How media businesses in Asia are using AI | Global Journalism Seminar with Rishad Patel, co-founder, creative director and head of product, Splice Media

11:30AM May 24, 2023

Speakers:

Mitali Mukherjee

Keywords:

ai

newsroom

audiences

generative

people

media

journalists

work

chat

journalism

newsletter

happen

terms

policy

organization

point

impartiality

experiment

news

fact

Welcome to the Global journalism seminars. This is the briefing on November searches last year, open AI quietly released chat GPT a language model capable of human like text interactions for generative AI. Microsoft reads to add it to Bing. Google tripped over its feet to launch Bard, Facebook and Apple announced a billion dollar investment in their own generative AI and Amazon fast tracked bedrock Western media were equally breathless box called Chat GPT stunning and wild said it would change the way we interact with the world. Gaiden warned it might be used to spread propaganda. The New York Times worried that might be used to manipulate in Southeast Asia. The news industry has been using generative AI for content generation, translation and localization as well as other AI technologies for content recommendations. That checking and more at the straight times rappeler and the Jakarta Post to name a few. You polled 13 Journalist funding just ask if their news outlets were using generative AI technology 63% said no 37% said I don't know what that is. For an SCA Chinese companies AI bustled this week we're turning to spice media's reached out to the tell the season product and design professional based in Singapore who writes a weekly newsletter on product development bias frames. That's the briefing. Let's begin.

Hello, and welcome

to another one of our global journalism seminars here at Reuters Institute of journalism and no surprises. We are going down the AI path for good or evil. It's a packed house online. So welcome to everyone who's joined in for this conversation. It is going to be scintillating. For many reasons not least because Rashad is here to speak about AI and journalism but also for two things that come to mind. One, the fact that I think this is one of the first conversations happening, that is really exploring what's happening in the non Global North networks, not the large media organizations, not the ones that really pull a lot of heft but the smaller nimbler organizations and what they can do with AI and journalism. And second to the fact that I hope Rashad and I can chat about which is what does this mean for journalists per se? I don't mean the tips and tools that you're using. But what kind of stories should we be telling in a responsible fashion around AI to really build awareness for what is ultimately your key goal informing your audience better sureshot welcome. All introductions to your core skills have been done. But I may add another one which is a funny story about how Rashad and I met, which is that we follow each other and we know each other's work but we finally got to meet in person at the media conference in Perugia and the both of us spent the better part of one hour going up and down the street, trying to recognize who we were from very, very small thumbnail sized images of each other, which kind of feels like the conversation around AI at this point Rashard where people are sort of scrambling to try and identify something they may not have, or no one does perhaps have a clear handle on right.

I mean, I feel that that's a very good segue into this conversation. Yes, we we knew each other from thumbnails, but then I resorted to, to to the crowd. And so I put the word out. So it was the wisdom of the crowd that that saved us. And I was like, well put the word out. Do you know what Metallica kanji looks like? Here's, you know, here's what I'm trying to find. And sure enough, I had enough people come to me and say he or she is and then we met. And I think that's kind of where we're gonna go with this. Chat about about AI. Thanks for inviting me.

No, not at all. And I think in one of our previous chats, you sort of lined it out quite succinctly in terms of the mood, which is that it swings between his Tyria versus nothing. Is there a snapshot even to define how AI and journalism are sort of engaging with each other at this point?

I feel like yes, indeed. There is. You know, there is a massive amount of hysteria and you know, while we should we should have a sort of diplomatic nod towards it. I, I trust we're not going to go to either extreme. You know, either the hysterically excited and you know, announcing the second coming of whatever it is this new technology brings, but also the, the fear mongering that we see so much of and that's the theory at the other end of the spectrum. I, you know, obviously, you know, as we know, two things, two big things have happened since, in fact, we last chatted about this. And since since we agreed to this talk are one is that opening I announced that chat GPD can now read the Internet can access the internet, which is you know, the big, the big. The big scary move towards real time knowledge. And then the idea that there are now third party plugins and this now opens it up to different use cases. For a whole different area of access. There's a whole bunch of obviously, a whole bunch of applications that come from there. But I think the interesting thing when I pulled a huge bunch of, you know, a bunch of the folks that we work with that splice splice helps, you know, helps people build viable media businesses. And so as a result, we work with a lot of media startups. We're interested in the journalism creators in Asia. We work with the media startups, mostly in Asia, and around the world. And we work with a global ecosystem to media ecosystem to make all of this happen. And so of course, you know, the nice thing that happened when I spoke to these folks and Mongolia and Nepal, and you know, all of that was I like that people are currently the good news, of course, is that people are currently in experimentation mode. Nobody's making big bets on AI right now. At least not around content. I think that's good news. I think that people are not ready to operationalize this yet. At least among the folks that I spoke to, and that might be a good thing. So folks are using it around podcast show notes. There's a tentative use of all of this, you know, how can I experiment with this around podcast show notes? How can I use it to summarize so you know, the the phalanx of tools that now allow you to summarize stuff? They've been around a long time, social media illustrations, little posts, landing page introductions, and populating CRM tools or databases and stuff.

