Today’s episode will be transcribed using machine learning from webASR, a free service provided through the University of Sheffield’s Machine Intelligence for Natural Interfaces (MINI).
Buyer Beware: This is not a perfect transcription! We highly recommend you listen to the episode.
Segment [112.73 sec.-137.58 sec.], Speaker 004
All right everybody i’m here with dorothy’s fitzpatrick e. me energy principle from go he is gonna join me today to talk about a topic that i’m really excited about because it’s something that we hear about in the media lot these days is wearing it are gonna ee i. going to talk about the hyper ve so dorothy’s tell us a little bit about yourself in what your topic is.
Segment [137.59 sec.-155.82 sec.], Speaker 007
O. k. robert dah it’s patrick i’m a managing crisp like co based in chicago and leader with the date and alex team here a cool and for those you don’t know where the largest management consulting company that folks is just on the financial services industry.
Segment [158.11 sec.-274.89 sec.], Speaker 009
You want to know what is so interesting about artificial intelligence and why i think it’s hyped up a little well um as with any new technologies is a period of understanding what exactly it is and how it can be applied original intelligence has been around for a very long time they call time was coined back in the sixties and what’s interesting about to me is that it’s going to be this this cycle right now there’s hope cycle um i love going from something as can solve all the world’s problems to the realities of hot classically done with it today um there was back in the seventies and eighties a lot of problems round here i am indeed we’re back to the thirties people have been talking about machines taking over their jobs through our nation um and what really struck me recently was so came up to me and said i’ve got to steal set that we need some help with can you just get some out one of your your one of your e. i. machine only i grew them to work out and fix it for us and what really struck me was even within our own industry was in our own company there are people that don’t quite get what’s possible what’s not possible official intelligence so um so that’s that’s why i think it’s interesting topic i will go ahead to find artificial intelligence for you and the way that i see it is that it is and a ticking the function of being able to make decisions um and by decisions i mean anything around pattern recognition or helping to decide whether a credit card should be issued an art or um you know determining whether an egg is bad not and minute that in some way using using machines and and.
Segment [274.93 sec.-308.48 sec.], Speaker 009
More so there are two different ways of looking at official intelligence one is that you simply take a decision and codified in some it was rules um or you can go step further and ashley have machine and try and learn the rules itself through neural networks for example machinery and then and come up with the answer as the second case is very interesting because it may come up
with unexpected patterns and for what for what i think the way that he thinks the world works.
Segment [309.02 sec.-508.14 sec.], Speaker 011
Absolutely i’m glad you find it in that context because i know the term i means many things to many people in for some people the scenes of really starts to pop up when we start seeing things like g. i. and so for the purposes of this discussion we’re not talking about eighteen eighty four he generally i like some philosophers talk about i think were working through any more to a business context is that a fair assessment i think we talk about the hype around us um that is that is true the recurrence why i was estimated that around twenty twenty five will be something called a singularity which is when human and machine intelligence will come together and this is an important topic of people don’t all fun talk about one went on with the realities of a. i. it’s not in human versus machine context it’s in an augmentation or current context so around twenty three five that yes by the the combined intelligence of machines and humans will explode and and indeed she’s me take over humans and in terms of pottery able to do but today were you yeah we’re talking about more realistic things within certain colours are confined conditions such as a game of chess one anne and deep blue was able to gary casper off chest and more recently being able to at google’s arthur go and beat up the game that is the hardest game for humans to play apparently more recently and then even more recently than that i believe that the reason and he either learns the game of chess from scratch and was still able to beat the best chess pastures out there either human are machine in record time so um this is where the grey area is when i’m talking about tara but we are talking about some very smart machines right and i think that one of the differences here is specialised e. iris is generally so if you think i’ve got a calculator or like you said the machines it had been trained to play specific games like chess or go in i know it will has recently built a machine that they were able to teach it boat butchers and go so that’s a little bit of and all that mean but it’s sir it’s a far cry from the intuition that human possesses today it’s a tell me a little bit about why you picked this topic the hype of i at this particular point in time against it’s a case of um we’re hearing a lot about it in the news and there’s a lot of excitement particularly from the scientist side of things all possibilities but also a lot of fear as well what people think about their jobs and have seen numbers about how many jobs may be automated over the next twenty years there’s a lot of concern about that so one of the things that we need to do we talking to our clients is reset that and say yes you will probably see automated trucks on the road and five years but no it’s not likely that medicine is going to be on was the next twenty years so those sort of things make it for nick this a very interesting time and so that’s why i picked up.
Segment [508.49 sec.-522.32 sec.], Speaker 004
And you mentioned you really really you feel like the concerns about the piece of learning are hybrid now will become a reality if that will be thirty five isn’t it.
Segment [524.28 sec.-529.86 sec.], Speaker 003
Um yeah i guess so is is that proven.
Segment [531.64 sec.-610.79 sec.], Speaker 011
Case where people tend to underestimate when things can be done over time and that experts underestimate the dreamers over estimate but the experts to undress what can be done so we were living in arthur c. clarke and william gibson time’s right now and so i do believe that it’s out there i do believe that it is it’s possible probable will happen and it may not be perfect but there’s a lot there’s a lot going on and what you may have just heard the background is my google assistant kicking in i’ve got me then aye that’s cellar oh qualities i don’t understand i had it just sing because time i here absolutely there you go o. k. signs there’s no issue off something you mentally i’m going to try to put some insurance together for this in a to a pipe see and hear us he is a philosopher for an answer it’s been a good amount of time talking about he ain’t some of the things that you’re talking about with with chess and go in the see he touched on our recent one of his pockets all into to that if anybody’s interested in listening doric tell me a little bit about what’s going on in this space ring out for you.
