Zuck, in a memo written to friends and colleagues in 2017, in his own words.

By Mark Zuckerberg

Building an inclusive global community requires establishing a new process for citizens worldwide to participate in community governance.

I hope that we can explore examples of how collective decision-making might work at scale.

Facebook is not just technology or media, but a community of people.

That means we need Community Standards that reflect our collective values for what should and should not be allowed.

In the last year, the complexity of the issues we’ve seen has outstripped our existing processes for governing the community.

We saw this in errors taking down newsworthy videos related to Black Lives Matter and police violence, and in removing the historical Terror of War photo from Vietnam.

We’ve seen this in misclassifying hate speech in political debates in both directions - taking down accounts and content that should be left up and leaving up content that was hateful and should be taken down.

Both the number of issues and their cultural importance has increased recently.

This has been painful for me because I often agree with those criticising us that we’re making mistakes.

These mistakes are almost never because we hold ideological positions at odds with the community, but instead are operational scaling issues.

Our guiding philosophy for the Community Standards is to try to reflect the cultural norms of our community. When in doubt, we always favour giving people the power to share more.

There are a few reasons for the increase in issues we’ve seen: Cultural norms are shifting, cultures are different around the world, and people are sensitive to different things.

First, our community is evolving from its origin connecting us with family and friends to now becoming a source of news and public discourse as well.

With this cultural shift, our Community Standards must adapt to permit more newsworthy and historical content, even if some is objectionable.

For example, an extremely violent video of someone dying would have been marked as disturbing and taken down.

However, now that we use Live to capture the news and we post videos to protest violence, our standards must adapt.

Similarly, a photo depicting any child nudity would have always been taken down - and for good reason - but we’ve now adapted our standards to allow historically important content like the Terror of War photo.

These issues reflect a need to update our standards to meet evolving expectations from our community.

Second, our community spans many countries and cultures, and the norms are different in each region.

It’s not surprising that Europeans more frequently find fault with taking down images depicting nudity, since some European cultures are more accepting of nudity than, for example, many communities in the Middle East or Asia.

With a community of almost two billion people, it is less feasible to have a single set of standards to govern the entire community, so we need to evolve towards a system of more local governance.

Third, even within a given culture, we have different opinions on what we want to see and what is objectionable.

I may be okay with more politically charged speech but not want to see anything sexually suggestive, while you may be okay with nudity but not want to see offensive speech.

Similarly, you may want to share a violent video in a protest without worrying that you’re going to bother friends who don’t want to see it.

And just as it’s a bad experience to see objectionable content, it’s also a terrible experience to be told we can’t share something we feel is important.

This suggests we need to evolve towards a system of personal control over our experience.

Fourth, we’re operating at such a large scale that even a small percent of errors causes a large number of bad experiences.

We review over 100 million pieces of content every month, and even if our reviewers get 99 per cent of the calls right, that’s still millions of errors over time.

Any system will always have some mistakes, but I believe we can do better than we are today.

I’ve spent a lot of time over the past year reflecting on how we can improve our community governance.

Sitting here in California, we’re not best positioned to identify the cultural norms around the world.

Instead, we need a system where we can all contribute to setting the standards. Although this system is not fully developed, I want to share an idea of how this might work.

The guiding principles are that the Community Standards should reflect the cultural norms of our community, that each person should see as little objectionable content as possible, and each person should be able to share what they want while being told they cannot share something as little as possible.

The approach is to combine creating a large-scale democratic process to determine standards with AI to help enforce them.

The idea is to give everyone in the community options for how they would like to set the content policy for themselves. Where is your line on nudity? On violence? On graphic content? On profanity?

What you decide will be your personal settings.

We will periodically ask you these questions to increase participation and so you don’t need to dig around to find them.

For those who don’t make a decision, the default will be whatever the majority of people in your region selected, like a referendum.

Of course you will always be free to update your personal settings anytime.

With a broader range of controls, content will only be taken down if it is more objectionable than the most permissive options allow.

Within that range, content should simply not be shown to anyone whose personal controls suggest they would not want to see it, or at least they should see a warning first.

Although we will still block content based on standards and local laws, our hope is that this system of personal controls and democratic referenda should minimise restrictions on what we can share.

It’s worth noting that major advances in AI are required to understand text, photos and videos to judge whether they contain hate speech, graphic violence, sexually explicit content, and more.

At our current pace of research, we hope to begin handling some of these cases in 2017, but others will not be possible for many years.

Overall, it is important that the governance of our community scales with the complexity and demands of its people.

We are committed to always doing better, even if that involves building a worldwide voting system to give you more voice and control.

Our hope is that this model provides examples of how collective decision-making may work in other aspects of the global community.

This is an important time in the development of our global community, and it’s a time when many of us around the world are reflecting on how we can have the most positive impact.

History has had many moments like today. As we’ve made our great leaps from tribes to cities to nations, we have always had to build social infrastructure like communities, media and governments for us to thrive and reach the next level.

At each step we learned how to come together to solve our challenges and accomplish greater things than we could alone. We have done it before and we will do it again.

I am reminded of President Lincoln’s remarks during the American Civil War: “We can succeed only by concert. It is not ‘can any of us imagine better?’ but, ‘can we all do better?’ The dogmas of the quiet past, are inadequate to the stormy present. The occasion is piled high with difficulty, and we must rise with the occasion. As our case is new, so we must think anew, act anew.”

There are many of us who stand for bringing people together and connecting the world. I hope we have the focus to take the long view and build the new social infrastructure to create the world we want for generations to come.

It’s an honour to be on this journey with you. Thank you for being part of this community, and thanks for everything you do to make the world more open and connected.

Mark

Thanks for reading. I’ll end this project by echoing what I said up front.

The news industry has survived the worst years in its existence, and major change is afoot with the break-up of Google and its ad dominance.

That’s all happening so fast it’s practically on autopilot.

But we are talking too little about Meta, as it ricochets from connecting communities, to silencing news, to building a synthetic AI future.

What Meta does impacts every publisher because it influences half the world’s population.

That’s why researching his musings and evolving strategy matters. I hope my work has been useful, and I shall be following developments as they come.