This week: talking to books, bots and pods; echo chambers; and fish tanks and Musk’s couch in other news. Sandra Peter (Sydney Business Insights) and Kai Riemer (Digital Disruption Research Group) meet once a week to put their own spin on news that is impacting the future of business in The Future, This Week.
The stories this week
The myth of the online echo chamber
Other stories we bring up
Science study on lies spreading faster than the truth online
Seth Flaxman and Oxford colleagues study
Elizabeth Dubois Ottawa University study
Christopher Bail and team at Duke University study
Future bites / short stories
The blindness of social wealth
Hackers exploit casino’s smart thermometer to steal database info
The internet is buying Elon Musk a couch (It’s a Musk)
You can subscribe to this podcast on iTunes, Spotify, Soundcloud, Stitcher, Libsyn or wherever you get your podcasts. You can follow us online on Flipboard, Twitter, or sbi.sydney.edu.au.
Our theme music was composed and played by Linsey Pollak.
Send us your news ideas to email@example.com.
Disclaimer: We would like to advise that the following program may contain real news, occasional philosophy and ideas that may offend some listeners.
Intro: This is The Future, This Week on Sydney Business Insights. I'm Sandra Peter and I'm Kai Riemer. Every week we get together and look at the news of the week. We discuss technology, the future of business, the weird and the wonderful and things that change the world. Okay let's start. Let's start.
Kai: Today on The Future, This Week: talking to books, bots and pods; echo chambers; and fish tanks and Musk's couch in other news.
Sandra: I'm Sandra Peter, I'm the director of Sydney Business Insights.
Kai: I'm Kai Riemer professor at the Business School and leader of the Digital Disruption Research Group.
Sandra: So Kai what happened in The Future, This Week?
Kai: Our first story is from Engadget titled "Google AI experiment has you talking to books". So Google research is obviously experimenting with language processing in machine learning and it has now released two more of its experiments to showcase what its technology can do and one is about hooking up machine learning algorithms with its Google Books platform.
Sandra: The Google Books platform is actually one of those Google products that is absolutely fantastic and that we don't get to see much or hear much about and this time around Google is allowing you to simply ask a question and the platform will find various types of books that have content that is related to the sentence that you're asking it but not necessarily features the exact string of words that you are looking for.
So the way this works is that the algorithm that Google is using has been fed with billions of conversational type pairs of sentences so that it can now identify passages in the books that are...
Kai: ...in the spirit of these conversations so what it basically does, it's not keyword search like you would expect from a Google search engine. It is based on AI and therefore the results that it comes up with are more relevant than if it was just looking up those keywords because it takes care of context and it tries to guess the nature and intention of the question.
Sandra: The article gives an example for instance if you might ask about the greatest detective that ever lived. The answers that you might get back might not have any of the words in those sentences, it might refer to things like investigator because it knows it's related to detective, it might not have the greatest but might have other synonyms or things that are related to these words rather than direct synonyms that you would get. And one of the other games that Google has released this week actually gives you a glimpse into how they have thought of developing this. The other one is called Semantris and is a word association game that looks a bit like Tetris but it's trying to get you to recognise both opposite and neighbouring concepts so for instance if you have a word like photography then you will think of cameras, you will think of pictures, you might think of frames and so on. So trying to get you to teach Google the types of word associations that we would normally make and some of them are curious and interesting and lead to a product that actually works quite well.
Kai: So it is a kind of word association algorithm, one that is based on teasing out which words occur in the vicinity of others right so this thing doesn't really understand anything but it is really useful because it makes use of the way in which people use words in sentences. Now we thought we'd give it a go and have a look at the talking to books experimental platform and so first of all we want to say it's not really a conversation, it's not like you're talking to books, you input a sentence or a question and then the books talk.
Sandra: So you're talking at books.