Segment [610.97 sec.-634.21 sec.], Speaker 006
So um ones the one of the interesting realities that we have to face is that um were all within a lab oh aye ayes certainly possible it’s being used in very specific circumstances very constrained conditions but in the general case for most oh.
Segment [634.51 sec.-763.35 sec.], Speaker 006
For most users in most businesses are simply not quite ready yes to utilise it i so google can do things with voice recognition it’s very smart there microsoft is getting really good image recognition as well um the car manufacturers are getting pretty good at putting thomas vehicles are introducing them to a different level so you may have assisted cruise control for example but in very specific circumstances and it’s going to be a leap to to actually bring that into the more general sense of business were he notes most of the work and decisions have been done by by people these days and one of the realities of his coming to bear is our days are simply is not the area and by data i just mean regular dear i mean the information that’s that’s produced by processes um or maybe produced by now since and there are coming out the systems and it is if if you are aware of that the work is scientists do right now they often complain that over half the half the job is just cleansing dated cleaning date it and it’s not just that the days it is dirty and it may be from generated with a certain perspective in mind so for example it may come from a sealed system the say system is good at knowing what was sold it may not be good at knowing what was purchased it may not be good at knowing what the credit worthiness of the customer was or who they were buying it four and so it’s really important to know where the diggers coming from um some did me just peel off for environmental reasons for mum to not be correct and this is very important the internet of things um and in other cases and very importantly for me i think the long run is that somebody do
it comes with inherent bias and a good example is um there was some say done.
Segment [763.74 sec.-919.3 sec.], Speaker 006
Avon the care of of of patients in hospital with certain conditions and they found that those that had asthma i had better outcomes and the simple reason was not because they were less secure and certain type patient but it was that that when they came into the hospital they were being treated differently right from the outset because they had asthma they were they were being treated much more carefully for for when they were diagnosed with influence so they had a lower mortality rate than the regular population who was who was it was diagnosed with its influence and so if you just look straight up you you would think that it’s o. k. to give that those with with as less care because they have a lower likelihood of of of of passing when in reality they take extra care was already built it so that’s us an example of bison so it could be dirty it could be off because of the way that was captured or it could have inherent spices in it so that is one sort of reality that we were dealing with um the piece of it as well as that when you try and bring different pieces of data together they don’t always match up just right and so um that’s another classic problem that you know you and i were talking earlier today with a client of ours there is now string of days having different perspectives on the data is is critical to this as well so that she can make the proper decisions and oftentimes this is um this is something that the people are work the dealer building or they may have just you know they may have got idea of how things should go based on their own experiences and pass it was a very difficult again to to codify into business rules so all that taken into context texan of course that was the joke that you know there’s no chance of was will take over the world we can’t even work out on coffin sick so yeah so could it could you know all these things were to go into that yes there’s great promise and yes there are specific cases of how it’s working very well for us today but in the more general sense and there’s still a lot of work to be done and it’s not until two businesses are able to generate good quality data that they’ll be able to make good quality decisions.
Segment [919.73 sec.-942.57 sec.], Speaker 004
Thanks at there you think your honour i’m gonna read say and sharon seen so what you say is there is there’s almost a world of basic blocking tackling that we have to do with our seed of data to get it ready not only four and ever said to be questions that people may want to ask you ringing on and on table two but one of the other challenges is how do we.
Segment [942.79 sec.-953.48 sec.], Speaker 004
Track dealing cash now for questions that we don’t know we’re going to ask in the future and so i think that’s a really interesting piece to think about it if you any thoughts on that.
Segment [954.01 sec.-992.61 sec.], Speaker 003
Yeah um yes i’ve got a traditionally dealer whereas it back round on data hoarder by nature as much as much data and admire as much history to that date as
i can and oh it is it is true that it’s particularly bringing that historical aspect things change over time you may introduce extra products maybe more countries created or less currencies available of the theft throws everything out of whack a little bit and all that basic block and tackle and is can be challenging him yeah you’re right that um that is his his be.
Segment [993.54 sec.-1015.33 sec.], Speaker 009
It is difficult to know now how the information is going to be used the future to and therefore how to avoid the challenges come forward so i think the basic rule is going all eyes wide open and one of the key qualities of a deed of scientists is to understand their domain and therefore understand dealers were.
Segment [1015.68 sec.-1022.41 sec.], Speaker 004
So tell me a little bit of doubt what if we use cases for rear fang chaucer squires.
Segment [1023.73 sec.-1026.68 sec.], Speaker 009
For free i am.
Segment [1026.76 sec.-1102.69 sec.], Speaker 009
It’s still reasonably straightforward stuff so um is is get the work of of um quantities of them analysts as they may have been call at one time or dear the scientists now our statisticians it’s it’s looking for specific patterns in the data again that because of by a stupid book up something that’s happening within maybe markets or something that is happening on the website so for example um very typical use case is the credit decision process in deciding whether somebody should get a credit card and after a alone done what i ideal rages for that arm another example is potentially looking for that intrusions on the website looking for parrots and to see if he’s coming in from a strange part of the world that your customer has never visited before is unlikely to visit aunt and fraud of course there’s another big area to to try and identify the patterns of spending have have changed significantly enough to comment toot toot toot throw up a red flag and contact the customer and ask them what is this really you.
