The Future, This Week 02 March 2018

Disclaimer: We'd 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.

Sandra: Today in The Future, This Week: lawyers v. AI, logos all the same, and ring my bell. I'm Sandra Peter I'm the Director of Sydney Business Insights. I'm Kai Riemer professor at the Business School and leader of the Digital Disruption Research Group. So Kai what happened in the future this week?

Kai: Our first story is from Mashable Australia and it's titled an "AI just beat top lawyers at their own game". The article reports on a contest that was hosted by a company called Lawgeex that creates an AI for legal stuff basically. And what these guys did they created this algorithm that is capable of reviewing non-disclosure agreements. These boring yet everyday documents that every company and professional have to sign on a regular basis. It's a lot of work for lawyers checking these documents. It's an obvious target for automation because it's such a high volume task and considerably expensive and so these guys set themselves a task which was the AI should beat a group of legal experts.

Sandra: So on the one hand you had the AI law platform and this new study together with professors from Stanford University, Duke University's School of Law, and the University of Southern California. Against on the other hand 20 very experienced lawyers and they get to evaluate five non-disclosure agreements and to identify 30 legal issues, things like arbitration or confidentiality of relationship or indemnification. And they were scored on how accurately they would identify this issue.

Kai: Yeah. And so the AI scored a 94 percent accuracy rate in spotting all the various things that had to be highlighted in those documents. It's worth noting that the AI was trained with thousands of documents but the five documents that were actually used in the experiment were not part of that training data set. So AI scored an accuracy of 94 percent and the human lawyers on average 85 percent.

Sandra: We should also note another slightly bigger difference - the AI had completed this task in 26 seconds while the human lawyers took on average 92 minutes.

Kai: Yeah. So the fastest lawyer did it in 51 minutes while the slowest took 156 minutes to complete the task. That's a lot of six minute interval charges if you had to pay for that service while the AI completed it in 26 seconds so we might have to come up with a new interval charge like six second intervals or something. Of course you would have to do some work around those 26 seconds because someone has to vett it but it would cut out a considerable amount of time from this task. It is worth noting that the highest performing human lawyers scored on par with the AI. But that the lowest performing one only got a 67 percent accuracy so by in large the AI performed really really well.

Sandra: We should note that the types of tasks that were put forward in this study are the tasks that are very similar to what many many lawyers these days do every single day of their lives: review various types of contracts. The majority of documents that they will look at are these types of documents: non-disclosure agreements, wills, operating agreements for various organisations, purchasing contracts and so on. So this type of AI actually goes to the core of what for many lawyers is their day to day work.

Kai: Absolutely no one is suggesting that the AI could build an argument or build a case to be used in court, but the bread and butter of the law profession is the rather everyday boring task and AI is making fast inroads into this space which is also shown by a second article this week in The Next Web (TNW) which reports on a different AI - a free platform that is available developed by a Dutch entrepreneur which essentially does the same. That article on the other hand also notes some of the difficulties, okay the Lawgeez case is really interesting and it's got a scientific study behind it. But obviously it's portrayed in very positive light while this other article actually goes to some of the difficulties and the nitty gritty work in building such an AI platform.

Sandra: So what we want to do here is first point out some of the issues with actually creating this technology or having this technology available in the first place. It sounds fairly straightforward the way it's set out in the article but there's quite a bit of complex work that goes into training some of this AI. And then we want to look at what would it mean to have this in the law profession as a commodity or as a service.

Kai: Yeah so the TNW article points out a few things: first of all the lawyer who created this AI points out that the algorithm very much embodies his own interpretation of those documents and the way in which he would red flag certain things. So he says that whoever does that task of annotating the documents with which the AI is trained their preferences or biases will be embodied in the AI and we've talked about this previously so this is nothing new. But it also points out that it is a lot of work because every document, every sentence has to be meticulously annotated whether they deserve to be red flagged or whether they are innocuous. And so all of these kind of things have to be done before the AI can be trained with that data set.

Sandra: And then there is a number of jurisdictions that would have to be taken into account even for a nondisclosure agreement, we have different legislation, in different parts of the world. You would also have to make sure you do not over sample some parts of the world versus other parts of the world when you're trying to train some of this AI which can result in some interesting effects.

Kai: So yeah in the article it says that once trained the AI for some reason put a red flag on any NDA that was formulated under California law and the guy went back and looked at the data set that he had used to train the algorithm and it turned out that there was a disproportionate number of documents from California which turned out to have difficulty so the algorithm came to conclude that anything under California law had to be red flagged and so once rectified with a different set of training data that problem went away but it just shows that whatever the AI does it very much depends on what was already in the training data. And it's often only after the fact that those problems become apparent. And so whenever we create a software a lot of work goes into testing but we will have to come up with very different testing routines I would say for these AI because of the way in which errors and biases become embedded in those AI in very non obvious ways.