And so I want to dive into all of these one, the experimentation mode that you're seeing in some of these smaller Asian newsrooms, and to see if we can keep connecting the dots saying here's the experiment and here's the learning because I think as you know, it seems quite evident from the people you've spoken to and the people I've spoken to, there is a degree of nervousness about taking that leap. You're not sure if your pockets are deep enough to absorb any potential losses. You're not sure if you have a full handle on what the end point and the outpoint looks like for small organizations, and organization. And most importantly, you're not sure if you have the acumen to really steer through something that is completely new, you know, new waters for everyone. I know that we discussed beforehand that we, you know, chat through some of the experiments too happy to kick off with anything that you'd like to in terms of what you saw what you liked about it, and what you think a learning could be for a smaller news outfit.

I think you're right about about those three things. I've got to add a fourth, if I may. I think there's also a little bit of nervousness around credibility. I don't know how many reporters and journalists right now and I could be wrong. I'm a product guy. I'm not a journalist, but I've worked closely with famous I know I can see a whole bunch of people dropping off right now. But what's a product guy doing here? But I think that there is a credibility risk with any kind of new technology, any kind of new tool or workflow or editorial structure. And I think that perhaps it's yes, I don't take it very seriously. You know, and I think that is some of it. I mean, I'm going to use a quick case study from not the global South, not Asia, but you know, gazetta We bought this polish newsletter did this famously. The deputy editor has a cool quote that says, you know, refusing to use generative AI is like not using search engines 30 years ago, so very much in you know, in the positive, you know, camp so they started using chat GPT to write articles a while ago in in early April, one reporter and he wrote a text called famously, Elon Musk versus open AI, the race to create a new human computer interface. You know, and it ran on the website and ran with the reporter's byline, no explicit display disclosure. And the goal of experiment was, of course to find out if Chad GPT would be able to write a quality article in the style of the reporter. To everybody's relief. It turned out not to be the case. But I think that once in a while, we've all played around with that. I've tried to get it to write my newsletter. Full disclosure, it was, you know, but horrible, horrible. Frankenstein's monster. I think that so I think that there is that, that slight embarrassment, and I have a feeling that that's going to change. When you're happy to talk about some of the

examples that have worked. I know that you've looked at a few in India, for example, like cupboard and strollers Well, what is it that you're working for those for those outfits?

Interesting. It's so Kabul the Harrier was one of the first that I asked and this is just to put it in context and Indian newsroom for and by women in rural India. newsroom that we admire very much at this place. And I asked them and you know, she said that they've been experimenting with tragic beauty for socials, which is a refrain of her through the research I've done through all of our folks, our community. Designers been using mid journey and dream for a couple of months now for for social images, experimentally. And that's it. They've backed away from using it operationally in the newsroom. What I really loved about scroll on the other end was added scroll down and of course, is a digital newsroom in India. I spoke to them quite extensively. And the person I spoke to said that, you know, they've done a lot of, they had a problem to solve, which was interesting, very specific problem, which was, they did a lot of layoffs in 2019, especially around video stuff. And then they got funding to build a multilingual text to video tool. So that's a big one that they're working on. So that means a whole bunch of things. They're going to be able to re rework and revitalize their eight year old archive of articles, a total archive of images. And then the working theory here is that is generative AI going to be able to make short videos to feed that demand in India for short video, but around multiple languages?

Let me pause before we jump to the next experiment and sort of summarize this and you know, correct me if I'm wrong, which is that one good way to approach this is to start small two or three projects don't be very ambitious. The other thing to do is get the right people in in terms of this conversation of what you're trying to construct. And the third perhaps is to as you pointed out with the example of scroll, secure your data sources, you know, does that look like a good few stepping stones would you add some more?

