Digital Works Podcast

Episode 033 - Ash & Katie, Bytes #1 on Twitter's demise, ChatGPT's lies, and some research into social activity

• Season 1 • Episode 33

Share episode

The first in our new series, Bytes, where Ash and Katie will be discussing 3 things from the latest Digital Works Newsletter.

In this episode we'll be discussing some of the links in the newsletter that went out on August 29th.

We talk about:

You can sign up for the newsletter at substrakt.com/digitalworks. 

Speaker 1:

<silence>

Speaker 2:

Hello and welcome to the Digital Works podcast, a podcast about digital stuff in the cultural sector. Today's episode is the first episode in a new short form strand for the podcast. We will be digging into some of the things that we highlight and mention in our regular Digital Works newsletter. If you're not already signed up to the newsletter, you can sign up on our website at subtract.com/digital works . Joining me today and for all of the episodes in this series is the person who puts that newsletter together, my colleague Katie. And today we are going to talk about the newsletter that was sent out on August the 29th, and I'll put a link to that in the show notes for this episode. So, hello Katie .

Speaker 3:

Hi Ash. Good to be here.

Speaker 2:

How are you today?

Speaker 3:

I'm okay, thank you. Post Covid. I've just had Covid, which was not pleasant, but I'm fine. I'm back.

Speaker 2:

Wonderful. So I've picked out three things that you mentioned in the most recent edition of the newsletter , and today we are gonna be talking about Eugene Way's article titled How to Blow Up a Timeline, which is a really interesting dissection of , uh, just how and why Twitter has fallen apart in the way that it has. We're gonna talk about chat, G P T and Simon Wilson's piece on lying machines basically. And lastly, we are gonna talk about social media usage and the report that you shared, analyzing how that has shifted through the pandemic and through to today. So the first thing, Eugene Way , how to blow up a timeline, it's a long form piece. He's clearly a really smart guy who knows a lot about TikTok and Twitter. And as a Twitter power user, like you and I, you know, we've both been on Twitter for a decade plus, and I think he does a really good job of dissecting just why it has fallen apart in the way that it has. Yeah. He goes into quite a lot of detail about how Twitter works, why Twitter works, sort of the structural, technological, theoretical, philosophical underpinnings. How important do you think it is for digital folks to think about their tools like this? You know, think about them on more than just, how am I gonna use this, but actually, how does this thing work and like, deliver the function that it does? The

Speaker 3:

First thing I would , I would just say if , if for people who haven't read that article, I would go and seek it out. Eugene Wei is a, you know, a kind of tech heavyweight . He really knows what he's talking about. And there's been so many pieces recently about Twitter and how it's falling apart and you know, it's all Elon and everything and that's fine, but it's not hugely helpful in a way. Whereas I think this article is, which is, I guess what you're partly alluding to, to answer the question, I think it's not necessary that people kind of understand all the granular detail of how a platform works and you know, why it's built in a certain way or, or or whatever. But absolutely the thing with Twitter in particular at the moment is it's , um, Elon taking it over and all the changes he's made has fundamentally changed how it works and therefore there is an impact in terms of reach, in terms of engagement. And when you read the article, it crystallizes why those changes have had such a big impact and what that might mean going forward. And obviously, if you are pouring lots of hours of your time and the teens time into running things like Twitter accounts, you really want to make sure that you understand, you know, what the current position is and what's likely to happen going forward. Because, you know , <laugh> , the reality with Twitter is unfortunately it is completely going off the boil. And, and I think for a lot of cultural organizations, it will end up being almost a pointless platform, I would say. So I think you don't have to understand everything, but I , I do think at this point in time, it's really super important to kind of educate yourself about what's happening with, with social media.

Speaker 2:

Inevitably organizations are going to be perhaps casting their eyes about to try and identify a replacement or evaluate alternatives. You know, there's so many new social media platforms popping up. I think it's been interesting, you know, when you and I have been talking about the decline of Twitter and it's been, I mean this sounds, this is melodramatic, but you know, there's been, something has been lost there, you know, in, in the way that we both discover new things and interact with our networks. I do think it's also important to recognize, as Eugene's article says, that Twitter was a relatively niche platform for

Speaker 3:

Sure. Yeah.

