Skip to content
Pandora's Box: Taming AI episode artwork

Jun 2, 2023 39 min

Pandora's Box: Taming AI

from Fourth Estate Hosted by 2SER

Our first show of 2023 focused on AI. Four months later, we revisited this topic with the same guests, Dr. Sacha Molitorisz and Charlie Lewis. Since then, significant developments have occurred in the AI space, and our discussion this time addressed the challenges we now encounter with this emerging technology. Produce and hosted by Anthony Dockrill

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

Pandora's Box: Taming AI

0:00 39:51
of matches

Transcript Auto-generated, may contain errors

Hello and welcome to For For State, a show about journalism. We're coming to you from Tiresian Sydney on the Gadicle Lands of the Euro Nation, right across Australia on the community radio network, and directly through the Vice-Across-the-Glide podcast. I'm Anthony Dockwell. Our first show for 2023 was a look at AI.

Back in March, the abilities of GPT-3 were reverberating around the world, and during that time we asked what this meant for the future of journalism and the news. Fast forward four months later and a lot has happened since. In short, GPT-4 and Google's Palm have shown not only are the abilities of AI moving ahead, but more concerningly that they are accelerating. Some estimates now have AI equal or better than the average human at all language tasks.

That's all tasks. It's probably fair to say our emotional response to AI over the last few months has moved from alarm to excitement to despondency and then back to alarm. Last week, in the US, the stock market fell reportedly due to an AI-altered photo of an explosion outside the pentagon. Are we ready for a flow-to-fake news?

And what happens when the fakes are as real and believable as the actual news? To discuss this and more, we are lucky to have Dr. Sacha Melatoris, who's a single lecturer at UTS, and in a previous life as a feature writer for the Sydney Morning Herald. Charlie Lewis is the tips and merma editor at Quarky.

He also presents SpinCycle on Triple R in Melbourne. Dr. Sacha Melatoris and Charlie Lewis, welcome back to Forf Estate. Hey, I'm thinking that's empty.

One of the cliches of radio is ending a show saying, we'll have to get you both back to talk about this again. Well, for once it's actually happening. Now, our first show of the year, I spoke with both of you about AI. Now, a light has happened in the last four months.

I started the first show asking you both if we were on the cusp of a revolution. And at the time, both of you were a little bit unsure, but noted it was clearly something that had that potential. Four months on, is this the start of a revolution? Oh, gee, I don't know if I feel any different to be honest from four months ago.

Clearly, it has the potential to be transformative in a whole lot of ways. There are many potential benefits and there are serious, I reckon, enormous risks as well. You know, when we had this week, I'm sure we'll get to this later, but this week, some of the AI leaders coming out and saying there is a possibility of a risk of extinction coming out of all this technology that we're developing and the speed with which it's developing. So yeah, there are serious risks and I have some very specific and strong views about how we need to respond to those risks.

But I'll let Charlie have his response. Yeah, no, and I think certainly the pace in a way we should have known that the mountain months following our chat would have such a strong and quick set of advances. And I think also part of the irony of this is the reason that journalists are writing about it so much is that it's more and more likely that part of this is going to be costing them their jobs. So I think that's part of the reason why the existential sense of everything is being fed into the coverage.

I see, obviously, this morning that the government has announced plans to attempt to regulate this in some ways. I think we will agree is necessary on one level or another. It's interesting. I think one of the points that I thought was interesting in relation to that letter that's actually referenced before the fact that it should be thought of as a potential extinction category level event is that there's a bit of a school thought that I've seen from some experts in this area that feel that's almost a little bit of a slight event is that by calling attention to the fact that or by calling to mind a far off and at the moment hypothetical world in which we are we are terminator-style slaves to the machines does actually stop people that take the headline and people talk less and less about the fact that AI in other ways has been woven into our day to day life quite a lot, particularly in the area of say surveillance and a lot of the issues in terms of the confirmed bias of the data sets and things like that are already well well and trained.