Very much so. I think, yes. The only thing that I would add, and perhaps through the conversation is that I think the huge you know, content is one tiny part of the conversation I think the huge huge up opportunity we have here with not just generative AI but and not just general AI for content is to be able to look outwards from the newsroom and outwards from content and use this technology to make what we're doing this journalism away doing. Either more relevant relevant for existing audiences, or to be able to to reach new audiences, the power of AI to currently especially with these two new, new factors, and I'm happy to address that later on the idea that this, you know, connects us to the internet. And so we're current and up to date, of course, with a whole bunch of with a whole bunch of factual errors, but also that there are third party plugins that allow us to make our offerings in this day and age when you know, so many media outlets are struggling to survive and allows us to make our offerings, you know, reach the right people. And my big, you know, we're in the business of seats at the table. We're in the business at supplies of making sure that journalists make money to put it bluntly, media organizations make money and they diversify their sources of income and revenue. My big nightmare as I we've done splice for eight years. Is that what if you're doing the best journalism in your life, and it's not getting to the people that need it most? So that's where I'm seeing the massive opportunity here. Are there examples

that come to mind that have been able to do the same some more a large degree?

Not just yet. I'm, I'm, I'm I'm going to bring up my friends at the Nepali times. in Kathmandu, in Nepal, obviously. And this is a newsroom that's been around for ages. Since I think the early 80s But I need to fact check that these folks when I asked them, you know what you're doing, and they've been very much a print product. It's been very much a traditional, what we like to call legacy media. And what these folks have done straight up is to say that they were looking for new audiences by doing Instagram, AI generated Instagram videos, but from the existing stories. Now, the assumption there, of course, is that there are younger audiences on Instagram. That may or may not be true. They may or may not be interested in us, but I like the direction that that's taking.

Let me jump off that to make another point. You know, and hear your thoughts in it, which is, is it essential for a news organization to identify the pain points they're trying to address when they're thinking of how to integrate AI? Or should they just cast their net very wide and say, Let's experiment here or there or there and see where we land?

I feel that it's, that's a great question. Thank you for that. I feel that it's important to be able to cost wide but but also skim shallow look at look at the areas where you can use this. I think I worry about sometimes solution in search of a problem. And I don't know that AI you know, there's a lot of like, Oh, you've got to use AI before AI uses you. And there has been some talk that before it replaces you. I don't think that that's there are very real problems out there to solve for our audiences. And I think that to go back to the second part of what you were saying yes, I think it is a huge opportunity to be able to use that content and use AI and not just general to the AI to be able to explore where your products, how to get them to these audiences, how to solve problems for them. How do you we have this superpower now available to us being able to ask specific audiences what they need help with? So things like, you know, how do I vote? How do I make sense of an election? How will you help me make this budget this national budget relevant to my life? So can I use AI and all the tools that are at my disposal now to serve that audience? How do you make do I do rent or mortgage? Do I move my kid to the school less clear, closer to home or to the better school much for the way and ruin their chance to have you know, a life with their friends, etc? There are real problems to be solved. And I think for audiences, and I think that I think there's another problem, which is even within our journalistic circles, and the conferences we all go to, for example, to use AI to make sure that our newsrooms, our references or news gathering sources, experts, panels, and products are inclusive, diverse are they truly representative of the people that meet so

before we get to this very, very important point about audiences to share just one final sort of observation or comparison from you, if you if you will, Reuters recently sort of released a letter in terms of their approach to AI and they said, were very comfortable with autonomous content. It's something we've done before. So our approach is going to be be skeptical and verify. Is there a marked difference that you see in the approach from large global organizations versus the more nimble footed in terms of what AI is doing in their newsrooms? Or is there not a big difference? It's a scale that that matters.

I find that and if we're talking about the scale, of course, you know, we can't help but bring up BuzzFeed and their AI generated articles and the whole ridiculousness around that. But have we grown to accept the expect this from BuzzFeed over the years perhaps? Do we take them seriously? Do we not is it AI or is it something else that maybe added that bit out? But I love that I love Reuters is Reuters memo that they send to their staff? I think that's a I think that's at the grown up end of the spectrum that's at the non hysterical, you know, middle of that spectrum, and I love that it that memo was about responsibility and transparency, and, you know, values of skepticism and verification, right up front. I love that. The scroll down in did exactly the same thing. And I have deep respect for anyone like that. I was told they were in the middle of drafting the use of AI policy. And that that should be out soon. I'd love for that to be a standard that we could use. Across collectively across across newsrooms.