Speaker 2:

You know, it was not used by a lot of people, but the people that it was used by their influence was oversized on sort of society at at large. Mm-hmm <affirmative> . And I wonder how cultural organizations should be going about evaluating which platform to spend time learning, building a presence on. Because I think it's always tricky to try and draw a line between your own individual and personal perspective on something and actually how that works for your institution and their audience. Because the old adage of you are not your audience feels important to bear in mind in this context.

Speaker 3:

Yeah, definitely. Especially with Twitter, it's so sad to speak about it in the past tense. I mean, obviously it still exists, but at one point it had such an outsized influence on society because of who was using it. And I mean that generally, so journalists, celebrities, people who have a lot of visibility. And so that brought more visibility to Twitter than you would've expected it to get, given its user size comparative to something like Facebook. I always used to say, you know, if you go out onto the street, you know the , the omnibus whatever, and you ask 20 people, do you use Facebook? Probably 18 of them would say, well yeah, I've got an account, whatever. And probably two would say Twitter and then people started using it obviously for customer service and stuff like that. But yes, absolutely, it was one of those platforms that gained this traction that meant, again, arts organizations, cultural organizations, brands, companies were like, we need a Twitter account. But it was never really clear to me the massive value of that, like genuine value because all the Google analytics accounts I've ever looked at show that Twitter was sending an absolutely minute amount of traffic comparative to things like Facebook and whatever you think of, you know, Facebook philosophically or morally, it was doing a better job of that. Of course, social media was never just about directing traffic to your website. And so, you know, it was about awareness, it was about building relationships. I think what we come back to, which probably isn't rocket science, is organizations really thinking about what is the value of this channel of communication if I'm using it to sell tickets or to kind of reach a mainstream audience. And for example, in the case of Twitter, that's probably not gonna be that successful if I'm using it for relationship building with journalists or for profile building with, I don't know , stakeholders, funders, whatever. Yeah. Maybe fine less so now. So I think it's about, you know, continuously coming back to the , what is the point of this channel and then how am I actually going to assess whether it is doing what we need it to do? Because again, particularly with social media, it's so easy to get into a pattern of just rinse and repeat post, you know? Yeah, I've posted today fine and not really have a sense of what it's doing. So yeah, I think, and now more than ever, it's really working out what are all these things actually doing for us? And if the answer's not much being brave and <laugh> resigning them, I guess.

Speaker 2:

And if you're interested in reading more on, I guess the decline of Twitter and perhaps the reasons behind that, there was a, a good thread on Twitter from Ben Sauer where he's used Carl Popper's Clouds and Clocks theory, which is basically saying that, you know, Elon thought that Twitter's problems were a clock's problem, an engineering problem, and actually they're a , A clouds problem, a human behavior sociology problem. And he's a clocks guy , he's not a clouds

Speaker 3:

Guy . Yeah. And I think that the other thing about the how to blow up a timeline piece that I think is so interesting to think about is the sort of difference between a social graph and an interest graph. So you know, Twitter, Facebook, that Facebook is a social graph. Twitter originally was sort of an interest graph. So what we mean by that is, you know, something like Facebook, you connect with people that you know and that's how you build out your network. Something like Twitter originally was it, well actually it was a bit random, but it was sort of about things that you were interested in and the people that talked about those things. That's actually changed now, which we won't necessarily go into now, but TikTok has really nailed the interest graph approach. And the algorithm is just incredible at saying, oh, this person likes videos about dogs doing stupid things. That's just show and , and actually the people that you follow is not irrelevant, but it's nowhere near as important. So again, I think there are some implications of that for cultural organizations for sure, the slight tangent . But

Speaker 2:

The next thing I wanna talk about is the Simon Wilson article that you linked to, which is titled, we Need to Tell People Chat , G P T Will Lie to Them, not Debate Linguistics, which I thought was a really interesting and important piece because certainly over the last couple of months you've seen a real rise in the number of AI assisted things. You know, even on LinkedIn now, you pop open a status update says, do you want AI to help you? There's AI in all of Google Office tools now in Microsoft Office Tools binging has AI assisted search, but as Simon Wilson says in his article, you know, chat G P T cannot be trusted to provide factual information. It has a very real risk of making things up. And if people don't understand it, they are guaranteed to be misled. And I'm just wondering what you think people should be mindful of as they perhaps start to incorporate AI assistance into their workflows?