So I was happy to see in the report that the government's not that they was talking about one of the issues being that there are these in-built biases in built into other ways that the AI kind of works and it leads to see what is going to be done to regulate I suppose. Well, there's a lot to unpack this. Let's jump into the call that's gone out from a number of leading figures in the AI field to basically put a pause on the development of these models. It strikes me that this call to pause this development, it's already too late.

These things are already out of the box. So how do you feel about these calls for pause and do you actually feel that it's actually sincere? Yeah, look, I just see that there are enormous potential risks. So how do we respond to that?

Do we say, look, people are going to innovate, companies are going to innovate and it's going to happen anyway? I frankly, I don't think that's good enough. When people have come out now and said, you know what, we should have had more guardrails around social media. Some of the harms that have been caused and still continue to be caused are really serious and ongoing, you know, from misinformation to teen mental health to what have you.

I'm just a big believer in, yeah, okay, it might be hard to work out how to regulate this. It might even be impossible to enforce regulation perfectly or even really well, but we still got to try. And I think, you know, the risks are great that there are certain basic regulatory moves that we should make and need to make, you know, and Charlie mentioned that the Australian government's made an announcement in this space over the last day. And I think that's really good.

We need to get into having some guardrails as AI is it as the AI developers of themselves admitting? Yes. Let's talk about why we need these guardrails. I mean, one of the things that has been alarming me more and more when I look into AI, I've been following a few blogs.

And one of the things that keeps popping up time and time again, in the case of JPT-4, is that it has abilities outside of what it was designed to do. So for instance, JPT-4 can write music. It was never trained on music. As of last week, JPT-4 has been made to play Minecraft autonomously.

This is a language model interacting in a 3D universe. So it does feel like it's time to also drop this idea that these large language models are just predictive text. And to some extent, we can just ignore them. It also seems that to me that the genie is out of the bottle.

JPT already has 100 million users. It's reached that level in a timeframe that took Facebook two years. So it does seem that we are dealing with something completely new and something that is outside of what we even were told what it was four months ago. We were told that this was just basically a thing that arranged words by statistics.

Do you both feel that we're now in this space where we're talking about a technology that is actually something new and so new that we probably should be pausing and taking a moment to actually know what we're actually dealing with. Oh, absolutely. I don't think there's any doubt that that is the case. And of course, with issues such as AI, you get to slightly head spinning territory when you start to get the section of once you create something that is as intelligent or more intelligent than human, what is that then capable of creating this more intelligent than it is?

And that's when you get into the kind of doomsday scenarios that we've been discussing. I mean, and there's so much even just sort of pure practical non-philipical level, there is now so much research planning up in the US and elsewhere that sort of indicates how many jobs can actually be sort of wiped out and destroyed by technology that saw an employee of the public consultancy actually quite breezily and happily announcing that they believe that by the end of this decade, large language models will be able to write final drafts better than a professional writer would and create code better than a full-time developer would. And that starts to get a bit mind-boggling when you think about it in those terms. I've heard that there's an academic incorporated lot of chat JPT into his course and had students sort of reflect on it and have students generate work using that technology.

And they had some really interesting reflections. And one of the students said, look, I'm not really that worried yet that the AI is going to match human intelligence. I'm all worried that we're going to regress to AI's derivative, generative type of approach, you know, because we have to keep reminding ourselves that what is going on with chat JPT and all the other generative AI is that it just keeps scraping the content that's already out there and then mashing it up in a new form. There is nothing fundamentally new.

It's just a new kind of blend of what is already out there. So in some ways, it's not such a step forward at this point. It's kind of, it's locking us into a particular point in time when the internet is at the point that it's at that point. And that's what's being scraped in mind.

So it's at this point, hopefully, doomsday scenarios are still a long way off with it here at all. But yeah, this is the point now when it is getting pretty critical and we need to take these. Have these thoughts about what we're going to do? And so the specific measures, two of them that I argue for, and this is something I've written a short piece in tomorrow's, that's Friday, June to send up a media transition newsletter, which you can access through the website, cmt.uts.edu.au.

This is the moment where we regulate in a couple of specific ways. One is, I think we just need to switch the whole mindset of the law from a carby-art, empty-door approach to a carby-art vendor-door approach, right? Instead of buyable ware, we're thinking more sellable ware. We just need to put the responsibility fundamentally on the companies who are letting loose this technology into the world.