Sandra: In this case however this will be slightly easier to catch because these systems will make suggestions of how to amend contracts, of how to adjust them, and the errors and the embedded biases might be much more obvious than they are in other industries.

Kai: Which brings us to how are these things going to be used? And across those two articles it seems to suggest two different ways. One is in the hand of lawyers or law clerks as a way to very much speed up their work or as a free tool available to businesses or customers as a way to have a quick check done on a whole stack of NDAs to see which ones deserve to be sent to a lawyer for more scrutiny.

Sandra: Interestingly AI in the law profession now can become a commodity something that is sold to 20, 30, 50 law firms and that's just a basic package that everyone will own and have much like other software packages that are now available in this industry. But also the potential to be a service whether that's a free service or a subscription service allowing for a different type of business model in the law profession.

Kai: Which brings us to what will be the downstream effects of this once this technology becomes widely available not just for NDAs but for any kind of checking of standard documents. So in the first instance you could argue that it will make life and work of lawyers and clerks easier. And of course that might actually in the short term increase their profit margins because they could charge customers their standard fees but cut out a significant chunk of work. But of course because we live in a competitive world this will have an effect on bringing prices down for these services especially once we have free services which will then have a downstream employment effect for sure.

Sandra: So obviously law students and junior lawyers will need to not only be aware of these developments but also know how to employ them, know how to make most of them in their practice, but also for the general public this can be an exciting development if that increases access to some of these technologies and to some of these services.

Kai: Yeah absolutely. So we would expect that law firms down the track will hire less people to perform these tasks and instead have people who are capable of training AI because I would expect at least that rather than having pre-trained AI become a commodity and a thing in the market that we will actually AI as a commodity and that law firms will use their past data to train their own AI and compete on the basis of who has the more effective AI system so two scenarios that we can envision, a fully trained systems as a service or AI as a commodity that will be used to custom train their own AI systems but this will become a thing in the law profession and related work.

Sandra: Definitely a technology that's starting to come into its own. We've talked about AI and the law profession for quite a long time but it's finally starting to impact some of the core work that lawyers do rather than really interesting things at the fringes as it has done in the past.

Kai: Absolutely and for the first time we can actually see in more detail how it works and a trend for these things to indeed become a commodity.

Sandra: Which brings us to our second story of the day...

Kai: Logos becoming commodities.

Sandra: So our second story comes from Fast Co.Design and it asks the puzzling question of why do Google, AirBnB and Pinterest all have such similar logos. And the article goes into examining the logo of AirBnB and Spotify and Pinterest and Google and observes that these companies have all moved to remarkably similar sans serif fonts with no other embellishments, no other design elements added to the logo.

Kai: Yeah absolutely. So as always we put the link in the show notes so follow the link, have a look at the article where it shows the logos with which these companies started out. All of which were quite unique, had their own quirky little fonts and colours and everything. And they really all looked very similar in terms of topography these days. So the article asks the question why is that and they literally asked this question of a number of branding experts.

Sandra: The article interviews a number of CEOs and chief creative officers and creative directors at a number of design companies to try to understand what is happening. And a lot of the theories they put forth has to do with this need for simplification in a world where we are besieged by various logos and images and branding and so on that all of these companies clarity and simplicity has been a core of their strategy hence this is why they are simplifying their logo and making it more generic.

Kai: I quote Thierry Brunfaut creative director and founding partner at Base Design says all these bold and neutral logos are telling the consumer the same message: our brand and our services are simple, straightforward and clear and extremely readable.

Sandra: Andy Harvey from Moving Brands talks about the fact that these tech brands deliver their personality through their service and their content and their voice. So they do not actually need these logos to be the bearer of that message.

Kai: Yeah so look my first inclination when I read this article was to call bullshit because when I read the language around you know why they had to become all simplified and clear and the same in a world where people are bombarded with visuals and which is very busy. I want to say if it was otherwise and they went from generic to more complicated logos we could probably say the same in a world that is bombarded with visuals and has become very busy they had to come up with very standoutish kind of logos that are unique, that speak to the individual customer and their personality. So sometimes I'm a bit frustrated with this language because you can explain everything and anything. But on second thoughts I think there's something deeper in there that speaks to how social media has developed and some of the things that we've discussed on The Future, This Week previously and what is actually happening in the market. So as a visual representation of a broader phenomenon I think it's actually quite useful.