Speaker 3:

Yeah, it's fascinating really, isn't it, where we are at the moment with ai 'cause it's absolutely like 2007, 2008 with social media. It's, we can't understand what the full impact is gonna be, although there are, you know , thousands of people commenting on what they think it's gonna be. So what's great about that article is just the very stark reminder that chat G P T and tools like it are not the same as a search engine. They're not looking into a database of knowledge. They are literally making things up based on what they've learn . And you , even if you know that on a sort of intellectual level, what Simon's article does is really bring that home quite starkly. And particularly he references, if you give chat g b t some URLs and you say, okay, you know, summarize the data that's from this U R L , it doesn't go away and do that. It just makes it up basically. And again, what that reminds us is that these tools are super useful, but they are not necessarily what we might initially think they are because we're so used to thinking about things like search that's so embedded in our consciousness now and like how search works and what it is and where it gets its information from. And we just need to keep reminding ourselves that it's not that. On the positive side though, there is a great video that I will include in the next Digital Works newsletter called Hallucinations for Fun and Profit. And it's about how if you accept that a tool like Chuck g p t is what they call hallucinating. So we might call it making things up, but yeah, the technical term hallucinations, if you accept that it's doing that, then actually it's a super useful tool for creativity because in the video the guy gets the tool to come up with suggestions for new products and for brand campaigns and things like that. Again, that's not a new idea, but again, it's about thinking what is the best way of using these tools? And are we all sure and clear as an organization perhaps, you know, what's our policy for using these tools and what's the most useful way that we can use them? I , it comes back to that thing of not just going blindly into these things, but actually giving it some thought. So I think we're at a stage where we're flipping from huge excitement and positivity about these LLMs large language models to nervousness and negativity. And then obviously the reality is it's somewhere in the middle and you know, we just need to figure out the best way to use them and the best ways not to use them. I

Speaker 2:

Think it is really interesting that, you know , six, nine months ago when there seemed to be a real spike in companies starting to visibly integrate these tools into their products, it mostly seemed to be related to, as you say, information retrieval. You know, is is AI gonna replace search? Mm-hmm . <affirmative> . And actually it seems like, you know, research that's coming out now is, is showing AI has the greatest current applicability when it comes to aiding creativity and aiding innovation and using AI as a way of supercharging creative activities rather than as a replacement for factual tasks.

Speaker 3:

Yeah, exactly. So I think there were absolutely some, you know, all the things that people use it for around, you know, summarize what's in this information and put it into 10 bullet points for me, those sorts of , uh, you know, admin type tasks can absolutely be useful. I think the crucial thing is just not, it's not getting confused that a tool like this, when you ask it a question, we're so used to asking questions of search and knowing subconsciously that it's going away and, you know, finding the data to answer that question. I mean, it's not always correct, but basically, and because we're generally asking questions of things like chat , G P T, I think mentally people have an assumption that it's doing the same thing, but of course it , it's not. And so yeah, I think definitely that sort of aid to creativity and also the admin tasks and, and the more that we all become sophisticated in things like prompts and how to ask it, how to, how to get the best out of it, I guess. Yeah , we have a lot to learn collectively on that.

Speaker 2:

I think that prompt practice is gonna be an interesting area because again, an article that you shared with me about AI creativity was showing that AI is most useful in that context when you constrain it very specifically, when you give it quite a verbose prompt, a lot of creative constraints in there, actually, that is when it's going to be most useful and most novel. And as you've said, we've all trained ourselves to be brilliant at getting things like Google to retrieve the piece of information that we need, and actually those inputs and that behavior doesn't work with AI in the same way, and we're gonna need to train new skills around that.