Let me pivot then to the point about what's happening with our audiences. And you know, let me put two or three things in front of you one, I think I heard you correctly when you said the Polish news outfit, publish that story with no explicit admission to the audience that they had done it. Second, your point about this being relevant for audiences and actually impacting their lives in a meaningful way, either by informing them better, or by changing their lives. And the third, which was the recent Open AI Senate hearing which was not at all, as as sort of lively as a Facebook hearing might have been or a tick tock hearing might have been, where Altman actually called for regulation, but there was some other points which he made such as, let me find that exact quote. The public would learn to adapt to an onslaught of AI aided false information and media. Are we treating our audiences with enough respect and care when we're looking at AI from inside a newsroom?

I think I think the worry here is multi fold. And I think that the worry concern here is that yes. Are we expecting people to be grownups and to be able to tell, you know, whether the Pentagon was in fact bombed or whether it was just complete rubbish. As as we now know, are we expect

you know, journalists from my land may disagree reshot we had strategic experts describing what they saw. But hey,

so that's what I mean. Are we are we now you know, and we keep talking about that spectrum of hysteria, the hysteria spectrum, but I think that, yes, I love I'd love to as a product and audience guy. My work is around how to make journalistic and media products relevant to audiences, existing audiences, but to be able to find new audiences. To to serve. I think there's a there's also the the tendency to treat an audience as though it was this monolithic, single mass audience. And I think that there is we need to very quickly it's been 2023 Ai, oh, no AI, we need to be able to move away from that, that, that idea of publish it and they will come they want this. They don't want that. The number of times we've heard people in Myanmar won't pay for new people in Poland won't pay for news, it it's there is no one single audience. So yeah, is there is there an audience that will believe everything you you say about AI and panic when you tell them to panic and you know, trust when you tell them to trust? Yes, of course there is. But I think there are separate audiences who wants to be able to understand this better. And as journalists and as media organizations, we're in this incredible position. To be able to be that Representative entity that makes sense of the world, you know, to be able to be a whole different kind of gatekeeper. You know, to change that. That old, that old idea that we decide what's good for you where the fourth estate we tell you what's democratic or what's not. So I think that there is there is this, again, this huge opportunity to be able to collaborate with audiences, to be able to go to them for once and say, Listen, we're creating all of this. What can we do for you? How do we make journalism, that service industry that was always meant to be?

You raised a very important point about credibility reshot and I might add impartiality to that and trust to that. Are there examples aside from you know, building policy towards AI that come to mind, where newsrooms have actively done that in an environment where, you know, it's AI or its algorithms, where at some point, the human being is saying, Hey, hang on, can you read my mind? What's going on? I'm seeing exactly what I was thinking about.

Yeah, I think I think we have, of course, we've been suspicious around that, you know, we have had that urban myth of, you know, there he was talking about so and so and suddenly I've got a follow at you know, on my phone. I think, again, it's going to happen. We've had that technology well before generative AI or this, or this generation of AI, made itself clear back in November. It's been around for ages. You search for something and Google wanted to tell you that you did a little search for where to eat Korean food, and then you got Korean food ads for months on Facebook. I think that's always going to happen. I think the conversation of disinformation and misinformation is going to be freshly fueled. Now, I think, with this idea of AI is going to ruin our life. And it's it's the next coming of Satan or whatever variation on that. But I think that there is a I think it's possible to be hugely optimistic about the fact that this allows us again, not speaking as a fanboy. But I love that the current for example, a Pakistani video newsroom, run run by you know, one of our folks in our community called Maryam Chaudhry. I love that. They're experimenting with video news, but in Urdu, and they're still working their way around that, but they're using it for how to stories. And a lot of people are using it now for guides. So when you start populating you know the you know, the the all your content surfaces with here's how you apply for your Pakistani passport renewal online, which is an example she gave me. How do you get your driver's license? Here's how you apply for a visa. How do you pay your taxes? This is now you're getting into an area where you can do this news reporting in a responsible way. And you can use technology that's available to you, but you're also building trust. And there's two parts