Sandra: So there are two things that are happening here: first is and the article picks this up as well is that we have since 2007 moved to a world of smartphones, of tablets, of devices with smaller screens. For which actually simpler, flatter designs that do not have a lot of embellishments do work better. So technically there is a factor in that.

Kai: That's a good point I hadn't actually thought of that.

Sandra: It's in the article.

Kai: Is it?

Sandra: Yes. If you scroll down after all of the quotes from the various people and you think it's over there are three ads including a free sample pack of...

Kai: Oh my God you're right. But it says oh...but it says about the author. This is usually where the article ends. Okay well. You go on, tell me about it.

Sandra: So simpler logos do work better on smartphones. And that's one of the arguments that we can make for this. The other thing that it starts to pick up is really this idea that what's happening with logos in technology companies, it's just a type of mainstreaming that has happened to all other industries. So think about painkillers and whisky and other products where most of the logos in those industries kind of look the same. They're part of the same category and they're not trying to stand out but rather convey the same signal to a very large number of people and this is what's happening to tech companies today.

Kai: So the idea that social media has become mainstream means that these companies these days they do not speak to a particular clientele they are not addressing the needs of a particular niche or a hipster group or an older target group but they have to speak to everyone and therefore to no one in particular and so to move to a very generic, very inoffensive logos makes complete sense but to me it also mirrors the move away from being a quirky start up with a particular culture to going corporate basically, to becoming a white wash, in certain respects sanitised, corporate giant that offers a convenient service that is easy to use. So in that respect the language used is quite correct.

Sandra: So the reason we wanted to look at this story was to try to highlight the fact that quite often we are tempted to go into these novel complicated explanations just because something is grounded in technology, if it's a tech phenomenon it must be new, it must be different to how we've done things in other industries, it must be something where there is a novel explanation or another set of circumstances. And sometimes a lot of these technology companies as they mature they follow the same sort of dynamics that other established industries have before them and they find the same types of principles apply to them as large corporates and a lot of the strategies that we've learned provide an advantage in other industries actually have quite a lot to offer to tech companies as well.

Which brings us to the next story of today. Amazon is spending $1 billion to buy doorbell/camera startup 'Ring'. A Shark Tank reject that has turned into a massive success story.

Kai: Yeah. So this company, as many companies before, started out as a struggling startup in a garage. And the founder Jamie Siminoff took this idea to the Shark Tank TV show where he basically spent what was left in the till to build an elaborate set to demonstrate what his solution then called Doorbot could actually do. So, here's how it works - it's a little camera based device that goes next to your door, someone rings the device, camera springs into action and hooks up to an app on your iPhone where the picture of the person at the door appears and then you can decide whether to actually take the call or reject, pretend not to be home, which might actually be the case because you can answer the doorbell from wherever you are in the world as long as you've got connectivity and so it's a way to increase security but also convenience because you can answer the door and tell the delivery guy to leave the parcel by the door, even if you're not at home. Now the Shark Tank people weren't impressed and didn't invest in it but the segment aired nonetheless.

Sandra: On the show the judges had a variety of reasons for not wanting this. Well it's not really an internet play, it's so much more a consumer device, was one of the arguments. The founder of the apparel company Fubu said oh I just really don't think I see how this could work. It's just not for me. I don't see the progression of how this could make more and more money said Mark Cuban. So really did not see where this device would fit into the ecosystem of devices that customers would have today.

Kai: Well, turns out first of all Jamie was worried that given that the investors didn't bite his segment wouldn't air but since it was put on the air eventually it had a really positive effect on sales at the company because customers seemed to disagree with the judges and actually liked the device. And so sales kept coming and the company grew significantly after the show went to air.

Sandra: Well let's remember this aired back in 2013 and smart homes weren't really at the centre of conversations. These conversations were just starting up.

Kai: Yeah and what else has changed... back then a whole lot of focus was on software because that was what was supposed to scale right. So when he said it's not really an Internet play what they mean is that anything that is just internet and software and app based can actually scale to millions and millions of users aka Facebook and Pinterest and all of these kind of platforms so hardware wasn't seen to be something valuable back then but that has changed quite quickly.

Sandra: So as the company grew it simplified its logo and its name and became known as Ring and began being distributed at top retailers, places like Target or Best Buy in the U.S., and by the end of 2015 they attracted the attention of Richard Branson the founder of Virgin who became an early investor into the organisation. But what did the Shark Tankers miss?

Kai: What they missed is the growing ecosystem of smart devices around the home and that a lot of companies building these platforms are looking to integrate as many useful devices into their ecosystems - Google Home, Apple's Homekit or in this case Amazon.