Segment [1102.83 sec.-1165.08 sec.], Speaker 009
So those are some pretty pretty straightforward examples based on you’ll notice that all three of them are based on the soil idea of pattern recognition and trendy looking for out liars no calling out who’s out liars um unfortunately this is where it gets interesting again the inn in these cases um there are parties out there that will try and gain the system they will um they will accept our fake identities so they can apply for credit cards and they’re not just doing it to smile nominations is thrown to the next they’re not just doing it once or twice he doing it many hundreds of thousands of times per second so they can try and gain the system and find loopholes and and try and get through quickly and so for example if a bank which traditionally tries to lower the
barriers to its customers for spending money um they will issue a credit card that having gone through a fall.
Segment [1165.16 sec.-1246.52 sec.], Speaker 007
Credit check process they’ll say you know all we know you want to spend the money right now we know you’re going to spend it very quickly um so we’re gonna issue it to you right now and then we’re going to follow up and it just make sure everything’s good um so these bad bad parties can take advantage of that and apply for a credit card to get approval spend the money on before the crash the cars shut down once he find out that’s you know that’s not a real person um and this is is yeah what is there in the pinks have a huge motivation to do that because if they don’t true cos she’s going to go to the next institution right yeah correct i mean so for example you mister be standing in a best buy um register with a three thousand r. t. v. that you’re just itching to bring home and they can to tell the difference between that and another and another was some sort of hacking consortium and attempting to to problem so yes very difficult um get an extremely smart with the way that they do these things and when organising when using on operation a i can they can turn it back against itself and down so that i was his pupil humans are suffering because er because of the people using our tools against us.
Segment [1247.33 sec.-1255.13 sec.], Speaker 004
So when you ask you a question i want you to touch on robot process on a mission or pied and how that differs from me i.
Segment [1255.76 sec.-1266.45 sec.], Speaker 004
But when you know when does the line start to blur there because you know one could argue that basic work here is you’re giving it a task or apparently you wanted to perform you.
Segment [1266.65 sec.-1277.81 sec.], Speaker 008
But you would you call that basically i were you think there has to be deficiency sierra yeah i think i think there’s a difference between um.
Segment [1278.67 sec.-1371.87 sec.], Speaker 009
There’s a difference between automation of decisions and artificial intelligence um and it’s a very good question saw a. by process on a nation basically is the oh more advanced form of what we used to call flow tools and and it’s more advanced insofar as it can do a lot more it can not only incorporate the manual processes that we know of all but if you actually some reaching into systems and do things with screen scrape that weren’t possible before it is it is just it at its next generation it’s not revolutionary my mind oh but where does get interesting is is the level at which you can incorporate this and some artificial intelligence in there but um what we’ve seen is that it’s still pretty it’s kept separate saw p. a. will have certain rules in their business rules a little run like a rolls engine to get those things done it’s very strict and he doesn’t make its own decisions everything has to follow that pattern that
have been programmed into it but when you get into um a iron and deep worrying machine learning where you’re using think is structure such as neural networks and to help you make decisions and it’s not quite as black and white that component is normally bolted onto the r. p. a. at that stage so it will go through the process of arm.
Segment [1371.95 sec.-1471.95 sec.], Speaker 007
Doing everything for her mortgage application is a great example so will go through the process of collecting all of the information it will send out emails and record and look for responses back will ask you to send documentation the ducking he shall come back it me use optical character recognition of some sort of other and for to look at that documentation then process and certain way but it’s not until he gets that stage of actually making a decision based on a laurels um very balls are so i very type of decisions and looking at other much more complicated information as people tended to ken and to then they could make a decision and it jumps back into the market cross again with its decision so i’ll give you i’ll give you a good example here of um of what to what level we’re talking so your pew sir o’clock forward and do it step step step step step cos the a. i. side because of its power and breath and the things they can breathe bringing it can be considering more things than just i don’t know your income it could be considering the amount of time that it took for you to produce documents it could be considering um any number of variables that are shown too too have an impact on that decision um in marketing it even gets more interesting because if you were if you go to our website for example arm you can do something called motto very testing the website that world hope you understand.
Segment [1471.95 sec.-1597.92 sec.], Speaker 010
The best wait to present things to certain people that you may have never met before but through their clicks through your website you can start making to terminations has to offer of cohort they fall into one say fall into and you can have an impact on whether they will buy or not called going through the funnel in a website and the collar of a button or a particular image that shone to them or particularly out it can have significant impact on whether or not to going to buy a classic example starter lawrence played with the colour of a button and to offer a baggage it was baggage adults like seventy five bucks they would deliver your bank directly to your door so you wouldn’t have to wait at the airport um and they found there was er a one percent uptake in people taking that offer because they choose the coral the button that one per cent for delta airlines is tens of millions of dollars and that’s just from change the colour of the button so any idea what you’re reduced size it worked on there what’s going to impact the that that the uptake of the people if you want you know what’s the to people flashy clicking on that button and it also helped with the other side of it with his which is how come i had to identify people are going to be receptive to that signal to make that mash do it sort sort of work into areas and that’s um that’s worse of you go beyond just or p. a. and go into very smartly segmenting somebody and then show them something that’s very smartly gone to help them do what you are looking for just increase revenue be interesting to think of the
canna example of that we’re someone might use this knowledge to actually discourage someone for making a purchase like if they’re about to reach capacity on a flight or something they start discouraging that in the hopes of encouraging someone to take a different why for and on to buy a similar but slightly different product it is has more star or more cycles to you.