Kai: Yeah so you ask, they respond basically. So we thought we'd ask Books "is podcasting any good?". And here is what Books comes up with: So our first entry is a book "Podcast solutions: the complete guide to podcasting" which tells us the fact that you can listen to podcasts whenever you like is another important part of the growing success of podcasting. Well that's useful. We're also being told that "many people find it similar to radio on demand but podcasting gives far more options in terms of content and programming than radio does" from "Teaching English as a Second Language", not a book we would have looked at if we wanted to learn about podcasting. "Podcasting is a little more forgiving, you can launch a podcast of limited duration, shut it down and archive it" from the New Influencers - A Marketer's Guide to the new social media" or "We absolutely love podcasting, it's not for everyone and it's far from necessary but if you have a big mouth it can be a fun way to connect with fans". So there you go.
Sandra: Speak for yourself Kai. There's a number of examples on Google Books that showcases the range of questions that you can ask and particularly fun one is a question of what smell brings back great memories and it's got passages from a variety of books, some fiction, some nonfiction, on how smell is linked to people, to places, to things throughout literature and science.
Kai: Oh this brings back memories for me of one of our first podcasts on the great horse manure the crisis of 1894 that you used to explain innovation.
Sandra: And that's exactly the point. So these types of questions are open ended questions where you do not know for sure where you will land where you might come up on interesting insights, interesting angles that you have never thought about. So this is one of the few examples of such conversational interfaces that allows you to ask a fundamentally different type of question than simply a search query or a retrieval or a pre-programmed answer.
Kai: So the idea plays on the concept of serendipity, that you find interesting bits that you wouldn't normally have searched for or that you wouldn't even have known how to search for and in a way that yields slightly more refined and intelligent results than your social media stream would normally provide you with. And I can totally see the usefulness in researching certain phenomena like for example I did a study on innovation in the music industry last year. This service would have helped me a lot in finding books that weren't on my radar that would talk about say you know the early days of Napster or how people used MP3 so I can even see some really interesting applications in researching pop culture themes and other types of topics.
Sandra: So I think this is a good opportunity to maybe reflect a little bit on where we've come with conversational interfaces because conversational interfaces and in particular things like chat bots have been on the radar in the news for the past couple of years and there's been a lot of hype around them and just to clarify before we begin when we're talking about conversational interfaces we mean chat, we mean voice but we also mean any other type of natural language interface where we might interact with technology and chat bots obviously were the earliest incarnation of this but for the longest time in the area of chat bots it was fairly simple type of tasks that they could fulfil. So things like retrieval for example I, in trying to purchase a magazine, I actually asked the bottom one of the platforms "do you have the Harvard Business Review or do you have the New York Times and does the discount apply to the New York Times" and the chat bot was quite good at performing these very simple tasks.
Kai: Well incidentally there's another article this week this one in Tech Republic titled "Chat bots are dead. A lack of AI killed them". So this article makes the point that chat bots have largely failed to live up to expectations and our listeners might remember that just about a year ago they were the rage. That was the big topic at the Mobile World Congress 2017 where people coined the term conversational commerce and chat bots were supposed to be the next big thing in how companies connect with their customers. We would now order a pizza we would just about do anything - customer service, ordering, browsing products via conversational chat bots and we could connect then in Facebook Messenger, in other platforms Snapchat and the like. So a whole new world of doing commerce and that topic has died down fairly quickly and the article quotes Dave Feldman Vice President of Product Design at Heap who says chat bots didn't just take on one difficult problem and fail they took on several and failed at each one. So the point that is being made is that because the industry almost took for granted that these chat bots would live up to expectations they put these technologies on to too many too difficult tasks and what we want to say is they can be quite useful when the task is really narrowly bounded, information retrieval, as you said as interfaces for other services but they really didn't live up to expectation when it came to engaging customers, building brand reputation and actually doing more than the simple straightforward order process.
Sandra: So what we're seeing today with developments such as talk to Google Books is the move from retrieval type tasks to more generative types of answers. So where on the one hand you would be able to ask a chat bot "do you have the Harvard Business Review?" you are not moving towards the type of questions that would ultimately result in me being able to ask "so what should I read today?", more open ended questions and the range of conversational interfaces has expanded from simply chat bots to as we've mentioned before a variety of mediums that include chat, voice, images and other types of interactions and we're seeing the maturation of some of these services, some in the chat bots space with things like Bank of America's Erica that is used for money transfers and financial advice and so on but also with a combination of such technologies in platforms such as Slack where they come together to not only help you pass information but to achieve tasks that you might want to do to coordinate a group of developers to places like Google Sheets where rather than typing in a formula you might ask the spreadsheet to give you a range within a couple of columns and so on and Google Sheets actually writes the formulas for you so things that are more intuitive to use but also that are slowly, slowly starting to deliver on the promise that such technologies made years ago.
