S1#4 Die transformative Wirkung von KI (zu Gast Charlie Beckett/ KI-Experte UK)

00:00:00: Ich denke, das ist ein wichtiges Moment,

00:00:02: nicht nur wegen des Geschäftsmodells,

00:00:05: sondern auch generell, was für Journalismus ist.

00:00:08: Ich denke, es ist über zwei Dinge.

00:00:10: Es ist um diese Technik zu benutzen,

00:00:12: um effizienter, effektiv, mehr akkulär zu sein.

00:00:15: Aber auch um zu denken,

00:00:17: vielleicht müssen wir mehr um die Menschenjournalisten machen.

00:00:20: Echte Insights aus der Medienbranche.

00:00:23: Hier sind deine Innovation-Manage.

00:00:25: Direkt aus dem Mediale Bayern.

00:00:28: Hey, schön, dass du dir Zeit nimmst und der Innovationraum gibst.

00:00:31: Mein Name ist Sabrina Harper.

00:00:33: Und heute habe ich, wie du schon im Intro gehört hast,

00:00:36: ein englischen Talk-Gast bei mir.

00:00:38: Ich habe ChatGPT gefragt, wer er ist.

00:00:41: Und das ist die Antwort.

00:00:43: Er ist Gründungsdirektor von POLLIS,

00:00:45: einem Think Tank für Forschung und Debatte

00:00:48: über internationalen Journalismus und Gesellschaft

00:00:51: in der Abteilung für Medien und Kommunikation

00:00:54: der London School of Economics.

00:00:56: Und forscht unter anderem zu dem Thema "Künstliche Intelligenz".

00:01:00: Ich mache es ein bisschen kürzer.

00:01:02: Die Rede ist von Professor Charlie Beckett.

00:01:04: Und bei Media Lab Innovation Festival

00:01:06: hat er ein Panel zum Thema KI gehalten.

00:01:09: Danach haben wir uns getroffen

00:01:11: und ich habe ihm dann quasi die Gegenfrage gestellt.

00:01:14: Wer ist ChatGPT?

00:01:16: Hey Charlie, so great to have you on the show.

00:01:21: It's lovely to be here, really is.

00:01:23: You do a lot of research about AI and also about media.

00:01:27: Here in Germany it seemed like the media landscape

00:01:30: was quite surprised about ChatGPT becoming so important, so fast.

00:01:36: Have you been caught by surprise about that too?

00:01:38: I'm very excited about ChatGPT.

00:01:42: Just because my business is media change.

00:01:44: So anything that brings the possibility of change

00:01:47: is always interesting for me.

00:01:49: And I think coming to Germany,

00:01:51: I mean here at the innovation lab, the media lab,

00:01:55: it's just fantastic because you've got a bunch of people

00:01:58: who are really interested in the topic intellectually, if you like.

00:02:02: But they're also trying to do something.

00:02:04: And so we're at the heart of this storm,

00:02:08: if you like, around genitive AI.

00:02:10: We are really in the midst of a huge experiment.

00:02:14: So of course there's a lot of hype and a lot of uncertainty.

00:02:17: And the best thing you can do when that happens

00:02:19: is get together with other people who care about you, know about it,

00:02:23: and share your experiences.

00:02:26: Yeah, but what to do now?

00:02:28: From your perspective,

00:02:29: how should the media landscape react on this?

00:02:32: Or even better, act in a transformative environment like this?

00:02:37: It was a surprise, at least in a sort of marketing way,

00:02:41: that the genius of the ChatGPT is that it's so easy to use.

00:02:48: It's just one line, you put in a prompt, and something happens.

00:02:52: And it is quite kind of magical at a superficial level.

00:02:57: This is so much more accessible than other forms of artificial intelligence.

00:03:02: Machine learning, natural language processing, data analysis, all that stuff.

00:03:07: It was quite actually quite laborious and quite complicated.

00:03:11: And it was in this kind of so-called black box.

00:03:13: It was hard for, if you like, ordinary people or ordinary journalists

00:03:17: who weren't tech clever to play with it or to understand it.

00:03:22: It was left to the tech department.

00:03:25: And I've been working on that for four years,

00:03:27: and I can see how that's had a big impact already on many news organisations.

00:03:32: So when this came along, it was a surprise, because they did rush it out.

00:03:37: It emerged because there's this competition,

00:03:41: there's a race going on between Microsoft and Google in particular.

00:03:45: But the other big tech companies too.

00:03:47: So we're getting to play with this technology before it's properly evolved.