Segment [1598.55 sec.-1603.13 sec.], Speaker 009
Oh yeah that does get more complicated because.
Segment [1603.5 sec.-1669.48 sec.], Speaker 007
There are there are various things and play just like going into a store and days the assistant says i’m sorry we don’t have blue socks but if you go next story i’m pretty sure they have blue socks in the short term and they’ve lost that sock sale but in the long term they may have increased loyalty so that was a very that’s a really good example of um i was sort of a soft number that you’re making a bet on we’ll play off in the long run and sold as this idea that a lifetime value for customer is more it is is uses this concert of loyalty it’s not just about the revenue for an immediate self respect his lifetime value so um that goes into zoo at this i was i was he some psychology behind that um there’s also a partnership o. i. d.’s as well so that we can sell it to you but somebody else can and so yeah this is when it comes becomes very complicated and not only from the day prospective but you know if you’re a company like amber’s on erm you were you’re you’re.
Segment [1669.86 sec.-1726.21 sec.], Speaker 007
There are potentially erm big impacts to doing stuff like that so and answers we cats are to you but we can and they know they can fly that merchant with requests the marshes going to feel are the she has bad reviews recently then how does that impact of the clients experiences for the royalty to you so that that could be it becomes challenging but up i was on and glass brother because it’s a very interesting example you know that i was on the only real way that you can see sun exposing its its use of such things as and i know of two areas one recommendations of products and whether the weather is deeper and i’m not exactly sure if it’s more than other people like you things like this out but the arty put a lot of work into it needs to simplify it to that idea of recommended products and the second thing is that he’s around um.
Segment [1726.47 sec.-1831.44 sec.], Speaker 008
The shows that they recommend and that flicks famously runs oh competition to try and improve its recommendation engine which to me seems a lot of work for something that isn’t still isn’t very good erm you know it’s it’s it and throws in all sorts of things as well my kids wash you know some kids show and suddenly i’m getting recommendations from work consumers i don’t you know since he clicked on the current profile on the mary or you throw one kid’s toy in your right in your court in new recommendations are all there they were the explorer for the next six months right but but we’re dealing with very smart people here were dealing with very smart companies and so i think it’s indicative um yeah sorry one as reality checks that you have to call sometimes inclined to say it was so simple don’t you think that somebody
bigger than you would be doing it right now oh so that’s that sort of like you know that one of the things you sometimes you have to say is yes it’s a great idea but um he’s just too difficult right now aren’t we advise you to to hold off maybe think of difference plan the planet that technology so you’re touching on some challenges here inside do want you to go into it on this let’s take the example of work is on so for instance i met my wife and i we use one account for her entire family and i would be interested to know your take on do leader purcell activity in divided into different segments of profiles like you know maybe i do most of my purchases between nine p. m. and eleven p. m. at night where is my wife does most of hers between twelve and one o’clock.
Segment [1831.7 sec.-1833.55 sec.], Speaker 012
You know how much.
Segment [1833.96 sec.-1971.18 sec.], Speaker 003
Power do we have rain out and builder’s segment asians even within one warden in where do you see people doing that there were some challenges with that they’re not generally do that today and there are simple challenges i’ll give you an example oh there it is again in banks want to keep things easy for the customer so they here and there are banks that will make it easy for you to sign up for a joint account and in doing so they say o. k. just get personal data to take in their name oh click the box yes i agree and then get person beats type in their name and click the box yes i agree that building you’ve got a joint account with a problem with is there’s no guarantee that it was actually two people that did that it could be the same person doing it twice and that is quite literally there and compliance because it is required by law for you to have a contract the signature suckle soon a your car as a contract with your customers and if you can prove that there were two costume separately saying and then you’re you’re a compliance so that’s a simple example of a way isn’t it sorry sorry challenging um the there are certain signals that can be used for she not very reliable so i should talk to me sees about such things and how how to do this and it’s just it’s simply and too difficult to to make things oh hello genius certainly in other words make it a nice little split between one group and another group and it just it it’s too hard and me mean not be you and your wife for me that you may be purchasing for your friend she may be purchasing for a friend’s daughter you may be percy if your dad and so it’s too easy to make assumptions about it then so yeah we’re still faced with the you look for that watch and hammers on and then everywhere you go from that point forward the next three months you got to see ads for watches all the time and only watchers almost the point that you feel coerced into buying a watch but that’s exactly maybe what they want so and it it’s it’s still very it’s still very difficult because the world is more complex place them down and computers can quite understand.