Sandra: So why is Amazon only a few years later paying more than a billion dollars to acquire this company? In only a few years, Ring actually turns out to be a very interesting component to add to Amazon's play to own the smart home. After the introduction of Alexa that already gave Amazon access to inside your home and to your shopping habits and to your shopping basket, Amazon has introduced Amazon Key which actually connects the inside of your house from a security and privacy standpoint to their ecosystem. So Amazon Key relies on a security camera combined with a smart lock to let couriers deliver your Amazon packages while you're not at home. Ring comes to complement that and to give Amazon access to your front door.

Kai: While you could argue that this is just another component in the smart home portfolio where Amazon positions itself against Google and Apple in this new emerging field of competition, Amazon does have a particular interest in the front door and that is also visible in the fact that they have already bought two other startups in the smart doorbells space. So this is the third one that they've purchased. They have Amazon Key as you just said and so they're really building an infrastructure around allowing customers to choose from a few different devices all of which supposedly will connect with their Amazon Key service and the idea that customers can easily give Amazon couriers access to their home and have packages delivered while they're away adding further components to the whole service and convenience experience around ordering casually by talking to Alexa and then having stuff delivered while they're not at home to just make sure that there's no barriers to actually shopping from Amazon.

Sandra: And indeed you mentioned the word convenience and I think that is at the heart of what maybe the Shark Tankers missed because back in 2013 if you called someone then very soon you will let whatever stranger is part of the delivery service that Amazon owns or uses, walk in to your house and deliver packages while you're not at home people would have told you nobody will ever give a key to a complete stranger to drop off things while they're not at home. But very slowly this idea of convenience...

Kai: You don't find that creepy?

Sandra: Depends what they're delivering. Yeah it is. There was a very interesting article in The New York Times last week that spoke about the tyranny of convenience and the fact that we find ever more efficient and easier ways of doing a lot of the tasks that we used to spend quite a bit on, things like shopping, finding, deciding on where to eat or where to go and that convenience has emerged as a very powerful force in shaping our lives and that it actually trumps our true preferences. So we decide for the convenience rather than the thing that we actually would prefer best. The fact that convenience comes to dominate every aspect of our lives and also the fact that we are actually willing to pay a premium for that convenience, this doorbell is not cheap it's between I think one hundred and eighty dollars to five hundred dollars for a doorbell. Also highlights the fact that it is actually quite hard to foresee upfront the types of changes in social norms or cultural norms that many of these services have brought to our lives. So when you try to imagine the future of the doorbell, it's very difficult to imagine it as a different kind of doorbell fulfilling a different type of function in people's lives which is what has happened in this case.

Kai: Yeah so the whole idea of the smart home is based on the convenience idea but also on the control idea. The idea that I'm in charge and I can do everything at once and I can do it now says this convenience instant gratification, I don't have to get off my couch to actually see who's at the door. I can just order in passing a product I want from Amazon Alexa and The New York Times article makes that argument quite beautifully that in many respects convenience these days trumps everything. The article then goes on to paint a rather bleak picture of what that means for our day to day life and our identities when we outsource everything to machines and service providers and really live in this convenience bubble. And that's an argument that we're going to come back to at a different time. I wanted to highlight a different aspect here and that is the security one. Because it has been shown previously that these devices are vulnerable. They can be hacked. And so the more we are letting corporations, devices take over control of essential aspects of our life which is protecting our front door for example we're also outsourcing a lot of that security. And while these devices are being sold with security and control as a feature, what if these systems are vulnerable and someone else can actually take control of your front door and let someone in remotely by hacking into the data stream. So these are the things that really are unanswered when it comes to smart homes.

That's things that are just in their infancy and we've discussed previously on The Future, This Week that even cars can be hacked and you can take over control of cars with self driving capability. So this ties in with that larger discussion around not just privacy but also security and vulnerabilities.

Sandra: Before we finish up this week's podcast, let's just have a quick round of Future Bytes. My first one is and we have to mention this on the podcast is that Snapchat stock lost one point three billion dollars after one tweet. This was Kylie Jenner's tweet last week that said "so does anyone else not open Snapchat anymore or is it just me? Oh this is so sad". And this has resulted in a 6 percent drop in the stock market price of Snapchat.

Kai: So who is Kylie Jenner?

Sandra: I'm not quite sure. I think it's one of the Kardashian sisters.

Kai: Yeah okay so look we can make fun of this but there's something serious that we can take away from this which is that these platforms are often talked about as tech companies, there's technology, there's apps, there's all of these kind of widgets and gadgets but at the end of the day these social media companies they're networks so the value that is reflected in their share price is very much the networks that they command. So when a celebrity like Kylie Jenner...