Kai: Yeah but for me what all of this points to is that it's not like the Tech Republic article says that the true potential of chat bots will require further advances in AI and machine learning. Sure, that might be one aspect but what has been missing in the past is that people haven't really thought about where those chat bots would add the most value and in what niches they might actually be applicable because we shouldn't forget that they are still only based on text matching, they're still only based on parsing how humans use text and then work on the vicinity of words and in different contexts in sentences to then come up with replies that are basically simulating a conversation. There's no conversation happening and so the broader the context is in which a conversational agent or a chat bot has to engage, the more disappointing the results will be because you'll just end up with things that are fairly random. But if you have a particular task to solve, these things can actually work quite well.
Sandra: Speaking of things that are starting to work quite well, we now have a range of smart speakers in the house such as Amazon's Echo with Alexa who you can talk to or Google Home or Apple's HomePod - all of which are conversational type interfaces that have now started to improve to a point where you can actually find use for them.
Kai: And guess what? There was another story this week on just that topic in Tech Crunch this one, "In defence of the HomePod - Amazon's smart speaker dominance won't last", and the author Lucas Matney he basically tells his personal story he says in my home, you know those three - the Home Pod, the Google Home and the Amazon Echo - they live within about 15 feet of each other and so he reports that he's got firsthand experience of all three and sort of discusses the usefulness and what these things can do and can't do and where things are headed. And so the first argument that he's making is in defence of the HomePod where recent articles in Bloomberg for example have said that the HomePods or Apple's stylish but very expensive, not surprisingly, smart speaker has not lived up to expectation when it comes to sales apparently. And so people are questioning you know is Siri up for the job. Is this thing too expensive? Has Apple left the introduction too late? Is the gamble not paying off? But the author says that actually there's a few things that Home Pod does really really well.
Also I want to add that this is a typical pattern that Apple products always are supposed to be bestsellers in their first iteration and neither the iPhone 4 nor example the Apple Watch which is now the most successful wearable device by a long mile have been really successful in their first iteration. But the author says that Apple has actually concentrated on a few hardware things that the other two don't do as well and one is surprisingly important which is listening. So apparently HomePod is able to pick up voice commands even when the music is really loud and so you can actually speak very casually and the thing will hear you. Whereas the author says that he finds himself shouting at the Google device more often than not.
Sandra: As a person who cohabits with the Google Home device, I can vouch to the fact that you actually do need to turn the volume off on the TV if the TV is on it will not understand what you're saying. So that definitely would be an advantage to not have to do that every single time or to have to yell at it to stop the timers that it puts on. But there is one other advantage that might arise in the competition between the three because as you've mentioned currently the Echo does have most of the marketshare at least in the US by a long mile.
But it's interesting to consider given our earlier conversation the types of questions that are being asked of Amazon Echo. If you're trying to expand this into a broader set of capabilities that are not just about providing information then Google actually has an advantage over the other two given the types of questions that Google Assistant is made to answer. Whilst most of the questions posed to Amazon's Alexa have to do with retail or travel or sometimes finance, the types of questions that we ask Google Assistant in those categories are actually a lot broader, the types of services, the types of information questions are not so much geared towards a retail experience which is what Amazon's Alexa is encouraging but rather a more a broad set of queries that we tend to ask Google Assistant, not only on the HomePod but also on Android devices.