Segment [1971.32 sec.-2051.9 sec.], Speaker 008
Well since i’ve been thinking about through another when you’re away so one of the challenges i have is that we like to use one in as i can’t like you say but my wife kept finding like she’d emails else yesterday give her for her birthday and things like that so yeah i started my own account in a row i
buy everything in that account she’s under strict instructions not to open evening that comes addressed to me that when i walk into that account there’s nothing on it that i’m ever going to want to buy for myself for a recommendation so it’s like you said my facebook is littered with you know things like she’s remained a racket ball in so it’ll be everything for wrecking ball worth a racket washes and i couldn’t care less about something like that so that the the the seventies is how we twist it for me yeah what did you do bring up least think so which is which is that they’re using these methods for trying to have somebody self select themselves and so um yeah different counts and her family accounts and family are accounts from now as well because there is protection for for underage um individuals online and so they have to be treated slightly differently and so you you know it’s it you he it’s a necessary for you to say um.
Segment [2051.94 sec.-2138.06 sec.], Speaker 003
This person is is i want dinner’s myself and and adults that can do things for herself and her financial responsibility um whereas my child you know me log in and there they can’t be marketed to um they shouldn’t be able to hear explicit music but stuff yes anything engines having those kinds of filter her lenses this you yeah yeah so yeah but but when you set it set up this family counts now ich ich as with everything else to get the very clever so if you use the microsoft ecosystem for quite a while you’ve had this family concept where i can gift um money to my son so he can purchase schemes but at the same time i can also call family safety i can also lock down his account so he cannot be you will not receive advertise means he will not receive invitations to become a friend he will not receive chats he won’t be able to go online and play any games at a rate of about fourteen and so i like as a parent i really want that but on the flip side microsoft lot are getting a really good picture of who my family is how old my kids are and what they can what they can do with that information is is limited sometimes but um does give a lot more signal to two what’s going on in the real world.
Segment [2138.19 sec.-2144.56 sec.], Speaker 004
So you’re touching on something today that wrote down on those you to talk about from.
Segment [2144.62 sec.-2216.25 sec.], Speaker 012
You saying some things that were going on the space and you talking about a different pattern recognition to do things like longer conditions and things like that and you’re starting out a touch on the ethics behind me i and i do want to spend a minute talking about that is that something that i think is very concerning because you know some of the things we talking about what on as doesn’t recommend me that’s more of a convenience challenge for me bye if something is going to recommend or exclude me from a world or you know there are some albums out there now that are being used to decide who makes for things like that it’s arson theory significant ethical clock consequences there right then depending on what the model is it’s a neural network forces just a simple pattern recognition you have varying levels of ability to understand why contribute to that decision in so what are
some things that you think you know for clients and just even in general a philosophical standpoint what worries you in the space so there’s creasing concern from people so keen on mask is probably the standard bearer for arm.
Segment [2216.46 sec.-2235.86 sec.], Speaker 003
Join concern very high in a general sense and he’s talking about everybody being to go to morrison’s that’s an interesting he’s office think that if we’re ready you dozy hole k. well yeah most my ten year old son said something about the matrix last night in such a context but um.
Segment [2236.32 sec.-2260.1 sec.], Speaker 003
He don’t they do have to he does have very good points and you’re right it’s about blondes oh there is another in with it thank you as concept of red lining um where they basically do things erm it’s like districts map of these districts and they say the dish to have to be one bag per district and the reason they have to do that is because of banks.
Segment [2260.71 sec.-2340.41 sec.], Speaker 003
Could decide where they put their own branches they’d only ever put them in high income areas high high income a low bankruptcy areas because those the customers have they what um but they are banking is at some levels a social right and so the banks are required by law to to build their branches in under certain areas to provide banking and services such as the ability to simply have a safe place to put your money and so they require a lot to do that again if you put an a. i. on that just like you said it would probably not do that they would just put everything in all banks and i think marie’s so why is also what how do we deal with that today is we to do without through guidance in compliance and regulations and this is what you are muslim and his fans are saying and i agree with that is that we have to think now about the regulations that need to be in place to make sure that the these these albums that are used are not these are not simply cold logical albums they do take um the the ethics and and considerations for different for diverse populations into place into play and so i absolutely agree with that that.
Segment [2340.42 sec.-2351.83 sec.], Speaker 003
They’ll need to to to have that there and this is it’s for looking yes but today it’s possible to two to do these things down again you know.
Segment [2352.26 sec.-2435.53 sec.], Speaker 003
My son goes o. k. goo and o. k. cool comes up and he’s able to trick isn’t it to responding to him um we’re not there the eyes not there yet to be able to be trusted to make sessions but it didn’t it will be at some point in time and when it makes the decisions it has to be for good and should not be able to be abused or should not be able to two and disenfranchised certain certain groups of population the the piece about this is you know i talked to see larry in and people come together in my view of of the way things the two progress he’s through he’s through a teeming her augmentation for now
i does need to be supervised and i think in the future it will still continue needing to be supervised at different levels until we’ve worked out of the kinks and i don’t believe there will be a case worry i will be truly truly a timeless and be able to make all all decisions i think there’s if you do look at necklace is a series called altered carbon i’m only doing the first episode of there yeah so you’ll see that there sonny eyes in there that are better the pride themselves and be able to manipulate humans and and does not worry wannabe.
Segment [2436.31 sec.-2450.71 sec.], Speaker 004
So would you say that brain now from an ethical standpoint do you think we’re keeping pace with the ethics concerns or do you think we’ve gone over skis all a bit weird he capabilities are outpacing.
Segment [2450.78 sec.-2463.79 sec.], Speaker 011
My from the store the ethics well you brought a proper parole example and it’s very um it is a very good case and we need to keep looking out for these um i think it is.
