- 4 months ago
Seeing is no longer believing. With a simple prompt, anyone can now create images, sounds or videos that appear perfectly convincing. Newsrooms, HR departments, cybersecurity firms and all of us going about our daily lives are all hitting the same wall: we need to find new ways to verify real-looking media that can be generated in seconds. How can we move towards a "provenance-first" culture, where the origin of media matters as much as the content itself? What guardrails, from digital watermarking to human verification policies, can companies and regulators adopt? And as the technology evolves, what happens to our legal and social systems when evidence can no longer be taken at face value?
Category
🤖
TechTranscript
00:23Thank you, great to have you all here.
00:26Wow, the purple stage is really purple, isn't it?
00:28And I love all this metallic, industrial chic.
00:33Welcome, everyone.
00:35So I was doing a little bit of research for this panel today,
00:37and I found that a Europol estimate suggested that 90% of online content is synthetic,
00:44and researchers suggest there are going to be 8 million deepfakes shared across social this year.
00:50So that's just the deepfakes, not the number of shares.
00:52Obviously, the technology is moving faster than regulation can handle, faster than detection tools,
00:59and faster, I think, than many consumers maybe understand.
01:02It's undermining our democracies.
01:04It's undermining trust in institutions.
01:07AI is obviously amplifying the problem, meaning that separating truth from misinformation is only going to get more challenging.
01:15Now, we have a fantastic group here with us today to address this
01:19and maybe offer, hopefully, a little bit of optimism on this topic and ways we can address it and win.
01:24I'd like you to introduce yourselves, please.
01:26Rona, if you don't mind starting.
01:28Sure.
01:28Hi, I'm Rona Murphy.
01:29I'm the chair of the board of the International Consortium of Investigative Journalists,
01:34and my background is as an executive running media companies in New York, mostly.
01:41And I'm still involved in consulting a number of news organizations, and I'm now based in Dublin.
01:48Hi, my name is Mario.
01:49I lost my voice yesterday, so hopefully you'll be able to hear me today.
01:52I'm the CEO of Readocracy, which is trying to fix the Internet.
01:57And I'm also the chair for the Center of Applied Information Health,
02:00which is creating the new science of infotrition, and I'm based in Toronto.
02:06I'm Bill Adair.
02:07I'm a professor of journalism at Duke University in the United States.
02:11I'm the founder of PolitiFact, the fact-checking website,
02:14and the co-founder of the International Fact-Checking Network, which is the Association of Fact-Checkers.
02:20I'm now writing a book about Wikipedia.
02:23Thank you so much, guys.
02:25So let's get into it.
02:26So I think maybe it's worth just addressing the word misinformation itself
02:30and trying to sort of pass that and sort of figure out what it means,
02:34because it has become contested, and we've sort of seen all these kind of, you know,
02:39various conversations about truth and freedom of speech
02:45and all these kinds of issues that we see being raised by all kinds of elements in our society,
02:51largely to justify a form of censorship.
02:53So I guess the question is how we protect truth,
02:56and Rainer, maybe you can take this to begin with.
02:59How do we protect truth without handing kind of a weapon
03:03to sort of like very powerful individuals?
03:08Well, do you want to take it, Mario, then?
03:10Yeah, yeah.
03:12I think the way we protect it is actually to go deeper.
03:15I don't think truth is granular enough.
03:17I don't think it's the right place to focus because it lends itself to conspiracy.
03:21It's too opaque.
03:22It's too debatable.
03:23And I think, you know, where I sit, this is why I'm so interested in this notion of infotrition
03:28and the ingredients of the information we receive going beyond true or false,
03:34because when we focus on true or false, we're focused on that aspect
03:38instead of what it does to our emotions in particular.
03:41For example, when we talk about polarization,
03:44which obviously often goes hand in hand with disinformation and whatnot,
03:48it's not just politics, it's politics mixed with an agitation vector.
03:53If we were all perfectly calm and tried to be reasonable with our perspectives,
03:57even if they were not true, we wouldn't want to strangle each other nearly as much.
04:02And I think that's a, you know, and you can go much deeper than that.
04:04This is why there's all these ingredients.