to it when you're looking at journalism to my mind to SHA one as you say is how am I using this as a tool to the best possible advantage of my organization and my audience, hopefully. And the other I think is how am I telling this story as it develops? I mean, I don't want to draw comparisons, but we've sort of seen that with the crypto boom. There was suddenly all these journalists reporting on what was the next best thing after gold biscuits. You know, this is crypto. This is everything. On a bad day. This is nothing. I'm not sure by the end of this entire journey, how much audiences really felt informed about what was happening with this industry, what it was made off. What kind of story should journalists be thinking about when they're talking about AI? You know, in terms of what is the development what's the context of this development? What does it mean for us really? You know,

one interesting, I have two interesting examples of of that, and you're right about that, you know, we were bombarded with, you know, with crypto being bombarded with you know, nobody says when three anymore or dares to say NFT any longer. We've we've been through that, and it's recent, this is all, you know, literally three years. We had all of that. But what I loved is to create little tiny case studies. One is the fix which is a little newsroom that we that I like to follow a bunch of friends in Poland. They have a level headed guide for journalists on how to cover AI. I liked that they're using this. I like that there. Is that the people are experimenting around benchmarks. And as with every kind of technology, including social media, from all the way back then we're going to have to find a way to make policy around it, how to make money around it, how to make how to how to find values, kind of REITs style guides. How do we use this? How do we tax it? And I think that I like that, you know, the average article around the fixed says the average article insured around Chad GPT has tended towards this stupid CEO centric hysteria of gushy determinism. How do we counter this? How do we write about the dangers of AI, but not from an ideological or emotional standpoint, but from technical elements? How do we how do we make our audiences comfortable with this? How do we share that language? Likewise, Casey newton of the platform newsletter, fame, he he I loved his recent experiment. He asked his readers and you know, in my book, that's awesome. Talk to you talk to audiences more about this. And he asked them, How do you think I should cover AI? And they told him and it was a beautiful list. It it sounds like a user's guide to how to think about AI. Be rigorous with your definitions, predict less, explain more. Don't hype things up. Focus on the people building AI systems, and the people affected by it. Offers strategic takes not emotional takes on products. Emphasize the trade offs involved, which is exciting. So be balanced and newest, newest, and tell us which trade offs make sense for us and tell us which don't. And then finally, remember that nothing is inevitable.

The term I like is level headed guide. I'm going to appropriate it can attach it to everything that I communicate with. I had a bunch of questions about policy as well but I can see questions pouring in reshot so I'm not going to hog all my time all your time here. There's one that's coming on LinkedIn where accurate Michael asks Rashad. How do you estimate AI will reshape and is already disrupting news SEO.

Oh, right. Absolutely. And this is this is an exciting thing. We're moving from now. SEO to citations and the interesting thing there is obviously you know, you you have this is devastating for publishers, obviously, you know, you now can't necessarily depend on your traffic coming from, you know, it used to always to be links and social and search, which is where you've got your advertising and your and the audience's um, I think that there is a there is now a whole different idea around what we should do about this and I think, again, being the eternal optimist around around media. I think there's three things we can do. And I think one of those is diversify formats. So it comes back to what your what your formats are. Publishers will need to productize around audio around video events and direct delivery vehicles, like email. I love that. Every time we talk about programmatic, you know, AI driven algorithmic driven media. We tend to come back with every generation of technology to the oldest tech around which is email delivery. And this explains the rise of newsletters right? Then your webpages are now no longer going to be able to be you know, cluttered. You know, pages that we're used to, you have to go. You plow your way through display ads, autoplay videos, please sign up for my newsletter, register here or subscribe there. Let me give you desktop notifications. We're going to have to do away with all that and make sure there's more dynamic content that's updated, and then personalize the experience is the third point I'd make around that. When we have this opportunity to be able to personalize it's fantastic. So static content bad generative AI, but personalized and altruistic. Might I add? So not creepy personalization. Hey, Rashad, nice shirt, you know, it's not what you want to hear when you take a walk down. But the idea is to sometimes do things that don't scale its place we do things that don't scale, which is we, we write to every single person that signs up for our newsletters. And we're proud of that. It's it's a lot of work, but it's a few seconds, you know, and we write we say, Hey, Daddy, thanks for signing up. Where do you work and what do you do? You know, love to meet for a coffee, that kind of thing?