00:03:52: Is it about time to concentrate on new USPs to define journalism?

00:03:59: Yeah, I think it's, we've been through these different waves of change,

00:04:03: you know, going online and then on come social media.

00:04:06: And now these AI-type technologies.

00:04:10: And what's interesting about it is not just that the technologies change

00:04:13: the way you do your journalism.

00:04:15: It also makes you think about what's the point of journalism in the first place.

00:04:18: You know, is the point of journalism to keep big journalism brands profitable?

00:04:25: Or is it perhaps to serve the public and give them information

00:04:31: and to hold power to account, all those other nice public service things

00:04:35: that journalism can do.

00:04:37: And I think we're at another point here where we're thinking,

00:04:40: well, if the AI can do the more routine journalism,

00:04:45: what does that leave us with?

00:04:46: What is it that's special about a news organization?

00:04:51: What can we do as journalists that the algorithms can't?

00:04:56: And so I think this is a really important moment.

00:04:58: We're not just because of the business model,

00:05:02: but just generally, you know, what's journalism for?

00:05:05: And I think it's about, well, it's about two things.

00:05:07: It's about using this tech to be more efficient, more effective,

00:05:11: more accurate, to reach more people in a way that, you know,

00:05:15: gives them the content they want in the right format

00:05:18: and at the right time.

00:05:19: But also it's about thinking, well,

00:05:22: perhaps we ought to be doing more human journalism.

00:05:24: We ought to be doing more reporting, more witnessing,

00:05:27: more mission driven journalism, perhaps.

00:05:30: Journalism that relates much more to human interest, if you like,

00:05:34: but it's a bit more emotional, a bit more engaged.

00:05:38: So I think it's a fascinating time for journalism.

00:05:41: And like today, this event, you've seen some brilliant startups,

00:05:46: which are using the tech to try and do something a bit different.

00:05:51: You did the panel about AI earlier today,

00:05:56: and you said we need to think about something else,

00:05:59: because AI is just a technical term.

00:06:02: What did you exactly mean?

00:06:04: What I mean is, I use the term "artificial intelligence"

00:06:07: because there isn't a better word for it.

00:06:09: And it's really an umbrella term that covers a whole series of technologies

00:06:13: based around this idea that the algorithm can be trained

00:06:18: to predict the next bit of language.

00:06:24: And in some ways, it's very routine.

00:06:26: You know, those big machine learning programs

00:06:29: that scrape the social media, looking for stories,

00:06:32: they're actually quite routine in a way.

00:06:35: With generative AI, it's not like some miracle.

00:06:38: We haven't suddenly got sentient AI.

00:06:41: We haven't suddenly got something that's genuinely intelligent.

00:06:44: But for a number of reasons,

00:06:46: development and software and hardware and servers,

00:06:48: they are much, much, much more effective.

00:06:52: And they are able, as the name says,

00:06:54: to actually generate content incredibly quickly

00:06:58: based on these huge, large language models.

00:07:01: They are definitely not perfect

00:07:02: because they don't know anything.

00:07:04: They are not truth machines.

00:07:06: They are language machines.

00:07:08: They are just like a sort of very enthusiastic

00:07:10: intern who's trying their best to give you what you want.

00:07:14: But they don't actually necessarily know

00:07:18: in a human sense what they're talking about.

00:07:21: Someone said to me that in the future

00:07:23: we will have some kind of magic buttons.

00:07:27: They seem like virtual, or maybe,

00:07:30: even in the real world, buttons.

00:07:32: We push them and the magic happens.

00:07:35: Magic made by AI.

00:07:37: You are really down to earth.

00:07:39: So what do you think about those magic buttons?

00:07:41: Is it fancy stuff?

00:07:43: As I said, the point of chat, GBT,

00:07:45: was that it was so easy to use.

00:07:46: You don't want to know all the tech details

00:07:50: if you're using this thing, even if you're a journalist.

00:07:53: You just want to be able to use it and trust it

00:07:56: and find it efficient and helpful.

00:07:59: It's the same with driving a car.

00:08:01: You have to learn how to drive a car.

00:08:03: You don't have to learn how to rebuild the engine.

00:08:05: You don't even have to know particularly how the engine works.

00:08:09: You just have to know where you're driving to

00:08:11: and to know that you're going to be safe in your car.

00:08:14: And I think it's similar to this,

00:08:16: that there's this explosion at the moment

00:08:18: of people coming up with special apps

00:08:20: that are using chat, GBT,

00:08:23: they're using generative AI.