04:06And I think that more from a first principles perspective would help us get to the real problems
04:11and how we're being manipulated and made to hate each other
04:13more than just this perpetual, endless, difficult debate around truth.
04:17So that's, I know it's not the perspective, it's one of them, but that's where I see it.
04:21So I think we need to, we do need to tell people what the facts are.
04:29And I think that's the role of journalists.
04:31And it's really important in an age where people wonder about so many things.
04:37Is that accurate?
04:41That journalists really need to step in and check the facts.
04:46And that's what a whole legion of journalists are doing all around the world,
04:52telling people what, when their authorities say something, whether it's accurate or not.
04:59And I think that's critical to have that ground truth to, so that we can make decisions.
05:08And not everyone's going to accept what the journalists say, but it's important to do that journalism.
05:14Right.
05:14So, you know, people can have their opinions, but they can't have their own facts, is basically the point of
05:19that.
05:19Roan, I'd love to bring you in now.
05:21Obviously, misinformation is often sort of like presented as a technology problem.
05:26But obviously, you've worked across traditional media in many newsrooms and executive level.
05:31Do you think that the media in some ways bears some level of responsibility for the environment we're now in
05:37in terms of the information sphere?
05:39Well, partially responsible.
05:41I would still would lay the blame firmly at the foothold of the technology platforms.
05:46But if we can go back 30 years, I mean, we were very naive about the power of what the
05:52platforms were going to look like
05:53and what the internet was going to look like.
05:55And we gave away a lot of our content, undervalued ourselves in a panic, really, because we didn't, we thought
06:02this was a new revenue stream,
06:03which it was, but we didn't realize it was going to come at the expense of another revenue stream, the
06:09bigger revenue stream.
06:11And we probably undervalued our pricing.
06:16We were slow to really, we being, I mean, broadly speaking, to embrace the digital revolution.
06:23We did catch up, but it took, by the time we'd caught up, the platforms had got, the algorithms had
06:27gotten more sophisticated and had kind of outrun us.
06:31So, in that sense, there is some blame.
06:34And then, in the current context, it is very difficult not to follow engagement.
06:41And obviously, audience engagement makes money rather than accuracy.
06:45So, there is some guilt on behalf of media owners about publishing things that they know are going to get
06:51engagement at the expense of accuracy.
06:54But I would say that serious newsrooms do take accuracy very seriously.
06:58And just to agree with what Bill just said, I mean, verifying is really the most important thing now to,
07:06you know, to basically build a business and build trust.
07:09Yeah, absolutely.
07:10The trust is crucial.
07:12Mario, I mentioned the number at the beginning, like Europol estimating that 90% of online content might be synthetically
07:20generated now.
07:21Staggering amounts.
07:22What do you think that deepfakes reinforce or break about our approach to misinformation and disinformation?
07:32I think I'll go back to first principles.
07:34And one thing I'd love to touch on, what I think it reinforces, because AI, you said it, AI, I
07:40think, magnifies, amplifies a lot of things that already existed.
07:43When we talk about misinformation in the context of AI, we act as if it's some sort of new problem
07:48when it's actually the old problem we never really solved, just with fuel poured onto the fire.
07:54And one thing I'd like to highlight that is reinforced is that I think the skill of the future that
08:00we all need to cultivate and cultivate in our children and try to cultivate in our readers and viewers is
08:06metacognition.
08:07Metacognition, you know, with traditional misinformation is reflecting on your own mind process and taking that extra step of who
08:16is profiting from my anger or from my emotions.
08:19And now a deepfakes, when you see something that by all accounts seems real, but if it makes you feel
08:24in a particularly strong way or it has a particular context, those alarm bells should be going off.
08:28And that can only happen if you've trained your metacognitive muscle.
08:32And I think that is something that we all need to really focus on.
08:36Well, you mean individuals have some degree of responsibility for this?
08:40Individuals.
08:41And, you know, I think there's something that even on the media side we can do to cultivate that and
08:44create that awareness.
08:45On that first question of when we talk about truth or false, it comes down to transparency.
08:50We have the ability to provide so much data and context now in real time.
08:54For better or for worse, you see on X what Grok has been doing, like on both sides, just like
08:58getting in there and staring things up with its automated context.