So you can't be a company such as OpenAI and say, look, we've got this massive technology. We're going to release it on the world and let's see what happens. What, fine. But then you wear the consequences.

If this misinformation leads to violence, it leads to horrible violence on a large scale. There's defamation. We saw a story about a local politician down in Victoria who found out there was the family content that JBT was generating about the fact that he had been imprisoned. He hadn't been imprisoned.

This sort of stuff is really harmful, right? And the company needs to wear it. So that's my first suggestion. We shift to a seller, the ware, and put the owners and responsibility on the companies.

The second thing I think is that the law doesn't need to get all of my new CI in place. Rather, let's have general protections. Already the Australian consumer law protects against misleading and deceptive conduct. That's something that's actually worked really well in protecting privacy in a few instances.

Google was hit with a $60 million fine for being misleading location data collection and not telling users exactly what was going on. So I think these sorts of general provisions against misleading and deceptive conduct, against coercion for fairness, which is something that is being considered by the ACCC and the government for robust consent being mandatory. For a certain degree of transparency, I think we make these general provisions that the courts can then interpret and they can potentially be really powerful. That's some very interesting points there.

And look, it does feed into a point that was made by Boxes Kesley Piper who pointed out that AI is moving so quickly. It's not that humans won't be able to adapt. It's at the rate that it's moving is going to stop humans from adapting. We saw that with the internet.

We saw that with social media where things change very quickly in a space of a few years and our legislative framework was unable or unwilling to keep up. That model that you were just talking about, it does completely turn the web idea on its head because the companies aren't responsible. We know that social media companies aren't responsible for the content that they actually publish. They'll deny they published it.

What do you think of the chances of? It does take a complete mindset change here and it doesn't sound like one that we do need. You're totally right. You're totally right.

It completes which in mindset. In the US, most famously, we have Section 230 of the Communications Decency Act, which is about 25 years old and basically gives digital platforms immunity or what the content that appears on their site. Now, even Facebook has admitted and other digital platforms have admitted we need to revisit that and it goes too far. I think we have a moment where AI companies, digital platforms more widely recognize that regulation is needed.

Mark Zuckerberg has come out since January, January 2018 and said, but we do need regulations about privacy, about misinformation, about harmful content and so on. There is a moment where we recognize there are major risks where even the companies are saying, we do need regulation and sure that part of that might be disingenuous or there might be ulterior motives that ultimately might help their bottom line in some way. I think this is a time when we need to step in and do something. Look, you both mentioned the announcement that came out of the federal government about a discussion paper and with that, there's a call for submissions and maybe that we're moving towards having a white paper on AI.

Look, there may be some guardrails. There's even a call for a dedicated minister for the digital economy. All these things to me sound quite logical. But Australia doing those things in isolation is not really going to change anything, is it?

Europe may move on this, but fundamentally, all this comes back to America. It doesn't seem to be any appetite to regulate and to restrict the AI industry. Do you see a different path forward? Well, I mean, I suppose it does come down to a little bit of a moral argument about kind of culpability and controlling what you can.

I mean, that's been the longest running argument in this country against any kind of action on climate change from government, as always. But we emit so little if China and India don't do it, then who cares? Nothing we can make a difference. But it's sort of like, well, ultimately, you can't let that be the guiding, you can't let that be ideal to make you know whether something ought to be regulated or not.

And also, it's funny we were talking before about this is an evil chance to regulate something that needs regulation at a decent time. It sort of reminded me a little bit, and I think if we could critique the need or a little bit of the gig economy era, when every new app that came along that could sort of you know, task by whatever, you would be reporting this real like Gee whiz, I think, and this is so interesting, you know, what an innovation way by the media. And the fact that this was eroding, well, over 100 years worth of employment or in this country didn't seem to get noticed until about five years after all these things had been vetted in. So I think, yeah, I think we're going to actually reach a question.

I think ultimately, I think you're correct. I think the real issues are in bigger markets, but that shouldn't still be often from trying to trying to intervene. Yeah, I couldn't agree more with you, Charlie. I think we just have to take responsibility for what we can in Australia and try to get the best regulatory framework.