00:08:25: Some of them aren't, to be honest.

00:08:28: And many of them are very simple use cases

00:08:31: that will probably disappear, like autumn leaves.

00:08:34: But there is so much happening, so much creativity.

00:08:39: And there's such a potential there at least,

00:08:43: that okay, you may not rush to build your news from around tomorrow.

00:08:51: But I've got no doubt

00:08:53: out personally that it's going to shape the way the information flows in the future.

00:08:59: And one of the key things is to do what journalism has always done, which is to make it easier

00:09:04: for people to get information.

00:09:06: So getting back to the roots.

00:09:08: Getting back to roots, getting back to basics.

00:09:11: If technology is super complicated to use, people won't use it.

00:09:15: It's as simple as that.

00:09:16: So something that a tech expert finds deeply fascinating and intriguing as a use case.

00:09:23: Is useless unless it's easily used.

00:09:27: And that's whether it's used by journalists or whether it's used by a member of the public.

00:09:32: So the real goal dust, if you like, is making this stuff reliable, trustworthy and easy to use.

00:09:42: Thank God there's hype.

00:09:43: That's what you said.

00:09:45: That's your sentence.

00:09:46: You as an expert, what do you think?

00:09:49: What is the next big thing to influence the media industry in the near future?

00:09:54: Well, I think the whole genitive AI thing is going to be absolutely fascinating.

00:09:59: And I don't think it's going to slow down much.

00:10:01: I mean, I like the hype in the sense that it's like an alarm bell going off.

00:10:06: And it's provoked this fantastic debate, which is much broader than perhaps even the debate

00:10:11: we had over social media.

00:10:13: And I think that's really interesting.

00:10:15: It's got politicians engaged.

00:10:18: It's got regulators engaged.

00:10:20: It's got university professors engaged.

00:10:23: People paying attention to it.

00:10:25: Editors are paying attention to it.

00:10:28: And I think that is good because I think there are kind of short term effects that are going to happen.

00:10:35: You know, that people are going to use new tools, perhaps we'll come up with new formats

00:10:38: and new kind of ways of servicing the public.

00:10:41: But in the longer term, there is also the bigger issues around, for example,

00:10:47: the quality of information in our societies, but also from an industry point of view,

00:10:52: what is going to be the structure of a newsroom or the structure of the news industry

00:10:58: in say five to 10 years time?

00:11:00: Because we saw how people predicted that social media or the internet would kill off

00:11:06: all mainstream media.

00:11:07: Well, it hasn't.

00:11:08: If you look around Germany, there's some very familiar, very old brands

00:11:12: were still going strong, in fact, stronger than ever, more red than ever before,

00:11:18: as still very influential.

00:11:21: So I'm not suggesting that this is going to kill off mainstream media.

00:11:25: But I think it's changed so much in the last 10, 15 years.

00:11:30: And I suspect that this is another phase of very intense change

00:11:35: because this is opening up the creation of content so, so widely.

00:11:43: And it's going to create, on the more optimistic side,

00:11:46: is an incredible opportunity for creativity.

00:11:50: The machine kind of does a lot on its own, but somebody has to train the algorithm.

00:11:56: Somebody has to check the data set it's based on.

00:12:00: Somebody has to put the prompt in, train, instruct the algorithm,

00:12:07: what you wanted to do.

00:12:08: And somebody has to edit it afterwards as well to make sure that it's what you want.

00:12:15: So the idea that suddenly journalism disappears is a nonsense.

00:12:20: But it does represent a huge shift in power to other people,

00:12:27: just like the internet did, just like digital did,

00:12:30: just like mobile telephony and smartphones,

00:12:33: broaden people's ability.

00:12:36: Anybody who is savvy about using the tech

00:12:39: could at least, in theory, be part of the information ecosystem.

00:12:43: So journalists once again have to think, "What's special about what we do?"

00:12:48: Somehow it reminds me about the shift of skills

00:12:51: that journalists had to learn about 10 or 15 years ago.

00:12:56: Before that, there were video journalists, newspaper journalists or radio journalists.

00:13:01: They all had a specific skill set.

00:13:04: Then when the age of multimedia was coming up,

00:13:09: as a journalist, you needed to be able to do writing,

00:13:13: but also video editing or audio pieces.

00:13:17: Now when I'm thinking about AI, maybe there will be no one man show or one woman show.

00:13:26: But it's kind of similar.

00:13:28: We will need a different skill set.