Kai: And so this points to another argument that the author makes where he says that I've always found the "smart speaker market" to be a pie that's sliced in a bizarre way. So the argument is that we're really comparing apples and oranges here. And while you can certainly compare apples and oranges, for the smart speaker market that means that those three devices actually have very different purposes and they are interfaces that connect us to very different services. So clearly the purpose of Alexa is to give us an easy everyday entry point into the commerce empire that is Amazon and make it really easy to buy shit whereas Google Home sits on top of a search engine so for Google it really should be about driving access to Google's service something that Apple can't compete with because Siri is not powered by anything like it. But for Apple, the Home Pod is really an extension of its music based ecosystem so that the integration of Apple Music with HomePod and the ability to integrate with other devices such as the Apple TV makes the HomePod a really great extension of Apple's ecosystem and I think that's the point - we need to understand these conversational agents, these chat services not as standalone things that can be compared easily but as useful extensions of the kind of services and ecosystems that companies are going to build. And on that basis HomePod will always be something else than Alexa and it will always excel at certain things and Apple has put its money on great speakers and audio abilities than the other smart speakers might be.
Sandra: And that is a space that Apple will surely dominate in terms of coming to market with an exceptionally good technology fit for a specific purpose. But when it comes to Alexa and Google Home, Alexa will probably continue to lose out as their only advantage given that they did not have a smartphone presence previous to coming out with the Echo in the smart assistant.
Kai: Nor does it have Google search engine or any other more holistic smart system behind it.
Sandra: So their only chance was to come very early to the market with a home assistant, managed to dominate that market through a very low price and being able to improve that service to a point where it would become the norm. Unfortunately now that Google is in the same space and actually quite competitive in that space but having access to the variety of search queries that we use on our phones through Google Assistant it has a very good chance of dominating the market in the more generalist smart assistant space.
Kai: We're going to our next story which is coming from the BBC - "he myth of the online echo chamber" and this article takes on a topic that we have discussed a number of times before and offers a different perspective. So it says contrary to popular belief we now hear more diverse voices than ever before. So it reports on studies that tries to do away with the simple idea that it is Facebook or Twitter that exposes us only to the opinions that we already believe in which drives polarisation in political discourse and is at the heart of the problem that we've discussed around the US elections and elections in other places as well.
Sandra: So we thought this is a very interesting story to look at because pretty much today this idea of having an echo chamber or the existence of a filter bubble as enabled by social media platforms is kind of a taken for granted, never questioned assumption. So let's have a closer look at what's going on here.
Kai: So the article reports on three separate research studies undertaken in the past few years that all put into question this truism of the echo chamber. So the first one is a paper by Seth Flaxman and colleagues at Oxford University and they examined the browsing history of 50,000 U.S. based Internet users and they actually found that the browsing behaviour is much more diverse than previously thought. So it is not the case that if I'm conservative Republican right-leaning, I'm not just being exposed to Breitbart or conspiracy theories or other things that we would normally expect. It is indeed the case that through social media Facebook, Twitter, and browsing the Internet more broadly, people are indeed exposed to news items and opinions that are quite contrary to their own.
Sandra: So a surprising finding given the perceived view. The second study they report on comes from a team lab by Christopher Bail at Duke University that looked at Twitter users' political positions before paying them a small fee to follow a bot that would retweet influencers from across the political divide and people who took up the offer of being paid to follow this actually rather than developing a more moderate or nuanced stance on issues that there were tested on, issues such as gay rights, they actually became more confident in their initial beliefs. So people who were conservative as in your previous example actually became more conservative, people who were a Democrat retained or strengthened the views that they had before.
Kai: And the third story is coming from the University of Ottawa reporting on a survey of 2000 British adults who were asked about the kind of news items that they would use. So the diversity of their media diet so to speak and the researchers found that only about 8 percent could be classified as living in an echo chamber whereas all others were actually exposed to a broader variety of viewpoints.
Sandra: Now the question really is: is it possible that we are still experiencing the effects of a filter bubble or of an echo chamber even though we're exposed to diverse views. And why is that?
Kai: So the simple hypothesis so far has been that the way in which people drift to either end of the spectrum and are more polarised than ever in their viewpoints is a direct result of a lack of diversity in the opinions and viewpoints that they're exposed to. This research however now says that despite the fact or maybe because of the way in which I'm being exposed to other viewpoints I am becoming more emotional and therefore more confident and therefore more set in my ways.