And look, I'm not someone who sort of traditionally argues for more laws, more regulation. So as I said before, I think, you know, some of these general principles can be really powerful, right? We need to kind of rethink this from a first principle perspective. And you know, against misleading the septic conduct, which is already enshrined in law for fairness, for a degree of transparency, these sort of things what can make a difference and the principles can then be applied as the technology changes and evolves.

And looking around the world, we won't be alone. You know, the AI Act is proposed in Europe, which you mentioned earlier, you know, it will have some significant powers and it follows on from other innovations in Europe, the ADPR very famously, which protects privacy. And then last year, the Digital Services Act came into place as well. And it has these quite sweet powers with regards to misinformation and letting users have more control over how algorithms affect them, which is really interesting.

So Europe is definitely moving in this direction. We will follow suits. We're currently reviewing our privacy laws as well, which is something that's connected. And the US, I mean, even the US, yes, they won't move as far as Europe.

We know that. But you know, Biden and Harris, their administration came out with an announcement a couple of weeks ago now. It's taking new steps to advance responsible artificial intelligence research, development and deployments. You know, so hopefully the US will move in this direction as well.

And look, there is history of this actually happening. Traditional media is reasonably tightly regulated. And, you know, it's like governments around the world have shown that they can do this. But it's fair to say with the internet and social media that it was abject failure across the board.

Do you both feel that the lessons of the last 20 years have been learned, at least in Australia and Europe, Sesha? Look, that's a good question. And of course, what we're really interested in, I guess, in this conversation is how this all ties into journalism, right? So the regulation of journalism is a really thorny, tricky issue.

So in the US, it's largely unregulated, you know, that we don't, in the US, there aren't really the equivalent of Australian press council law, you know, Australian communications and media authority. In Australia, meanwhile, we do have quite a lot of regulation, but it's really messy. And it's really inconsistent. And, you know, I've published a few things with Derek Wilding about that and about how reform is needed there.

We need a more streamlined co-hero system of news oversight. Yeah. And that should incorporate how journalism works with AI and users AI and what the kind of parameters there should be. So like, we need a lot of, we're in a lot of work in that space, too.

And that's really relevant to this whole larger conversation. Let's move on and we'll come back to some of these larger issues later in discussion. But let's move on to disinformation. A little bit more over, we could go a reportedly AI-generated photo of an explosion outside the Pentagon, sent the stock market falling, was retweeted all over the place.

You know, fake news and disinformation is not anything new. But with the promises of AI, it does seem we're on the cusp of an explosion of fake news and disinformation. And especially here's the Dorothy Dixa. Are we ready for what's coming?

Of course not. We're never ready. I don't really, I guess I go back to my earlier point about responsibility. So, you know, the platforms that are enabling misinformation to spread, and in some cases, say AI, where they're, in fact, creating misinformation, they're technology is generating it and then spreading it that way.

Those platforms need to bear that responsibility for what follows from that misinformation. And I just think in a lot of cases, it's not good enough to put technology out into the world that generates lies. You know, that's kind of a fundamental problem I have. Right?

So, the kind of ethical school of thought that I come from is really informed by a kind of start of thinking about respect and dignity and autonomy. And generally, lying is wrong. Right? You don't lie to someone because to do so messes with their autonomy and messes with their ability to be self-determining agents.

Right? So, lies are really harmful. They're fundamentally ethically wrong. And yes, there are exceptions.

Sometimes lies ethically acceptable. You know, lying to directly save a life or, you know, certain occasions you can justify a life. But generally, they're wrong. And to put out a machine, say, generative AI, that generates content in a way where lies are mixed in with truth in a way that you can't really tell what to do about it.

You know, according to JPC, I've written a couple books that I know unless my memory is really bad that I haven't written a product. So, that's what I stuff is really problematic. I think most of the basic principles, it's like, come on, you need to do better. You can't just put a product into the world that generates lies mixed in with truth.