Yep. That's exactly what Alan did when I signed up for your newsletter. That's exactly. I do hop across to our own conference room where global journalists fellows have lots of questions, but I just want to quickly slide in one Ryszard about cost. All these hot takes that we're getting from Chachi PT, for instance, are cool because they're free. What What should newsrooms have in mind in terms of how much this whole thing is going to cost them? That's where as a product guy as you know, it ultimately does settle very often.

I think I think there's hard truths involved there. And I think there's actually relief involved there. I'm not in a position to make predictions I but I think that some of the costs are you know, involve people actually being the best kind of gatekeepers to be able to fact check. Hallucinations from from Ai, which we know that it is prone to do. I think that we will need a whole bunch of folks around that to be able to incorporate that into not just our tools and workflows, but into our editorial policies. I think that's hugely important. And that's a cost. I think that there's also the opportunity to be able to share that cost with audiences again. And by audiences I mean, do you have a b2b audience for your what your newsroom puts out in terms of say, reports? Not the atom of news is not necessarily the article. You know, the atom of news can now be diversified into multiple articles for multiple audiences. You can you know, why do a story on a budget that is a picture of two two dudes in suits shaking hands with four legs, a copy and a headline when you could make me the user understand that budget way easier with a calculator? Why? don't publish a calculator? Ai makes it super easy to do. Why not have aI crawl your stories as a matter of workflow in your CMS with a plugin that says your gender balance is really off here? Why why do you not have more female sources? Why don't you Why do you not have more variance in terms of in terms of religion in terms of representation of all kinds? So I think that that cost as with all tech will become cheaper over time. And I think that that content has to keep pace with that.

Let me hop across that room now. I think you'll see is waking by with the question you see.

Yeah. Thank you so much for joining us today. You were discussing already earlier, our newsrooms are taken care of for the company they're using AI for that, from the perspective of when something goes wrong, for example, when false information has published the product coverage there, as you can see, is on well you think he's going to explain to our audience what's been formed and only their mistakes don't happen. Is there enough?

Thanks, you see, it's a slightly scratchy lines reshot if you want me to repeat any of that. You know, just say the word.

Yes, I had a little trouble understanding that the Yeah, the line is a little scratchy. But how do we get newsrooms to own their mistakes? When there has been, say, a hallucination or a, you know, a complete re rendering of facts. Is that Is that roughly where your question is?

You see if I may just take the liberty to read out what you've written because we can't hear you clearly. You were discussed. You Ryszard. How well newsrooms are taking their audience into account in the usage of AI. You see is interested in a situation where something goes wrong with coverage where AI has been used, for example, and false information gets published. How well are newsrooms then equipped to explain to the audience what went wrong and owning their mistakes when they make those mistakes? Is there enough tech savviness to be accountable in terms of AI? Usage?

I think, I think there's there's two possible answers there. One is that you you you have a very clear policy and this is why AI usage policy is so so important. The way Reuters did it and the way scroll down is doing it I think that that policy involves a two pronged approach. One is taking ownership of the mistakes that you've made. predates AI by hundreds of years. Should you take ownership? Yes, you should. That has nothing to do with AI. Would you be able to use AI ultimately to be able to catch the mistakes that AI makes? Or the the you know, the reporting fact checking mistakes that journalists make? Sure, why not? Why should you not? And I think the other so there's those are the two usage. Use cases that I would see is that you have a very clear policy. What do we do when this happens in the newsroom? That is very much shared in in the workflow and in the principles of the newsroom, as as a an important, our part of the editorial process and how do you actually factcheck which to my mind has very little to do with AI. fact checking and apologizing for those mistakes and making them right is a responsibility that I assume journalists tab by default.

Another question that's coming online reshard. From Maria Maria vein, I don't know if it's the same person you were talking about earlier in terms of her work. No,

no, no, that was Maria Chaudry.

How media organizations are coping with. And I'm sort of just summarizing what she's written, building a basic level of journalism literacy around the subject, especially when the audiences that you're reaching out to, as you said, are not the sort of singular block where literacy levels and awareness levels might be quite, quite varied, especially perhaps in some of the global south countries that we're talking about.