09:01I think there's a lot we can do there to actually just trigger a bit that muscle, but also explain
09:07it and certainly make it part of our school system.
09:09Sure.
09:10So, Bill, I'd love to come to you.
09:12In terms of sort of, you know, we just talked about artificial intelligence and the way that it can generate
09:17at scale.
09:18Surely, if we can generate at scale, you know, misinformation, we can also use AI to debunk and ensure that,
09:26you know, there's verification of facts as well.
09:28How do we deploy these tools in the favor of truth?
09:33Absolutely, and I think that's something that we have not looked at much yet, which is how can we take
09:40this incredible power that AI provides and use it to spread accurate information more widely.
09:49And I think the problem there is the business model.
09:53The people who are spreading misinformation have an incentive.
09:58It might be financial.
09:59It might be ideological.
10:02But they want to spread that, those lies widely.
10:08There has not been an equal push to spread facts, and I think that's something we need to explore.
10:17The problem is the business model.
10:20The business model for accurate journalism is not what it used to be.
10:27And so you have no one who is investing the sort of money that needs to be invested to explore
10:35how we could use AI to spread facts.
10:40Ron, I'd love to bring you, I'll come back to you in a second, Mario.
10:43Ron, I'd love to bring you at that point, you know, the business models.
10:45Like, what do you think a sustainable business model looks like now?
10:49Well, a sustainable business model for serious newsrooms is definitely fact-checked, independent news.
10:55So, you know, with the, there really is no other way.
11:00And I'm somewhat of an optimist and feel like the cream will rise to the top.
11:04Like, we're living in an age of misinformation right now.
11:07And absolutely, there are tools being developed by notably Google to help journalists figure out what's accurate and what's not.
11:15But they're not perfect.
11:17So the great thing about AI is for journalists, it's an incredible opportunity because it saves a huge amount of
11:23time with research.
11:24It also generates ideas.
11:25So those are two out of the three things journalists need.
11:28But the third thing they need, and the most important one, is accuracy.
11:30And it is wildly inaccurate.
11:32Not all the time.
11:33I mean, it's often accurate.
11:35But there are wild inaccuracies.
11:37So, and there's, I have myriad examples of journalists that almost published a story that would have gotten them into
11:44really big trouble.
11:45And then just checked with one more source and realized that a very sophisticated dossier they may have been investigating
11:51was, in fact, completely invented.
11:53So, and everybody has those stories.
11:56So, it's basically, you know, you're only going to make money if you can get consumers to pay for journalism
12:02or if you can get philanthropists to fund investigative journalism.
12:06And neither, nobody will do that if they think the news they're reading is not real and not fact-checked.
12:12So, I would say about AI for journalists, it's an incredible opportunity, but it's a tool.
12:18It's not a magic wand.
12:21I'd love to know from your perspective, you don't necessarily have to name outlets, but, like, who is doing it
12:25well?
12:26Who's kind of figured it out?
12:27Who's embracing it in ways that, the technology in ways that is actually benefiting the newsroom?
12:33I think anyone that's running huge data-led stories is producing investigative pieces that we really haven't seen before.
12:42Because what it's able to do, obviously, is condense massive amounts of information and squash it into, find the connections,
12:51and something that would take a journalist maybe a year to figure out, and find all the connections and squash
12:55it into, shrink it down into comprehensive information.
13:00So, it can link legislation with financial records, with individuals on a global scale.
13:09And there's been some fantastic investigative journalist pieces that have appeared, and the New York Times have done it, ICIJ
13:16have done it, some others have done it as well.
13:18And they're getting better and better at it.
13:20But it's really where, at the moment, it's where it's heavily data-led that it's been particularly effective.
13:27Mario, you wanted to come in when Bill was talking about how we use AI tools to actually debunk.
13:32Well, I just want to talk about the business model.
13:34So, if you want to come back to that.
13:36Let's go back to the business.
13:37Let's talk business model.
13:39I was just itching to dive in there.
13:43I think it is one of the biggest misunderstandings and fallacies in today's world that this is the only way
13:51the internet can exist, and this is the only way our information ecosystem can exist.
13:56How can we possibly say that journalism is not well-funded when we never had a metric for being well
14:03-informed?
14:04Power online and in society is based on how much attention you can get.