Segment [2464.33 sec.-2504.25 sec.], Speaker 007
Pass it’s so if you take it out of the picture altogether you can also remember the banks are fighting regulation and being imposed on them anyway and they just they want to do that what they wanna do and um the challenges that the ice is commit enable that to happen quicker so you see flash crash is happening because we have a rhythmic training systems that make decisions much quicker than humans can stop them and so i think that erm it was a cases of our schooling or skis absolutely and i think these.
Segment [2504.77 sec.-2533.15 sec.], Speaker 007
Events i’m going to be important to remember that we can try and avoid them in the future but we’re humans we make mistakes and we’re going to build machines that make mistakes and it’s always happened and always will happen so we do have to train off you know what we can do a sometimes it i sort of think that too there’s a couple of er analogies you can use with fire and one is that um.
Segment [2533.44 sec.-2569.79 sec.], Speaker 007
We fire is good for for cooking food but is not controlled properly can burn your hamster so so but we still use it we still use fire we still weak because it has certain benefits to us um and then the others are analogy that around fires that um again we can use it for cook for cooking things and keeping us warm but it can also be used as a weapon and that’s where we talked before about the i ivor say i if it helps good people do good things we can hope that people do bad things to you i really like that you feeling thankful i’m going to seal that one from you.
Segment [2571.42 sec.-2616.39 sec.], Speaker 008
So when you sum up time doesn’t child is your ethics of bias in regulation are definitely ones that i i feel like ferns and even individuals are coming into contact with those inherited always buy back to the high part of this tell me what some of the places are a world where he eyes just incredibly over hyped like people are saying that it’s going to be fired but it’s really just a spark from the lighter that’s nice erm it’s so i think it’s i think it’s in there in that again i mentioned some people being overly concerned with the chart at this point in time i don’t think i don’t think we’re quite there yes um zoo didn’t do that.
Segment [2616.75 sec.-2632.61 sec.], Speaker 003
Machines taking over the world terminator it’s not quite there yes um but it is cooked it haven’t you have to be mindful we have to get ahead of this um so there’s to of the negative side and then that positive side of it is.
Segment [2633.82 sec.-2702.04 sec.], Speaker 003
He’s in terms of what people think can be done today isn’t he’s just isn’t possible it’s o. i. don’t we have erm hey why don’t we have a robots that can down come in and help take care of elderly people which i think is a black mirror but the way we just have that today i don’t just when were just craving hate me i am robot thing that does this it’s fantastic to give me two and and are like is a simple case of i’ve got some cindy the can keys get get on your machine learning things to fix it for us saint we up the area so i think he’s the sort of the coca cola things considers the name of side the positive side of the fear in that promise are not quite aligned with three hours here and there but there is the other agent we haven’t talked about here are the vendors and the few tourists who it’s with in there it’s within their their own interests which hopes this thing up a little bit too much and so um i think yeah i heard this the start about.
Segment [2702.82 sec.-2744.91 sec.], Speaker 003
If you add the word block chain it’s year tear or cryptic currencies is the name company and it is ten times easier to get funding um and so it’s that sort of thing where people don’t want to miss out this fear of missing out right and we don’t miss out those hyped up by certain people who were who were you wanna sell newspapers on i was sell product on sale software they want to get funding for the company and um actually they had little bit too much er so i think it’s again week we are now living in a world of arthur c. clarke and william gibson um and.
Segment [2745.11 sec.-2761.02 sec.], Speaker 008
We were probably underestimating the power of what can be done a long time but this structure just we go keep great reality check your urine reminding me other company you with the name but they were like a mini fashion company listed on.
Segment [2761.06 sec.-2844.15 sec.], Speaker 004
The stock exchange and they were at the very bottom about to get a d. with sting if they didn’t keep their market valuation over now so they re branded as a cricketer currency trading company in there you instantly shot up and gavin back into the round when they needed to be anywhere else in there for about a week before people caught on and said how are these are these folks are we doing manufacturing the gonna make it in’t crib difference trading and if i start phoning back to white year to work the afford to they must but i think the mayor should be under investigation by the s. e. c. so i’ll add one thing here i think some of the promise that people are incorporating and their products of court and corey i and i see two issues here i can cyber security and in some of the day and then my personal fitness products for instance it seems like every company now has some kind of quorum call d. i. feature but i think there’s things that you can argue about whether it is or is not ee i. and b. what use is it like it like what we said before people aren’t even good enough to do the blocking tackling nerve see edie edie nor are helping me to support their fitness goals to do we really care about you if there’s any idea recommends who they should be friends with there on a fitness tracking website.
Segment [2846.32 sec.-2868.13 sec.], Speaker 008
So i’m kinda curious like this but it seems like there are some people that are just brilliant you his features in calling to me i media not here they may not have any nine um yeah yeah o. k. labels i mean in some sense yeah maybe artificial intelligence um but again.