It does seem to me the more that I think about this and the more people I talk about it, and again, the other cause of unknown unknowns. But as far as I can see, it's just, this is the supercharging and the acceleration of kind of moral and philosophical and societal maladies that we currently are seeing. So, the kind of increased, I mean, for example, the amount of fake news that has been promulgated on Facebook and Twitter and elsewhere in the last eight years, probably longer now is huge. And it potentially has affected hugely affected big elections and things like that.

So, it's not like you haven't already got misinformation that is very convincing to its intended audience, already out there, as you say, when it becomes, the more uncanny it becomes, the more life-like and realistic say, fake photography. I mean, people are still there. Anyone with a decent idea of these things is still pretty good at picking up. That's a shot, depending on what had been doctored.

It wasn't actually all that convincing on close inspection. But we are going to get to a point some stage when it will be able to seamlessly recreate these scenes. And ultimately, I don't know. I think that just that, for me, is just an acceleration of what we've already kind of seen.

There'll be greater siloing, greater kind of atomization of views. And people who believe that, that you could demonstrate that it's a fake too, and it will never change the view. And I feel like, yeah, that kind of divide between people who, yeah, we're taken into things, we'll just kind of get deeper if not wide up. But see, the one of the aspects I've been talking about is the amount and the quality of fake news and disinformation is going to increase.

But one of the things that I heard, and it needs to be verified this is going to happen, but the disinformation and the fake news will be soon being able to be generated in real time. So, you could be watching a press conference and you think you're watching the actual feed of the press conference and you're not. And it's indistinguishable from the actual press minus what's being said. That's the level of disinformation that's coming, which fundamentally goes to the core of the nature of just traditional journalism and fact-checking and those kinds of things we're completely unprepared for.

And we haven't even experienced, we don't have any defenses. Yeah, absolutely. Yeah, it's a terrifying prospect, real time misinformation. Yeah, and again, we just need to try to do what we can to prevent that happening.

Look, if I put on my optimist app, I can say maybe this is good for journalism, proper journalism, that journalism needs to reflect on its standards too. We know that there are terrible trust levels in journalism, in Australia, fewer than less than half of population trusts, news media, and that's sad. And it's terrible and it's really bad for our democracy and our society and our common public life. So, let's hope that somewhere through this, there's a re-evaluation in journalism standards as well.

Perhaps people will recognize that, okay, the Sydney Morning Herald, in my bias here is where I used to work. So, I'll probably see this in the morning, I'll slightly throw rose tins of glasses, acknowledge all its faults. It does do decent journalism. You have some people there like Kate McLean and Nick McKenzie, who are really good at their jobs and do really forensic public interest journalism.

So, maybe, and this is being very optimistic, but maybe we'll see a bit more attention based on what are the journalistic values that really matter to us, how are they best carried out by journalists and can we therefore kind of get back behind this and support good journalism again? I think that's all I probably just amplified that. But I think when you have a situation of real-time misinformation, and by the way, I feel like misinformation could be applied to send people to us because it's just we speaking normally, but it just goes back to the toolbox of journalism in the first place, which is skepticism, verification, fact-checking, making sure that what your reports, your readers or your viewers is actually what has occurred. Yes, and it does feel that way that the media could have a stronger role.

So, the Pentagon didn't actually explode because it wasn't on the front page of the New York Times. So, there could be that aspect. But I also see this kind of factoring and you mentioned Sylize before, where there's people like the three of us who are consuming a lot of media, but also checking our media, getting our media from multiple sources, making sure that the facts that we're getting are true facts. And then there's people who are living in a social media bubble who are seeing all kinds of things.

And so, you end up having two parallel universes, which, that's also the issue, which we have been dealing with. It's not a new problem, but it could get too bad charge, as we mentioned. So, look, I do remember back in March when we're talking about all of this, I put the question to you whether the job of being a journalist was safe. And I do remember both of you pushing back and saying that the role of being a journalist was an important one, but also one that a language model like GPT couldn't easily replace.

You know, GPT can't go to a courtroom, can't go to a press conference, can't knock on a door and ask an uncomfortable question. So, I'm guessing you both still feel that way. But one of the points has been made and it's I think a pretty hard point to push back against any job that can take place at home, can be replaced by AI. And because of the pandemic, we kind of know what those jobs are.