Speaker 3:

Yeah, and if you think back, I mean, I'm showing my age now, but like, you know, when Google first came out and sort of early , the early days of that, Google had to get it out into the world that rather than just putting in one word for search, you needed to put in phrases and you need to ask questions. Now, you know, if you are much younger and you've always known things like Google, that might sound bizarre, but genuinely, initially with search engines like that, Google like, oh, it works better if you actually like, make the phrase really specific. And so it's just sort of similar to that, isn't it? And a prompt engineer is definitely an actual role now, which is kind of weird.

Speaker 2:

For our third and final thing, I'm gonna cheat a little, and , uh, combined two, so you shared a report from G W I , which analyzes social media usage, and so I wanna talk a bit about that, but then I also want to talk about the fact that the A, B , C , the Australian Broadcasting Corporation has come off Twitter, you know, citing toxic interactions, the cost of being on the platform and the fact that they're getting better interactions with a, b , c content on other social media platforms. But first, let's talk about this report. What's it telling us? The

Speaker 3:

Kind of key thing for me out of this report is that social media has kind of hit a ceiling. That's their phrase. They saying that for the first time since they started doing these, I think their quarterly reports, I think they started doing it in 2012, social media daily usage has dropped, I think apart from in North America, which is a bit of an outlier. But the point generally is that there's a suggestion, therefore that kind of social media usage has perhaps hit a ceiling, which is, I think a very sort of interesting and, and important thing to be mindful of. If you are responsible for kind of communications and awareness that an organization, you know, there's no way that social media's gonna disappear. And of course it'll just like transform and adapt and whatever, but certainly we're a long way from like 2010, 2012, you know, the kind of peak of people being social media's everything. And there was a period when people was , would say things like, oh, you don't need a website anymore. You just need to be on social media. I think famously, again, I've show my age here , <laugh> , unless they got rid of their home page from their website and they replaced it with feeds from all their social channels, which I mean, bless them, it seems so, so naive now. And of course it got hijacked and you know, people were just posting all sorts of profanities. And so that very quickly got pulled. But the point is that it was the sort of techno optimism of social media and it , and it meant that we should go to where the audiences are. We shouldn't try and pull our audiences to our website or get them to sign up to a newsletter, you know, that was heresy. But the reality of it is, is that actually, you know, things are changing. Twitter is, you know, it's going down quite a dark route. I can see why A, B, C have taken that decision. Absolutely. There's so much toxicity on there now. And so the question is sort of, you know, really as an organization, again, you need to be very clear on a strategy for, you know, what do we do if of all these things disappear. I was reading a newsletter yesterday by Russell Davis , so he's a writer. He is wrote a brilliant book about PowerPoint. Again , we won't go into all of that. And he's got a new book coming out and he was actually joking in his email newsletter saying, I actually don't know how you promote a book now given that Twitter's sort of dying. So , uh, I guess , uh, buy it, please. You know, like I , so I think it is a really important time to consider what, again, as an organization, what are you gonna invest time in? You do need to consider, you know, how do we talk to our audiences? I mean, of course cultural organizations are doing that all the time, but the rapport and the sort of changes to Twitter really highlight that now is quite a seminal time. I would say .

Speaker 2:

Thanks for listening. We hope this was a useful exploration of some of the things that we cover in the newsletter. This is a new thing. So do let us know if you have any questions, comments, suggestions for any improvements, and we'll see you next time. You can find all episodes of the podcast on our website where you can also sign up for the digital works newsletter, subtract.com/digital works . You can also follow us on Twitter, which is what we will continue to call it. We are at Digital underscore works underscore Katie is at Katie Moffett and I'm at Big Little Things. Our theme tune is Vienna, beat by Blue Dot Sessions. And last but not least, thanks to Mark Cotton for his editing support on this episode. See you again soon .

Head Shot

Ash Mann

Host