Sandra: So it turns out that the reason we are seemingly in a filter bubble has actually more to do with the way we function as human beings rather than the media diet that we are being fed. Social media contributes to this, yes absolutely, but it is because of our innate characteristics as human beings that we actually end up drifting and still experiencing the filter bubble.
Kai: And we shouldn't forget that what was tested here was Facebook and Twitter which are probably better at actually exposing us to some more diverse viewpoints than platforms such as YouTube where it has been shown recently that the algorithm tends to lead people down a dark and often more and more extreme rabbit hole of content that leads to conspiracy theories and others so we don't want to say that this is true for all platforms but it does seem that Facebook and Twitter actually expose users to more diverse viewpoints but that still doesn't change the fact that polarisation happens.
Sandra: So let's look at what types of biases actually lead us to develop this closed or extreme view regardless of what we're exposed to. And two such biases are put forward in the BBC article - the first one is having to do with motivated reasoning and this goes back to a lot of studies that in this case looked at political identities that we might have or other types of advantages where we will actually devote extra mental effort to dismissing any evidence that disagrees with our world view so that we can be more sure of our convictions so motivated reasoning was one such theory.
Kai: And the second one is the psychology of self licensing which says that if I've just recently been exposed to a viewpoint that is contrary to my own I might tell myself it's okay to now expose a more extremist version of my own belief because I've already had some dealings with the other view and now it's okay to actually be racist for example.
Sandra: A third human trait that might lead us there is actually confirmation bias. And Professor Alan Dennis at the Kelley School of Business at Indiana University working with a number of colleagues is doing research on how to combat fake news on social media and we will be talking to him in a separate podcast about this but just to foreshadow some of that it turns out that confirmation bias actually plays a much bigger role in what we choose to accept as real news or as fake news even then how believable some of that news is. So regardless of our media diet, the things that we choose to pay attention to or we choose to believe actually depend much more on our traits and our biases rather than on the types of things that we are exposed to. But there is one more thing we need to discuss here. The recent study in Science Journal that spoke to the fact that false news is actually being spread much faster than news from respectable verified sources. The study highlighted that it wasn't actually so much bots, it turns out that it's actually people who tend to spread more fake news and more outrageous stories rather than the real stories on the Internet and contribute to amplifying the effects that we were discussing.
So there seems to be something else at play other than confirmation bias or motivated reasoning or self-licensing, something else happening on the Internet.
Kai: And the BBC article actually alludes to this but doesn't really expand on the fact that online when we communicate we are in a largely anonymous environment or at least in an environment where the things we say, the opinions we espouse, doesn't carry the same kind of accountability than it would have if we were to talk to our neighbours or friends offline. And the fact that we are spending more and more of our social time online rather than in our traditional communities where we wouldn't actually get away with polarising opinions and fundamentalist beliefs because we would have to moderate our messaging and we would have to be much more balanced in the way in which we act with others.
Sandra: And we would also have to entertain conversations that would be sparked based on the news that we propagate. People would respond and we are forced to actually enter those conversations. Whereas in the online space we can choose to attend to comments or not. We can choose to attend to the effects of us sharing those news.
Kai: Because when you're offline you actually have to hold a conversation and you're accountable for what you say you can't just shout and then walk away which is often what happens online where you can know espouse a radical message but you don't have to actually be accountable for it. No one will point fingers at you when you go to the shops or when you meet in church. So offline we wouldn't actually engage in the kind of behaviours that these channels let us get away with which contributes to the whole problem.
So by and large what these articles point out is that just attributing the polarization of discourse online and the way in which it affects life in political discourse and society more broadly to the fact that we have too little diversity in our social media streams is making it too easy. So as often in life the issues are much more complex.
Sandra: Interestingly the article also has an inbuilt solution because as it turns out it's actually up to us much more than up to the platforms. So whilst it is our own biases that result in these platforms driving us to more and more extreme views, it is also in our control to actually spend time being a bit more conscious about the way in which we engage with the news items or the sources or the different topics that we see online and the way we choose or not to share them, react to them, comment on them, like them and so on.
Kai: So let's go to our Future Bites/ Quick stories. Sandra what's something that you learned this week?