In the case of Australia, at 1.4% of what's happening at home. Charlie, you've been thinking about the nature of the world to come. 40% of work potentially on the chopping block. It's not a rosy picture is it?

No, no, no, it's not. I think that's something that I've been giving more and more thought to recently as more and more information about AI is becoming out. It's kind of what the social and societal implications of that are. And I used to, you can really look, I've been trying to look to history for indications of that about, yeah, things like, for example, the Industrial Revolution, which wiped out whole ways of life for people in say, pretty rural areas, who are all stuck towards cities where the jobs now were and all ways of working.

And not ways of being in the world, of being part of a community by the family, who kind of wiped out. And it's interesting, I'm talking to Greg Patmore, who's an industrial historian. And he was sort of saying that what you find in very difficult, it comes to work at least, when there are these huge technological leaps forward that replace our ways of being that there are breaking of bonds of family in community. And you've seen new forms of resistance inevitably crop up in response to that.

So, for example, in this revolution, that's a huge milestone in terms of the idea of a trade union and the ideas that sort of inform that and the way that they are organized often are sort of strengthened in that period. And the same with things like consumer collectives, those people looking to protect consumer rights as well, they kind of cropped up in response to this huge historical and technological change. So, I mean, you're seeing things like, I mean, especially referenced in four about the idea of cutting greater control, or personally how algorithms say, there's also things on data columns that are kind of dropping up around the world. So, people who are treating their personal data, the way that an industrial group would have treated their labor during the Industrial Revolution as things that they can collectivize and possibly mark them with.

So, I think it's a scenic genity or answer, but I think it's obviously a terrifying view, but it will be very interesting to see just beyond in terms of people's jobs and white dots, we've seen that with the rise of automation in many industries, but it's sort of what does that leave behind? And how do people organize themselves, and do they organize themselves in response to that? For most economists, the general feeling that when there's a technological change and people put out of work, more jobs are created. But if we're talking about 40% of all work or something in the order of that figure, it's hard to see that there's going to be a whole lot of new jobs that replace that, even if we are suddenly all more productive.

So, do you share that kind of pessimism or do you feel that given time, given time that society will be able to restructure itself? That's a huge question. It's such a big question, isn't it? But look, let me put my optimist, just put it off, take it off, put it on again.

40%. Okay, by my calculations, that means we each work a three-day week. What do you reckon? I think that could work.

I'm up with that. I don't know. That's too big. But this is kind of related, that's definitely related to our larger discussion, but NewsGuard, I'm sure you both heard of and listeners might have heard of, it's an organization from the US that recently arrived in Australia, and that gives news outlets, it sort of checks them, and it gives them a credibility rating.

But one thing it started doing is looking around the world to try to see what has sprung up that is AI-generated journalism, and just their press release from late May, so like two weeks ago, the 19th, it says NewsGuard now identifies 125 news and information websites generated by AI, and it's also then developed a framework for defining unreliable AI-generated news and infosources. So, there are already, at least 125, I'm sure there are more sites that just generate all their content using only AI, which is not surprising, but it's worrying, because obviously, this is a way to make a few bucks. It's interesting if you, this will be much more people who write for Bigger Places and Cranky, but if we had to look for our own articles online, you know, say you're trying to find something that you wrote a while back to your former, invariably you will find three or four copies of it that have been sort of possibly, so it's not entirely clear if it's AI, but it could probably put through AI, which had a few words changed, and then put out from behind the unpayable, various kind of wild lines, so it's always sort of happening in that way. It's interesting that there was quite that many that were just entirely generated out of in air.