14:08There is no metric for how informed you are or the quality of the sources you informed yourself with.
14:13It's not rewarded in a formal way.
14:15And I think it's a fallacy to assume that we can't make money in other ways.
14:18The digital advertising and data sale industry is $800 billion, which is not anything to sneeze at, besides the fact
14:25that it's estimated that only about 22% is legitimate.
14:29But when you look at the fact that people spend over a trillion dollars on degrees and certificates every year
14:34trying to prove what they know, and there's an entire industry of everybody desperately trying to signal what they know,
14:40every newsletter they make, every post they make, all this signaling, posturing of trying to show what you know.
14:45And we don't make that count in any way.
14:47There's no metric to show or validate that.
14:50If we could do that, there's over $5 trillion you could unlock.
14:53We just haven't done it.
14:54I don't think we can be surprised that journalism isn't valued when it's nice to have.
14:59It doesn't map to our status and our inclusion and our power the way attention does.
15:03And I think we can fix that.
15:04That is a choice.
15:05That is a choice.
15:06And I just wanted to speak to that because we tend to assume it's a given, and I do not
15:09think it is a given.
15:10Well, then maybe this is a good opportunity for you to talk a little bit about how you're doing at
15:13Readocracy and what that model looks like.
15:16I mean, the signaling, because presumably you're also the platform, if I understand it correctly, that is also about signaling.
15:23It's signaling that you're, okay, so tell us a little bit about it.
15:26So briefly, the notion would be there's basically a missing half of the internet, right?
15:31Every post you see in a feed, it's the bottom, the reactions, the likes.
15:35How did you react, you know, what came out of it?
15:37But where's the metrics on top of what went into it?
15:40Is this person even informed?
15:41Are they balanced?
15:42Do they know what they're talking about?
15:43Can you order the feed that way?
15:45No, you can't.
15:47And generally, the opportunity cost, I mean, first of all, we all obsess about how we feed our bodies.
15:52It's 50% penetration on fitness trackers.
15:54Nobody knows how they're feeding their minds, despite the fact we all complain we're in a mental health crisis.
15:59And lastly, all of you here will spend about as much time in your life consuming, discussing content as you
16:05would studying for about 35 college degrees.
16:07And you have nothing to show for it.
16:09And so our system is, it's like a fitness tracker for information.
16:12You can make all the articles you read, you know, count your podcast, your books, everything, so that you actually
16:17have the data.
16:17You can use it to understand yourself and to present yourself.
16:20And if we integrate it with as a platform, they can reorganize the feed, you can monetize around this, you
16:25can have brands who want to sponsor encouraging people to inform themselves well, instead of just stealing their attention.
16:31It's the difference between power being the quantity of attention you get versus power being the quality of attention you
16:39give.
16:40That's an inversion of the model.
16:41And that's how we're trying to integrate it.
16:43But if we can just think of any other incentive model, it doesn't have to be this.
16:47Just think of another, because right now we just have one, and then we complain.
16:50Let's be a bit original.
16:53I'm very interested to hear what you're doing at Readocracy.
16:56Bill, I'd love to just get into, like, you know, you've got deep expertise in fact-checking.
17:01Obviously, fact-checking is integral to journalism.
17:05What are the main challenges we're seeing now?
17:07Like, Ronan was talking about the fact that there's so much great data journalism being done now.
17:12Now, surely that means also with fact-checking we can use similar tools to ensure that we're, you know, verifying
17:18facts before they're published.
17:21Yes, and it's important to understand two different types of fact-checking.
17:26So there is editorial fact-checking.
17:28So if you're a journalist, you're writing a magazine article, you're going to check the facts in that article.
17:34But the other type is what I help start, and that is checking the facts that politicians say.
17:43And that's equally as important to hold them accountable for what they say.
17:48It is easier now with some of these tools available because you can use data in ways to verify what
17:58somebody says more quickly than ever.
18:01Where it's not easy is to detect AI-generated images.
18:08You know, there was a moment, you probably all remember, when you'd look at an AI-generated photo,
18:15and you would say, oh, that's not right because that person has six fingers.
18:21You know, and there was sort of this six-finger moment because AI, for some reason, didn't know how many
18:27fingers humans had.