Sandra: My first insight was from the New York Times and it was actually a beautifully titled article titled "The blindness of social wealth" and this spoke to loneliness and social isolation as a problem that we don't often see because those of us who have it who have a thick web of social connections, who have a social calendar, lots of friends at work and lots of friends that we engage with in both our online and offline world quite often do not see that a lot of people actually suffer from social isolation, from loneliness and that this social wealth that some people have make the problem virtually invisible to public conversations.
So we do not talk about loneliness, we do not talk about social isolation and the article talks about 40 percent of adults in the US, other studies have spoken to about 60 percent in Europe of adults who are feeling lonely or isolated which in turn could contribute to the issues we've discussed just previously.
Kai: And that article actually mentions that the big issue surrounding Facebook is not privacy it's that it and other social media companies are feeding this epidemic of loneliness and social isolation which harks back to the point we've just made earlier. I think that as we have less interactions in our local communities where you know we would have to engage in civil discourse where our opinions are moderated because we have to get along with people from other viewpoints, social media lets us get away with connecting with people who polarise and we are less accountable for our messages.
Sandra: So what's something that you've learned?
Kai: Well this one is a weird one. It's quite possible that our listeners will have seen it because it was all over the news. So I picked the Mashable Australia story which is titled "Hackers exploit casino's smart thermometer to steal database info". It reports on Nicole Eagan the CEO of a cyber security company called Darktrace who was on a panel organised by the Wall Street Journal who gave a few standout examples of how hackers gained access to corporate networks via Internet of Things devices. And there was one where a bank had their data stolen by someone hacking into a CCTV camera but the most significant one was this casino where hackers broke into the WiFi connected smart thermometer in the fish tank in the lobby to then hack its way to its database with all the accounts information of its high rolling guests and then found a way to actually steal that database and downloaded it via the fish tank thermometer, which is just I think not just a funny story but also a serious story because it points to the risk that we're creating by hooking up more and more often cheap consumer devices that have internet access and that connect to our networks.
Sandra: So my guess is this is just the beginning of some very interesting stories that will come out of the Internet of Things that is projected to take off incredibly on the back of connectivity and very cheap devices. So something we'll definitely keep an eye on in the future.
Kai: I'm still waiting for my smart fridge that will order milk but I understand you have another short story which is A Musk.
Sandra: It most definitely is A Musk. It comes from Tech Crunch and it turns out the Internet just crowdfunded billionaire Elon Musk's new couch.
Kai: So we had Zuckerberg's cushion last week, this week it's a couch.
Sandra: This week it is a couch. During an interview on CBS This Morning Elon Musk gave a tour of his Fremont factory and showed the couch on which he slept during the long nights trying to fix the problems with the production off the Model3. So the Internet has decided this is an unacceptable couch.
Kai: It's too small, it's too ugly.
Sandra: So a campaign has been created on Go Fund Me by Ben Sullins from San Diego. In the last two days almost 500 people contributed.
Kai: So when the article reported about fifteen hundred dollars had been put towards the campaign but the publicity has certainly helped to buy Elon a much nicer couch.
Sandra: Yep so it's is now two days later and we have more than six thousand dollars for a new couch, that will buy you a decent couch.
Kai: People came out and said "Are you stupid? Why would you purchase a couch for a billionaire? The man can buy his own couch" but some people came back and said "well he clearly can't because he hasn't". My view is if we can't educate him about his views of AI at least we can put together and buy the man a decent couch.
Sandra: So Elon Musk's couch, Zuckerberg's cushion, that's all we had time for this week.
Kai: Yep we're looking forward to seeing the new couch. Thanks for listening.
Sandra: Thanks for listening.
Outro: This was The Future, This Week made awesome by the Sydney Business Insights Team and members of the Digital Disruption Research Group. And every week right here with us our sound editor Megan Wedge who makes us sound good and keeps us honest. Our theme music was composed and played live from a set of garden hoses by Linsey Pollak. You can subscribe to this podcast on iTunes, Stitcher, Spotify, SoundCloud or wherever you get your podcasts. You can follow us online on Flipbook, Twitter or sbi.sydney.edu.au. If you have any news that you want to discuss please send them to firstname.lastname@example.org.
**End of transcript**