I just wanted to make one more point on the jobs point, because I think it's as interesting as I said, we've been at the state of the society where we could be working three days a week for a very long time, and somehow that never seemed to happen, they just came up with a few kinds of jobs to keep us busy, but also, yes, it's not just the number of jobs, and I think it was Forbes that kind of calculated that it will cost something like 85 million jobs, but it will create something like 69, so they'll be this very big shortfall in terms of the balance in that way, but it's not just also about the number of jobs, it's about the quality of those jobs, and jobs that give people meaning in their life, and achieve what they've kind of been dreamed and take some pride in doing. I've already heard the term prompt engineer, and I don't know that I want that future for my kids, I don't know you guys, but the prompt engineer, I look at the job, and so maybe I shouldn't be so down on it, but it doesn't really sound like a great future. No, I saw that too, I can't remember what the publication was, but I also thought surely that's not a full-time job. I think we also should probably, we have a bit gloomy, but there are obviously very positive applications that other individuals can have within traditional journalism, in terms of particularly data mining and things like that, they can make those processes a lot more efficient and quick, so we can't just talk about becoming jobs a couple of weeks, and that was in our first chat back in March, that we kind of ended on the idea that it's just a tool and it's a useful tool if used responsibly.

It's pretty clear from this discussion that we're all feeling that something bad has got out of Pandora's box, and is it too late to put it back in, it probably is, but we're thinking how we can deal with this. But let's see if we can end on a positive note. I mean, where do you see AI maybe five years from now, and how it's benefiting the practice of journalism, but also benefiting society? I think we're not in a great state at the moment, when I look at news media and journalism, there are some good outlets, but trust levels are really low.

We have statistics about a large minority of people who are turning away from the news. They feel like the news doesn't cover issues. They care about that the coverage is too negative. There are all sorts of issues for this.

That's a real problem. I think what we want to have is a public sphere where there is a selection of quality journalism, public interest journalism, and I'm not just saying this because Charlie's here, but I think he does a good job of that, and there are lots of other publications in Australia that's a good job of that. The ABC does a great job of that. It's the most trusted news source in Australia for a reason, but it's excellent most of the time, and it too makes sense.

We do have some good news media. There is a challenge. The challenge is coming both from some bad journalism that's out there. There's a lot of shit journalism.

It's also coming from AI. If I look forward five years and I try to imagine a positive scenario, what do we want? We want us to know we're sure AI is being used as a tool, including for the purposes that Charlie mentions of helping data journalism, but what I would have to see is that somehow AI has been part of the spark that has helped journalism to better itself so that we see more quality journalism, more public interest journalism, and somehow in the process of working our way through how we live with AI and get the best out of this tool, we also work out a way to oversee journalism in a more effective way. I don't mean that in a top-down, heavily regulated way, but just in a way where journalism has standards that it adheres to and complaints mechanisms are more straightforward and streamlined.

That's what I would hope to see that AI has been this really confronting challenge, but over the next five years, it's improved the public sphere and led to a situation where we have better journalism. I can only really echo what Satcha says. I think there's an element to which for all the philosophical questions that it raises and all the kind of future that you can envisage with AI, like most technologies, is not even of itself malign or malicious, and that any good or bad it is doing, it's only to an extent down to the health of the system that it forms apart. So the optimistic version of what we see AI doing isn't necessarily specifically down to AI, it's maybe part of systemic kind of reevaluation of the journalism landscape and how it's working, and that it's a healthy and useful part of that, that it's obviously sensibly regulated and not being put to kind of specifically malign usage or as control as we talked about.

Look on that note, Dr. Satcha Melatoris and Charlie Larris, thanks for being on Forfistate. Thanks Anthony, thanks Charlie. Thanks so much you guys, we enjoyed it.

I'd like to thank Dr. Satcha Melatoris and Charlie Larris for being on Forfistate, and thanks for listening to the program. This edition was recorded at the studios of Taurus-R and heard across the country on the community radio network. Forfistate is produced with the assistance of the Community Broadcasting Foundation.

Thanks to the foundation for their continuing support. Make sure you subscribe to Forfistate on your favourite podcast app so you can hear us talk about media, politics and a lot in between. We'll be back next week more, but in the meantime, you can stay in touch with us on Twitter. Our handle is Forfistate.iu.

I'm Anthony Dockrell, thanks for listening.

Frequently asked questions

How long is this episode of Fourth Estate?

This episode is 39 minutes long.

When was this Fourth Estate episode published?

This episode was published on June 2, 2023.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this Fourth Estate episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.