Yes, we've worked with a whole bunch of folks in in audiences in India, in Nepal, and in in of course in Pakistan. And it was exciting. We worked with an organization called looks to jog in Pakistan, perhaps you guys know them and we, we help them prototype a product around radio, and which is one of you know, audio is, of course, one one interesting way to be able to deal with problems around literacy. The beautiful thing that they were able to do is have this they created this radio show. And it was in Punjabi in Pakistani Punjabi. And the idea was that the you're talking about getting it not to why not just to wider audiences. I'm reading Marion's question now. When they're not used to digital media or the internet at large. You know, in situations like that doesn't matter. The audience testing was done around a large shady tree in the middle of villages. They they played the radio show, and they got super super specific feedback around. We don't like the female host. Your music is too loud. They go on and on. There's too many opinions. We need more facts. And we'd like more. We'd like different size, format sizes. So make a short show but also to do an option for a evening, longer, shorter evening. As far as I'm concerned, that is a news product done by a news organization. With an audience I don't I don't care if it's digital or not. I it doesn't make a difference whether the literacy is up or down technically is that something we can we can we can address with AI? I'd love to think so. And I love that Maryam answer question where you're saying your research revolves around media literacy, including AI and crisis literacy. So I'm very excited. You know that we have someone like you doing this work and I'd love to know more

than me how back to the room and hopefully there's better audio otherwise, I will step in as interpreter again, Philippe, you have a question?

Yeah.

Thank you. And as you mentioned, Acebo SEO will be active in the coming years with the integration of AI in the search engines. And generative AI also relies on news articles to generate responses should the company's developing AI interface be required to compensate for the company they are using?

Yeah, that's a that's an interesting one. And that, that goes back to the code in you know, the the recent news code in Canada and in Australia. I am I my answer for you. My short answer is I don't know. I think as we know, we're here we are. We're still trying to create policy around technologies over the years. We're still catching up. With social media, which been around for arguably, you know, two decades. You know, we haven't figured out how to tax it. How do you know where its boundaries are, what its geopolitical footprint is whether it has any, these things are bigger than countries and economies? We were getting there. I think that that the simple answer there is that to in order to generate something you have to feed the machine, and I think here we are. We're in the business of feeding the machine that we also use and I think at some point, we're going to have to figure out how this works together. That's that's that's all I have. I wish I knew more.

Sorry, has a question is when they go ahead.

I found no way.

Ask you, as you said, many media organizations are developing AI policies. I'm curious which ones are in your opinion, the best ones that we can learn from so far? And also is there a need that media organizations come together,

develop

work common practices of these policies?

I love this. So you know, one of the things that we do a lot of and thank you so much for that. So yes, a resounding yes. Is there need to come together and create a policy, not just now we've seen you know, just here we've seen two policies at work and I that those are the only ones that I know about. But we've seen the Reuters policy. And we've seen the scroll down in policy, which they're talking about, should we come together and draft a policy not just media organization by media organization, but perhaps is there a way to talk about this and draft this as a as something that will singularly help us all a sort of AP style guide as you would and one of the things that I asked saluto Raghu have have screwed up in was and I hope you know she doesn't mind is that I was so excited by the fact that they're drafting this policy. I said, you know, I think the community needs to learn from this number one. Number two, we have you know, we we do a media startup festival every year in Chiang Mai in November. And could you come and teach us how to write use of AI policy? And she was like, Wait, really? And I said, Yeah, we desperately need this. And I think that that will be one of the most well attended sessions at our festival. I would love ways to be able to collaborate on this perhaps using AI to draft the use of AI policy. So hey, send me ideas. I'll make it happen.

Can I be devil's advocate to reach out and ask Yes, please. As you said with Reuters, it's sort of the the grown up end of the response and the policy and the take on AI can we expect an entire community that is going to be using this to be the most responsible about their own accountability? To the use of this? I'm just asking. You know,

I think that I think that's inevitable. I think that, you know, unofficially, we're going to have the sub Reddits we're going to have the splinter groups. We're going to have the Watchdogs. I think that's that's going to happen, right? But I think that much like you know, we make these three assumptions right in the journalistic world. We must use generative AI before it uses us there is this thing that we we've got to do it? I'm not sure entirely sure why. Abdulaziz Khattak has a comment on the chat here, where he says as a journalist in a small media organization where manpower is an issue but content cannot be compromised on I find AI as as God sent thing, it could help us reduce our stress levels and do our jobs better and faster. Well said

the industry your stress levels are not going to change.