Segment [2868.34 sec.-2959.9 sec.], Speaker 003
You know going back to anything that could be on it in some way ah ah could be could seem to be oh smart so we’re talking at all this and i think it was marvellous came out and wired about a machine that was beaten maturing test for eighteen minutes and if you don’t know the true test is erm is that measure of a eyes out so we can have a conversation with the machine i think it’s human um and so according to that were already there but look at all things can do so it’s it thing gets mark ah ah ah ah just computer chips get faster we get more memory he gets more and more access to data so they can learn more um but um you know if you scale it back in there in the physical realm you’re looking at insects and can you say hot appointed insect has become intelligent are hot plants and animals become intelligent and as different measures that and so erm i think you know in learning busy world a world that’s sounds as labels and i think people will focus on um what it actually does what it does it is it it’s a spectrum thing right it’s not it’s not black and white is a tiara isn’t it is it molly high for me the measures is our fisherman and artificial ones are making decisions right incisions artificially incisions to me it’s a it’s a sort of a hurry.
Segment [2961.34 sec.-2966.09 sec.], Speaker 012
What are you most excited about in the space ring ow ha.
Segment [2966.53 sec.-2970.62 sec.], Speaker 003
I’m really excited about our.
Segment [2972.23 sec.-2998.05 sec.], Speaker 003
This revolution i do see it as a revolutionist i think the world economic forum as market as the fourth the fourth industrial revolution um i’m excited that this is going to arm provide a lift to what we do are beyond anything that we’ve seen before so um.
Segment [2998.75 sec.-3116.19 sec.], Speaker 001
You’re just like oh meeting lou took people out of factories to a certain extent because it was so who happened um but the way that um a flour miller a windmill which is the first case of an automated system um was able to take people out of the mess bills and out of having to grind the flour themselves um in some parts think that in some parts are tribes of to grind flour but way that it was able to live people three people up his most exciting to me so how that plays out with him grip grandeur contest what we’ve been talking about is and this is when you when you’re considering what charge against him what jobs are going to do the jobs are highly manual highly mean you’re i’m going to go and so those people are going to be enabled to do much more interesting things and it did the people that really like to do those things are going to potentially become artisans are craftsmen the people who didn’t like doing them can consider taking on other jobs and it’s not just about intelligence we’ve got remember sometimes it’s about empathy and those of the jobs that i’m going to go anywhere so if you’re a nurse you’re in a great place right now and you will continue to be a great place because um because that’s not something that can be automated that human touch that human relationship could never be replaced by by machine and so so they will be more i think there’ll be more empathy in the world more artists in the world and yell the more machines but those machines are gonna be doing things which we don’t want to be doing anyway i’m us again was your archers and you want to you know make that thing you’re suffering by hand and your people are very very successful and they still have one shops because they like making furniture inspector um but in a gentle kiss in john walker’s case it will lift everybody up and enable us to um.
Segment [3116.57 sec.-3118.88 sec.], Speaker 007
Two.
Segment [3119.04 sec.-3134.84 sec.], Speaker 003
Have certain things commented and by a. i. and so we can think about the more more interesting cases and um the more strategic cases as well.
Segment [3134.87 sec.-3303.64 sec.], Speaker 005
I think you may need to be some cost eerie implementation of these things were even if something tune completely replace the job it was done me before we have some method of transition you know like you said maybe we have an assistance phase in we transition the people that want to get out of that into something else read in just cove abruptly and in saying sorry you’re
got to retrain into something completely different yeah do it’s not gonna happen that way because because the lucky seven she easy to be trained at some level it is off on missions need to be trained at some lower need to be monitored so this is a whole cycle of take your process manual now reverse engineering understanding how it works sunny what the rules are in there and the people are doing it right now probably want to understand best but then there’s also these people who have never done it before but rains her fresh just blank slates perspective to ask the right questions to say oh that’s cool but how do you do that and you’re you’re something nice for oh i never thought about you know dot about going into detail on how i do x. y. z. anne you going to have that period of reverse engineering period recreating and so all the people that you know that create these things the colder as the engineers will have to create is um solicitors opportunity and cast to that piece and then once at this stood up and running then the past people that were the subject matter experts then get to sit back and admire this thing as as mentors are supervisors and make sure it’s running good for a period of time and and they again i don’t think the irish step away i think there there and there are jobs just become a little bit more interesting we’ll be dealing with the edge pieces a bit more so another analogy is if you walk into a doctor’s office these days these days you’re more likely to see a nurse practitioner who takes care of the eighty percent of the cases because eighty percent of the cases the scene out i know the doctor’s only deal that twenty percent of the more unique or difficult or um um challenging cases and are always there to back up the nurse practitioners in case case question but the doctors get to focus onto a hive that higher value more interesting work so i think the same thing’s gonna happen oh in in in automation artificial intelligence as well there will be significant investment in the transition and then even after the transition does people at war in that position before we just have more interesting work so you talked a little bit about this being something that could open us up to you know maybe slow down a little bit and focus more on the mark of a craft.
Segment [3303.75 sec.-3322.82 sec.], Speaker 004
Because we we have this a politician and then we talk a little bit about the d. i. been something nicking assist us and would potentially allow us to do more faster stronger how do we balance between those two in the side like halloween.
Segment [3323.14 sec.-3379.58 sec.], Speaker 010
Worth getting to a situation where allows us to slow down and be more mine for reading just forces us to do things even quicker and faster then accelerate the pace of art is more than they are today yeah so that’s gonna be a balance between the organisations that are driving this change in investing in paying for the change and um the workers who are big affected by it um it’s a doubt that’s going to be a bounce there that and i don’t think it is i don’t think he’s an easy way to strike a certain to determine this to be done um the and screw very very specific to circumstances so against consider some real world examples here um d. e. a. company here may use um knitting machines to knit sweaters.