Sandra:...with roughly 25 million followers.

Kai: So when she says she's no longer part of the network, the value of that network and therefore the value of the company goes down, it's as simple as that.

Sandra: So what's one of yours Kai?

Kai: So mine is a Musk basically. Elon Musk has done it again. He's surprised the world by announcing that he will be putting wireless service on the moon. Teaming up with Vodafone, Nokia and Audi for some reason - are they putting a car in space now as well? They're teaming up with SpaceX to put wireless 4G service on the moon in 2019 and the article in Ink then goes on to wonder why and they couldn't really figure it out. But supposedly space tourists will want to go to moon soon enough and so that everything is ready and prepared once they hit the lunar surface, 4G network connectivity is already there. So that Kylie Jenner when she hits the surface can actually open her Snapchat and tweet to her 26 million followers. I suppose.

Sandra: 25 million. So when do you think we're getting this on Mars which is where I want to go?

Kai: I don't know. But by the way, Elon Musk also got told off by NASA this week for putting this car in space because there's a concern now with space debris and space junk and they basically said it's not a good leading by example to just put a car in space for no apparent reason.

Sandra: So if you drive on the moon you can actually now get Spotify which reminds me we are also now on Spotify so look for The Future, This Week and Sydney Business Insights on Spotify.

Kai: Oh very nicely done Dr Peter. So Sandra, what's your next story?

Sandra: It's Apple opening medical clinics for its employees and I think this is quite an interesting one because Apple is quietly planning these AC Wellness clinics and these will be medical practices that will be part of Apple and will be made available to Apple's over 120,000 employees. A large part of that is obviously in the United States where we know there are various debates about affordable health care. So what Apple is planning to do is to hire a large number of nurses, medical practitioners, physicians and so on, to be able to improve the health and wellness of their employees.

Sandra: And this connects interestingly to a number of other stories we've done in The Future, This Week where employers of these large organisations actually have income that is well above the average wage that other companies can pay. So they already have quite a significant advantage which then in turn drives up property prices in those areas and so on they are provided with free healthy food, opportunities to do sports, meditation, relaxation, learn new languages, and now are also afforded health care. So they create these bubbles where large numbers of people in certain areas actually can afford a lot more than what other companies or organisations can provide to their employees hence driving a lot of the talent to these organisations as well but also creating these micro ecosystems of housing and of services that actually drive up inequality in parts of the US.

Kai: So your concern is that this will only aggravate the inequality we see for example in Silicon Valley, San Francisco and surrounds where house prices have gone up, where people on lower incomes are crowded out and relegated to the fringes, where people bus in cleaners and low wage service workers where we really see a stratification where people on high salaries have everything provided for by their companies while other parts of the population are struggling to actually provide for their most basic needs.

Sandra: So what is happening at Apple is also actually happening at a number of other tech companies. Last week or the week before Amazon announced a collaboration with J.P. Morgan I believe to look at solving the healthcare issue for their employees. And this really risks making things like healthcare for everybody no longer an issue that is at the forefront of what these organisations or their employers are concerned about. And you can't actually blame the organisations because in a way the fact that the US doesn't have universal healthcare leaves open a space on which these companies can compete and let's remember that these tech companies compete for talent. And so they offer these services to attract the best employees and of course the side effect then is that in the eyes of people working at these companies healthcare once no longer a problem is also no longer a priority in the way in which they think about creating new services for the broader population. But let's remember that the bigger problem is the lack of universal healthcare because in countries with universal healthcare companies wouldn't actually have an incentive to create their own clinics because healthcare is readily available to everyone.

Sandra: So Kai have you got one last story?

Kai: Yeah a very short one. In MIT Tech Review it was reported that in California autonomous cars are about to start cruising without a safety driver. So from April 2, autonomous cars in California can now drive around without a safety driver behind the wheel. Companies who are testing autonomous vehicles will still have to have the ability to remotely control cars but there will essentially be empty cars - no one in the front seat driving around the streets of California soon it's heralded as a big step towards creating autonomous vehicles because you can now test them in the wild. But the question is also raised: what will happen to pedestrians that are being stalked by cars with no people in them. Will it help people getting used to empty autonomous vehicles driving around? Or will it just freak people out? That remains to be seen. And I guess that's all we have time for today.

Sandra: Thanks for listening.

Kai: Thanks for listening.

Sandra: We'll leave you with this...

Audio: That's all done. Um I'd call it a day but it's already called that.

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 Fliboard, Twitter or sbi.sydney.edu.au. If you have any news that you want us to discuss please send them to sbi@sydney.edu.au.