I completely agree. But the second is, of course, that, again, that the only thing that generative AI can generate is content. It's it's not true. So I love this idea. of you know, we've we how do we talk about you know, policing, how do we talk about the mean, where do we get that central benchmarking to happen? Is there an you know, an opportunity for an independent body to actually do this. I also liked you know, Altman's take on it. You know, when when he spoke about this, he was kind of circumspect. They were warnings. He said that he welcomed regulation of the emerging technology. And I think as as you know, as one of the people that made made this entire conversation this possible across the world, and that is an awesome thing to be saying. He also welcome regulation of of the technology, which is spectacular. You don't often see a guy who has who has been, you know, who has brought startups out of nowhere, and, you know, proliferated them around the world, saying that very often. So I think, and he also said, you know, he drew parallels with Photoshop. I'm not sure that that was necessary. And I don't know if it's apples to apples, but he said when Photoshop came on the scene about what 30 years ago, people were fooled by Photoshop images. We we get hysterical to overuse that word, but we developed an understanding around that became part of our common plumbing and our common vocabulary. And he said this will be like that. But on steroids. Everything is there on

steroids, is it not? People and I'm conscious of time and I can see more questions. So let me squeeze in as many as possible. Hideki fellow who is visiting from Japan has a question for you it he go ahead.

My question. Know, AIB used to make giant more buses or to make it seem more impossible to arrive on us.

So, are you asking whether AI can be used for impartiality?

externalism more impartial? I think it's hidden keys.

So So I think again, it's its nature, nurture and tech, right? I wonder right. So I wonder if there this is, this is also a principle is it not? It's a human principle. Should an editorial policy in the newsroom must decide what impartiality means, and it must have buy in from the newsroom. It is part of a style guide. But more than style guide, it's also part of your journalistic ethics. And how you how you spread that around your practice of journalism. So I also think that that may or may not have anything to do with AI. Can AI be deployed in order to crawl through a story and figure out how partial or impartial it is? Perhaps, I think it's entirely possible. I bet you there's a plugin out there right now. Likewise, can you deploy AI to crawl through a story and say, this is completely skewed towards towards men? We you don't have enough women sources, your conference program, your fancy conference program has, you know, way too many people talking about AI, chill out a bit. And maybe you should talk about audiences and stuff like that. So sure. Can the technology be used to, to and deployed to to make this a more partial look for sure. Can I do a quick Mitali if I may, a quick shout out to you know, I had a couple of participants who may or may not be online. But I have a newsletter update the to two newsletter signups yesterday because of the session. And I had a guy from spray labs in Berlin. Hello, Peter, if you're there, and he I asked him what he did. He said they're building the A AI powered publishing platform. And this is interesting. It's a SaaS solution for medium sized news and magazine publishers. And I'm encouraged when I hear him, you know, read the rest of his email to me, which is there's an editorial CMS. There's, you know, there's a experience thing to build custom apps and stuff and websites. But what I loved was there's also a bunch of tools geared towards newsroom efficiency and audience development. Another sign up from the Danish School of media and journalism, hello, Petter. Who also may be on this. They're working on research projects that are around how AI can be introduced in the education of new journalists. Now, that's amazing. That's way you can feed this around impartiality. You feed the machine with impartiality. You feed it with an education around what an ethical editorial policy looks like, and how to serve readers. I think there's huge hole but that

which brings me to my final question, Rashad, which is how do people reach out to the splice gang? How do people collaborate with the splice gang? Don't go by the shots Twitter image, I can assure you that's not an accurate way to find Rishard at a conference or anywhere else. So what's a good way to hear more about splice and collaborate with you guys?

Well, that's very kind. Thank you. You know, we go to splice media.com I'm happy to pop that into the chat or if that helps. Come to splice better subscribe to our newsletters, and meet our community. Come to Planet splice.com. We have a little slack community of global media wisdom and kindness. So come on in. There's no AI at work there.

Sorry. Thank you very much. Rashard. It's been really excellent chatting with you. My apologies. I know there was a stream of questions. I tried to sort of make space for as much as I could. One hour was clearly short time and not enough. So find a shot where he pointed you to, and you can lay out all your questions over there. It's been really great chatting with you. Thanks for making time for this.

Thank you so much. Thank you, everybody. This was so much fun.

Likewise, and for our audiences. We'll see you again next week.