Segment [3380.04 sec.-3438.21 sec.], Speaker 007
Because they want to lift them close to where they’re gonna sell them or they want patterns to be done in certain ways and easily available under the machines where there other companies that aside you know what we’re just gonna ship this work offshore and get it done cheap cheaper um in the short run and but a cheaper resources and then get it shipped back here where we’ll sell us and so there’s all sorts of different things and play there that you have to consider i don’t think i think you’ll be every case may be different and will be measured in some regard and um you know we may seek out more unique things and and it will be done by by by people who like to who like to be artists and sometimes we just need to screw we don’t care working from an arm if if if it’s made by a machine that’s cool as long as it’s ready when i need it where i need it.
Segment [3438.29 sec.-3449.62 sec.], Speaker 004
I think he’d be interesting to think about the implications of it coming from different providers like come for microsoft forces who all verses as in white the.
Segment [3449.8 sec.-3483.01 sec.], Speaker 012
Be as far as quality of life verses utility um you talk you talk about it i have a personality oh little beauty but more so you know i think it was initially the garden but he i think he’s developed is going to be marketed toward use cases that the people that created them are most suited to so i would assume that he eye for facebook would be more socially rented in the eye from it l. u. s. may be more generally applicable to.
Segment [3483.13 sec.-3484.79 sec.], Speaker 004
A multitude of.
Segment [3484.85 sec.-3487.96 sec.], Speaker 004
He u. s. slash industry.
Segment [3488.03 sec.-3502.2 sec.], Speaker 004
Which yeah but but more so how the culture of the company that makes it transfers over into the eye itself either in vice personality ooh in here we encourage you to use the c. i. a. to you.
Segment [3502.4 sec.-3511.22 sec.], Speaker 004
To make what why dance better for us as we encourage you to use the c. i. to perform better faster stronger than you could ever before.
Segment [3511.61 sec.-3521.62 sec.], Speaker 003
So interesting is interesting sort of concerts i think you’ve been a few different things there so it’s a little bit layering yeah i see that um.
Segment [3521.9 sec.-3523.31 sec.], Speaker 003
There.
Segment [3523.87 sec.-3600.51 sec.], Speaker 003
The eyes will be the use cases in particular will be based on market demand much man needs and i do think the tools the name may be different um i think certainly based on who implements them um whether it’s a company or a certain set of consultants oh we’ll have an impact on the quality of those tools the quality of the a. i. i don’t eat a lot of personality no i think it’s going to be either a fears quality type of type of detention job ah um the so he panics and probably go this is to is to have something that’s free of bias as much as possible or at least he’s compliant with the regulations that will be imposed upon us um and the nice thing about the iron is that it should become easier to order and to test and um are down we’re working on a way to make um we talk talked deep learning are dark dark marion and andy either where it’s very difficult to see how decisions are made and there are people like you making that more transparent um so i think that’s.
Segment [3601.99 sec.-3607.14 sec.], Speaker 003
I think that is is the way that will go like.
Segment [3607.69 sec.-3657.75 sec.], Speaker 005
Some t. v. shows you’ll see their architect and people are that are better at doing that than others but again i think it’s on quality and fit to mention i don’t think it’s on her to personality dimension but think that it was wasn’t west world where they started talking about the creators and artistic sides coming through and the characters um it’s er it that’s interesting absolutely but i don’t think we’ve got it there area true and that is another show that even though it’s on my recommendation i have not gotten to that yet not a good shot well i was on a t. v. if it were particularly if it’s anything to do with that is that allergens i have a list of shows that are on my to do list to catch up on all three corrigan is in west brom is another.
Segment [3658.15 sec.-3666.34 sec.], Speaker 004
But i i’m overdue for a lot of those today i am from the t. v. seasons.
Segment [3666.87 sec.-3688.73 sec.], Speaker 012
So i think you so much for joining me today i do have one of the question for you but for someone ignored you fruit for seamus time with me and sharing your thoughts it’s been a pleasure chatting with ian i look forward to hopefully having some more discussions like this in the future with you course you’re welcome so i’m going to give you the last word here and tell me something about the space that we don’t know.
Segment [3689.08 sec.-3693.56 sec.], Speaker 009
Oh my gosh share so much already um.
Segment [3694.19 sec.-3698.55 sec.], Speaker 012
Wow let me think of something i didn’t ask you.
Segment [3698.66 sec.-3700.89 sec.], Speaker 012
You think i’m a solution no.
Segment [3703.51 sec.-3714.31 sec.], Speaker 003
Wow i am absolutely dry blank maybe for heavenly i to back me up who could help answer this question um yeah.
Segment [3715.03 sec.-3745.65 sec.], Speaker 007
Here’s an interesting fact is a game called dover to which does competition for million dollar purse and um it was considered to be a milestone when and here i was able to beat that game and beat big she would play and game that surprised me and i get it may seem this interest trivial but to a lot of people it was a big deal so we’ve got something you didn’t know there is definitely something i didn’t know think fish in there.
Segment [3746.2 sec.-3754.04 sec.], Speaker 004
All right that is gonna wrap things up for this episode thank you so much for the time and i look forward to chatting with you again soon.
Segment [3754.21 sec.-3756.98 sec.], Speaker 012
All right cheers sorry but.