18:29That moment, of course, is gone, and AI-generated images are so good.
18:34AI video is so good.
18:37And it is harder than ever for the journalists that do fact-checking to verify what images are real and
18:46which ones are AI.
18:49And what can you do reliably today using AI?
18:52What have you seen in the wild, as it were?
18:56You know, it is really good at sifting through large amounts of data, large numbers of documents.
19:05The work that the New York Times did and other media organizations, too, in going through the Epstein files.
19:14So AI still has a hallucination problem, but when you give it a data set, when you say, here are
19:2110,000 documents,
19:22and go through them and look for these things or look for patterns or whatever,
19:29you're much less likely to have those hallucination problems.
19:32So that's the biggest example I can think of off the cuff of how AI has helped.
19:39Okay.
19:40So I want to throw this next question open to the entire panel, really, and it's kind of like, I
19:44guess, a cultural question.
19:46So, you know, Bill, you talked about, you know, the ability to check, you know, lawmakers or, you know, presidents
19:55to whatever statements they're making.
19:57But, you know, we can see that, like, President Trump tells lies constantly.
20:03Is the challenge more than that?
20:05It's just that the fact that people have stopped caring what's true or not,
20:08and we're basically living in a kind of a vibes world now where people feel like something's true
20:13or they want to be within their own kind of information bubble with their own kind of people, their own
20:18kind.
20:19I think that has always been a problem.
20:22We don't want to believe that our team has committed a foul.
20:27I know this because basketball is very important at Duke and where I work,
20:33and I know that the referees are incredibly biased against Duke.
20:38I mean, it's just a fact.
20:41So that has always been a problem.
20:43I think it's more of a problem now with partisan media,
20:48and I think partisan media goes along and will say, will repeat falsehoods and will repeat false themes,
20:57and so, yes, we have a problem where politicians have no shame about lying.
21:04I just wrote a book about lying in politics,
21:06and one of my conclusions was that with partisan media and the Internet,
21:16that it's just so much easier to build support for a falsehood.
21:21Yeah, I mean, any views?
21:23Yeah, I was just going to say it has always been a problem
21:25because human beings, we are by our nature tribal,
21:28and we like to believe.
21:31When we're reading something, we like to agree with it largely.
21:35But I also think that people don't like being lied to,
21:38and all of the research, and there's extensive research now about misinformation in media,
21:43will say that people don't want to be lied to, and they don't want to be misinformed,
21:47and they're frustrated because they're besieged and overwhelmed.
21:52And sometimes it's not just the journalism,
21:55it's they are looking at the news, for example,
21:57and seeing that Wall Street is booming as one example, which it is,
22:00but they're not feeling that.
22:01And so they're thinking, is that real?
22:04Like, you know, why are these economies moving when I'm feeling so?
22:07They don't want to read about that when they might be feeling hard done by themselves.
22:11So it's a bigger problem than just journalism.
22:14I think it's an economic and governmental problem as well.
22:17But yeah, we're kind of at an inflection point,
22:21or a perfect storm, if you like, at the moment,
22:23of the public being frustrated with news information.
22:27So this is your perspective that people still want the truth,
22:32they still care about the truth?
22:34Yes, very much still care about the truth.
22:35And I think it's more like, if they're turning away from it,
22:39or turning away from trying to get to the bottom of it,
22:41it's because they feel like they're in a firehouse,
22:43and it's just, they're overwhelmed.
22:45But I think they still want it.
22:47And as I say, I think it'll come full circle again.
22:50And maybe that's the optimist in me speaking.
22:52And one of the things that AI has done is it has made people realize,
22:57made people wake up to the fact that,
22:59because especially with images,
23:00everyone has seen false images that are obviously false.
23:03It is incredibly difficult to verify them.
23:05I mean, not the obvious ones, but some of the less obvious ones.
23:08But everyone has seen them in their day-to-day lives.
23:11So there's an even greater awareness
23:12that there's lots of misinformation out there.
23:14And in some ways, that's helpful,
23:16because it makes people think,
23:17okay, I need to think a little bit about harder,
23:19about what I'm reading, and what I'm watching,
23:21and what I'm listening to,
23:22and ask more questions,
23:24and think about this a bit more.