00:13:30: Maybe that will be a team with developers, tech people and journalists altogether,

00:13:36: or something like that.

00:13:37: What should you think about that stuff?

00:13:38: I mean, one of the things that's come out of our journalism AI project over the last few years

00:13:43: is we've been working with teams of journalists and tech people

00:13:47: from different news organisations collaborating on innovation.

00:13:51: And it's been fascinating.

00:13:52: You can see how the skill set has obviously had to change.

00:13:56: You still have the traditional skills,

00:13:58: perhaps the things that made you into a journalist in the first place,

00:14:02: finding things out, communicating, understanding stuff.

00:14:07: But there's also, obviously, this technology

00:14:12: brings new opportunities to do things in different ways,

00:14:15: to research in different ways and to format content in different ways.

00:14:19: So it's really interesting watching people who often come from a more tech background

00:14:24: getting dragged into journalism because it's really interesting.

00:14:27: Journalism is quite an interesting career to be in.

00:14:30: And likewise, the other way, journalists who think,

00:14:32: "Hey, this stuff is a really good way for me to find out things or to tell a story.

00:14:38: I'm going to find out how to use it."

00:14:40: And that's been the story of the last 20, 30 years.

00:14:42: In some ways, it's really exciting.

00:14:44: When I was a journalist, you were a newspaper journalist or a TV journalist or a radio journalist.

00:14:49: And that was it.

00:14:50: Now, you're expected to have all these many, many different skills.

00:14:54: And that's deeply challenging, partly because you have to work much harder,

00:14:57: as well as train yourself much harder and keep yourself up to date.

00:15:01: And all that time, you're still getting paid really badly.

00:15:04: And everyone hates you.

00:15:07: So I can see there's a real pressure on journalists today.

00:15:10: Many of them will look at this and think, "Oh, gosh, here we go again."

00:15:13: You just want to think, "I've got to cross all this stuff. It's going to change again."

00:15:17: And I think it's a real pressure point.

00:15:19: The stress levels for journalists out there are really high.

00:15:23: And management needs to think hard about how it incorporates some of these technologies,

00:15:28: how it makes sure that its staff are both trained, but also managed well

00:15:34: in a way that doesn't overestimate the ability, the magic of this.

00:15:38: AI is not magic.

00:15:39: It doesn't do things completely on its own.

00:15:42: You don't want it to.

00:15:44: So there's going to be a lot of responsibility for journalists

00:15:47: and a lot of pressure on them, probably, to work harder.

00:15:51: I hope the technology can save them all at a time.

00:15:54: It can mean they don't have to do some of those boring routine things.

00:15:58: And it may mean they're able to do very exciting new things

00:16:01: to help make their work feel more meaningful and effective.

00:16:05: But yeah, it's going to be tough.

00:16:09: But then, gosh, I don't know if there's journalists out there

00:16:12: who isn't having quite a tough time already.

00:16:15: That's true.

00:16:16: Thank you for sharing your insights with us.

00:16:20: I'm going out of the conversation.

00:16:22: And what's left of me is, among other things,

00:16:24: that AI influences the power of power.

00:16:26: Like back then, for example, the phone, the internet

00:16:29: or social media, as Charlie Beckett said.

00:16:32: What does your editorial work look like?

00:16:34: Who has the data?

00:16:35: Who has the technology?

00:16:36: And how do you do that with AI?

00:16:39: If you're interested in technology and media,

00:16:42: I can recommend you the Tech Lab.

00:16:45: Developers and media people will come together

00:16:48: and develop together what the media industry needs.

00:16:51: If you want to know more, just go to media-lab.de

00:16:55: You'll find the Tech Lab.

00:16:57: And we'll hear you in the next episode

00:17:00: with the content creator Tuget Jong.

00:17:03: He specialized in trans people and diversity in the media.

00:17:07: Subscribe to the podcast,

00:17:09: and you won't miss the next episode.

00:17:11: I'm Sabrina Harper.

00:17:13: And I wish you a fantastic day.

00:17:16: These were your Innovation Minutes.

00:17:20: Direct from the Media Lab of Bayern.

00:17:22: If you want more,

00:17:24: visit us at media-lab.de

00:17:26: and start your own project.

00:17:29: Last but not least.

00:17:30: This podcast is possible

00:17:32: because we have great supporters and supporters.

00:17:35: Special thanks to the State Chancellor of Bayern,

00:17:38: the Bavarian National Center for New Media,

00:17:40: the BLM and of course my colleagues in the media.

00:17:45: (upbeat music)