23:27Mario.
23:28Just to add to that, it reminds me,
23:31I think about 10 years ago,
23:32I gave a presentation called Newsrunner 2029,
23:37which suggested that in the future,
23:39when everything is, the reality has collapsed completely,
23:41we're going to suddenly come around and value journalism
23:44that we can really rely on,
23:45and it can be our basis of reality.
23:47It'll be such a premium.
23:48We'll hold on to it so dearly.
23:50And there'll be scavengers, basically,
23:52looking for this information.
23:53And so I think there is something to be said for that.
23:55What I wanted to add is,
23:57I think everything is fundamentally human.
23:59We're fundamentally tribal.
24:00But I think in that way,
24:02we also have to fight fire with fire.
24:03I think too often,
24:04our efforts to combat the forces of misinformation,
24:08the toxic incentives,
24:09are very gentle,
24:10and we expect people to be their best selves.
24:13And media literacy in schools,
24:15that is a good start,
24:16but it expects you to remember the theory
24:18in platforms that were designed
24:20by the best designers in the world
24:21meant to grab the back of your brain.
24:23So you don't have time to remember the theory.
24:25And so I just wonder,
24:26what does it look like to...
24:28I think people will care about journalism and information
24:31if we make, I don't know,
24:32a profile about how they inform themselves
24:34cooler than their Instagram profile.
24:36If we make, you know,
24:38there's Spotify wrapped.
24:39Why isn't there a knowledge wrapped every month?
24:42You know, self-discovery is so addictive.
24:44BuzzFeed quiz to find out
24:45what Lotus character you are.
24:47Tell me how my information maps
24:48to like Obama's or something.
24:50Like, how do you make my intellectual journey
24:54actually rewarding
24:54so we fight fire with fire?
24:56Because that's what we're up against.
24:57If we take it gently,
24:58we're just going to get crushed.
24:59It's like a water balloon in a knife fight.
25:01So, yeah.
25:02Yeah, I mean, that's the challenge, really.
25:05It's just like the enormous power
25:07of the platforms against...
25:08I mean, Bill, I'd love to get your sense
25:10of like the kind of like
25:11the global fact-checking sort of, you know,
25:14ecosystem at the moment.
25:14Often, you know,
25:15these organizations are pretty under-resourced.
25:17They're pretty small.
25:18Like, particularly if they're working, you know,
25:20in inhospitable environments,
25:22maybe with, you know,
25:23governments that are not friendly towards them.
25:25Like, when you're looking kind of globally,
25:27what do you see?
25:30It's a very timely question.
25:32We, at Duke, my team does an annual census
25:37of political fact-checkers.
25:39And for the first time this year,
25:42for the first time in the 12 years
25:45we've been doing it,
25:46the number of fact-checkers
25:47has declined significantly.
25:50And it's really worrisome.
25:52And it's a function, as you were saying,
25:56of resources.
25:57The journalists who do this
25:59don't have resources.
26:01They are, for the most part,
26:03not commercial organizations.
26:05So they don't have the backing
26:07of a major entity like the New York Times.
26:13Some commercial entities that did it,
26:15such as the Washington Post,
26:17no longer do it.
26:18And the chilling effect
26:21of what Donald Trump has done
26:25in the United States
26:26has really, I think, had an impact,
26:28not just in the U.S.,
26:29but around the world,
26:31in that there is less philanthropic support
26:36of fact-checking.
26:37And that's the only way
26:39that this important journalism gets done,
26:42because there's not a commercial model for it
26:47that works and raises enough money.
26:49It's like investigative journalism.
26:51It has survived because it's been part
26:54of another larger media organization
26:58or because it had philanthropic support.
27:03Ron, I want to come back
27:05to the business model again,
27:06if you don't mind.
27:08Most of the examples
27:09that we can kind of come up with
27:10of, you know, successful models these days
27:13are, you know, titles like the New York Times,
27:17which, you know, require a subscription,
27:19and, you know, there are many others
27:20doing great work that require a subscription.
27:24Is the challenge really
27:26that we're also kind of creating
27:28a two-tier information ecosystem
27:30with kind of like elites having access
27:32to good information
27:33and everyone else just getting AI slop on eggs?
27:37I don't think so,
27:39because I think if you think back
27:40to the heyday of newspapers,
27:42which were the main vehicle for news,
27:45they always charged.
27:46So there was always, you know,
27:48there was always a cost associated with that.
27:50With the exception to that
27:51is public service broadcasting,
27:53which is free,
27:54but in places like Ireland and the UK,
27:56you still pay a license fee.
27:59But getting cut back every year.
28:01Getting cut back,
28:02and despite their relative scandals,
28:04they're still highly trusted.
28:06But I don't,
28:07I think you've, you know,
28:09serious news has always been a cost,
28:11I mean, a cost for the consumer.
28:12It's never really been free.
28:14It's just that there is more,
28:16there's more avenues
28:17to news inverted commas online,
28:21and a huge amount more video news,
28:24for example,
28:24which certainly is the most popular thing
28:26with youth or the youth audience.
28:28So, and in that context,
28:30it's hard to get,
28:32it's a question of education,
28:33it's hard to get that audience
28:34tuned into paying for news.
28:36But I don't know if it's a,
28:38you know,
28:39it's a two tier,
28:39as in the elites will get the news
28:41and the non-elites won't.
28:42Because I think it's always,
28:43there's always been a cost attached to it.
28:46Okay.
28:46In terms of the business model question,
28:48Mary, I'd like to come to you.
28:49Obviously, like,
28:51you know,
28:51the big platforms
28:52have based their,
28:53you know,
28:54business model on engagement.
28:56Obviously, outrage spreads
28:57much more quickly than accuracy.
28:59Do you think that fundamentally
29:01the business model
29:01of social media companies
29:03is incompatible
29:04with a well-informed public
29:06and building trust?
29:08Absolutely not.
29:10I mean,
29:10I mentioned it earlier,
29:11but there's,
29:13you know,
29:14the famous quote,
29:15the medium is the message,
29:17but the incentive
29:20is the medium
29:21and the message.
29:22And the metric,
29:24the metrics
29:25are the incentive.
29:27And so we just
29:28need better metrics.
29:30Whatever you make the metrics
29:31is what people are going to chase.
29:33Right now,
29:34the only metrics,
29:35again,
29:35are the quantity of attention you get.
29:37So it's a race to the bottom.
29:38How informed you are,
29:39the quality of your sources,
29:40how balanced you are
29:41is functionally irrelevant,
29:44which is why often
29:46it is to your detriment
29:47because you're just wasting resources
29:49to chase the good information
29:51and do long-form content,
29:52unfortunately.
29:53And I'd love to hear
29:54what you have to say.
29:54Like,
29:55how do we put other metrics
29:57that are going to make people
29:58have a race to the top?
29:59And we can't have that.
30:00Again,
30:00the money's there.
30:00I explained,
30:01and if you want to have more,
30:02I can send you the numbers.
30:04The money is there.
30:05We need better metrics.
30:06Well,
30:06I like this topic.
30:11If Mark Zuckerberg
30:12were sitting here,
30:14what would you tell him?
30:18Well,
30:19first of all,
30:19I would say
30:20stop pretending
30:21because this notion
30:24that a company
30:26that can beam
30:27internet
30:28to remote parts of Africa
30:29and has like
30:30a trillion dollar market cap
30:32can't design slightly better
30:34and healthier feeds
30:36and transparency.
30:37Now,
30:38I like to always use this analogy.
30:40I think what Mark Zuckerberg
30:41and a lot of the folks
30:42now are doing,
30:44it's like somebody has a boat
30:45and there's a hole in the boat
30:47and instead of
30:49plugging the hole,
30:51they hire 15,000 people
30:52to scoop water
30:54because they make money
30:55off your feet being wet.
30:57That is not,
30:58that is so easy to fix
31:00but there's always this,
31:01you know,
31:01this show of like,
31:02oh,
31:03we can't do it.
31:03We're doing our best.
31:04Let's set up a board,
31:06an oversight board.
31:07What I would say
31:08is just stop pretending
31:08and get to the core issue.
31:11Give people more transparency.
31:13Give them control
31:13of the algorithm.
31:14Let them see why
31:15they are seeing
31:16what they're seeing
31:17because without the transparency,
31:19but again,
31:19this is going to fall
31:20on deaf ears
31:21because this is going
31:22to make the business model
31:23for them less effective.
31:24It'll be better for the world
31:25but they'll make less money.
31:27Bill,
31:27what's your take?
31:28What would you tell Mark Zuckerberg
31:29if he was sitting here?
31:31The first thing I'd say
31:32is reinvest
31:33in political fact-checking.
31:35In the United States,
31:38Zuckerberg pulled back
31:39from supporting fact-checking,
31:42much to the delight
31:43of our president.
31:46And that has really hurt
31:48and the world's fact-checkers
31:52depend disproportionately
31:54on Facebook
31:55and so Mark,
31:57please continue to support
32:00fact-checking.
32:01It's really important.
32:04Rona?
32:05Grow a conscience.
32:07Rebuild the guardrails,
32:10the ones that were taken down
32:11that Bill just referred to.
32:12Yeah.
32:13I would also add,
32:15Mark,
32:16that you can actually
32:18make money with quality,
32:20especially these days.
32:21People are so desperate
32:22for quality information.
32:23Can you imagine a situation
32:24where instead of having
32:26your attention stolen,
32:28because you went
32:28and started informing yourself
32:30on nutrition,
32:30let's say,
32:31Whole Foods could come in
32:32and say,
32:33we love that you're
32:34trying to inform yourself
32:35from trusted sources
32:36on nutrition.
32:38Here's $10 off
32:39at Whole Foods.
32:40We want you to think
32:41of Whole Foods
32:41when you try to have
32:42a healthy mind.
32:43That doesn't have to be
32:44this adversarial,
32:45let me sneak into your attention
32:46and steal your time.
32:47That's like better
32:48brand retention.
32:49There's so many good things
32:50about that.
32:50You can do that.
32:51I think people would like that.
32:52Let's try that.
32:53Let's at least start there.
32:56Unfortunately,
32:56Jeff Bezos owns Whole Foods.
32:59Wrong brand.
33:00You're right.
33:01That's right.
33:02All right.
33:02We're nearly out of time,
33:03so just very quickly,
33:03I want to go around the group
33:04and I just want to get
33:05your sense of,
33:07how do we get this right?
33:09How do we fix this?
33:11We do have a lot of people
33:13with highly skilled.
33:15We have an appetite.
33:16Ronan was saying
33:17that people want truth.
33:18They want good information.
33:20What would be the one thing
33:22you would suggest
33:23that we do
33:24in order to sort of like
33:25stop the spread
33:26of misinformation?
33:27Bill, do you want to start?
33:28I think we need to explore
33:30different ways
33:31to present it
33:32more effectively.
33:33I think the old ways
33:35of presenting
33:37political fact-checking
33:38and other kind of
33:39corrective information
33:40are not working
33:41and I think we need
33:42to look at creative ways
33:44to take advantage
33:46of this moment
33:47and spread it
33:48much more effectively.
33:50Mario.
33:51We need to make it cool
33:53to be smart and informed.
33:55Until your feed
33:57and popularity
33:58can only be
33:59based on how loud
34:00and divisive
34:01and provocative you are,
34:03it will be detrimental
34:05to be intelligent
34:06and informed.
34:07And so we need to fix that
34:09and so that is what
34:10I would fix.
34:10Make it cool
34:11to be smart
34:12and to make an effort.
34:16We need to avoid
34:17serious news
34:18becoming like art,
34:20you know,
34:20like ceramics
34:20as opposed to
34:21dishes made in factories
34:25and otherwise
34:25it will become
34:26way too elitist.
34:28And in order to do that
34:29I think we need to
34:29engage audiences
34:30even more than
34:31we're doing now,
34:32bringing them
34:32behind the scenes,
34:34explaining what's real
34:35and what's not real
34:35and just not giving up
34:37on that.
34:38Great, thank you.
34:39Three levels of optimism
34:41for the future
34:42and I'm sure that,
34:43you know,
34:43we'll be back next year
34:44having similar conversation.
34:46Thank you so much,
34:47guys,
34:48great conversation
34:48and thank you for joining us.
34:50And thank you,
34:51Mark Zuckerberg.