00:00hi everyone welcome to the asic Cinzia
00:02podcast I'm sonal and I'm here today
00:04with two guests we have Tom Davenport
00:07who is a professor at Babson university
00:09and a research fellow at MIT and Julia
00:11kirby who is an editor at Harvard
00:14University Press and a contributing
00:16editor to Harvard Business Review but
00:17the reason we have them today on the
00:18podcast is because they have a book
00:20that's just coming out called only
00:21humans need apply and the subtitle is
00:24winners and losers in the age of smart
00:26machines which is a topic we talked a
00:28lot about as software eats the world
00:29welcome Tom and Julia thanks I'm glad to
00:31be here nice to be here the best place
00:33is just kick off is you guys have this
00:36section in your book where you talk
00:37about this ode to the AI spring and I
00:40think that's a really important place to
00:42start because but listeners on this
00:44podcast and people who've been following
00:46the world of artificial intelligence for
00:47a long time often talk about their scars
00:49from the contrasting AI winter before
00:51that do you want to talk briefly about
00:53that as you suggest we've gone through
00:56various cycles in the space and this is
00:59probably the most spring-like spring
01:02we've ever had in the sense of interest
01:05in the technology the number of firms
01:07that are adopting it one of the things
01:09that always fascinates me is that even
01:11during winter there were a lot of things
01:14kind of quietly happening I mean 10
01:16years ago I wrote an article saying that
01:19automated decision-making was really
01:21percolating its way through lots of
01:23insurance companies and banks and so on
01:25for underwriting and credit issuance and
01:28so on but now I think probably with big
01:32data and analytics gave a lot of impetus
01:34to the topic and everything is in full
01:37flower all over the place yeah and one
01:40of my former colleagues Brian Arthur I
01:41wrote this really compelling view called
01:43the second economy and his idea is that
01:46there's I I'm actually only now putting
01:48it in the context of the AI winter but
01:50this entire time that we've been waiting
01:52for AI to have its moment it's this
01:55collection of things that I've already
01:57become automated in our lives that are
01:59invisible to us every day like down to
02:01checking in at the bank teller to
02:03checking at the airport I mean there's
02:05so many ways that automation has to your
02:07point percolated and permeated into our
02:09lives I think the question that's top of
02:11mind though for people is that no matter
02:14as investors or researchers or observers
02:17of the phenomena I think what really
02:19people care about a thing of the day is
02:21how it affects their jobs and the
02:22reality is and we've talked about this a
02:23lot on the podcast especially lately
02:25because we just came out of a DC series
02:27where people are that theme that came up
02:29on every single podcast was the
02:31realities of the job market is how
02:33people especially in the US can adapt to
02:37this world and even before that what are
02:39the realities of the job market as
02:40software and automation eats the world
02:42the big fear I guess right now and it's
02:45justified is that a whole kind of set of
02:49us who thought that our jobs our
02:51livelihoods were kind of immune to this
02:53encroachment of automation are now
02:56having to to rethink that confidence you
02:58know so we've invested a lot of time and
03:01money in gaining those college degrees
03:03and advanced degrees so that we can do
03:05you know this sophisticated knowledge
03:07work we thought that that meant you know
03:10we're not gonna be like those assembly
03:12line workers or even frontline service
03:15workers in you know fast-food settings
03:18who might be seeing their their jobs
03:22gobbled up by automation and even by
03:24computers but with the advent of
03:26cognitive technologies we're now seeing
03:29machines capable of doing decision
03:32making so you could see this as sort of
03:34three waves of automation that first
03:36machines came along and they automated
03:38that dangerous work and then computers
03:41came along and they started to automate
03:42some of that dull work like
03:44transcriptions etc now we're at the
03:47point where they're taking over
03:48decision-making and the scary part is
03:50that it's hard to see what is the higher
03:53ground that you can move to as a human
03:56and still be able to add value in a
03:58workplace that notion of a higher ground
04:00I mean first of all I think it's hard
04:01for anybody in this economy I mean we
04:04have to also think about non knowledge
04:05workers but this idea now that nobody is
04:08safe from automation is a compelling one
04:10and so when you guys think about where
04:12is the higher ground that you can go to
04:14as this flood of automation comes in how
04:17do you guys think about that I mean what
04:19is the reality I mean I don't think it's
04:20enough to say let's just give people
04:22skills training and better stem I mean
04:24these are all realities that that we
04:26need to adjust to of course but
04:28what now what can people do how should
04:30they think about this we generally think
04:32that since there's no higher ground to
04:36which humans can retreat then they have
04:38to find common ground with the machines
04:41that are going to be their colleagues
04:42and so a lot of our book is around this
04:45idea that we'll need to augment their
04:49capabilities and have them augment ours
04:52rather than a set of activities that
04:55humans will always be better than
04:57computers we just don't think those will
05:00necessarily exist but at least for the
05:03foreseeable future there will be ways
05:04that we can add value to what they do
05:06and work with them as colleagues in a
05:10whole variety of fields and I think that
05:12augmentation oriented future we think is
05:15by far the most the most likely one for
05:18how a I travels through the occupational
05:21world and that's especially true because
05:23of the fact that when computers move
05:25into a workplace they never take
05:27anybody's entire job that what they do
05:29is they take away certain tasks and all
05:32of us have you know a certain percentage
05:34of our time that is spent on
05:36yes--it's knowledge work but it's very
05:38codify Abul rules based knowledge work
05:42that you know given the right algorithm
05:45could it be turned over to a computer
05:47because it doesn't involve a lot of
05:49ambiguity or creativity and so it's
05:52those tasks that are getting chipped
05:54away from jobs today so the reality is
05:57that every knowledge work job is going
06:00to see this encroachment of smart
06:02machines into the workplace and it's
06:04just a readjustment that people have to
06:06make to figure out okay what do I do in
06:10this equation what does the Machine do
06:12and how do we make best use of what both
06:15parts of that equation can do best so
06:18before we talk about how people can
06:20engage with it and this is a really
06:21important question I do want to pause
06:22for a moment on this concept that you
06:24guys are both reinforcing of
06:25augmentation versus automation outright
06:28because I think it's really important
06:29because it's the difference between
06:30reacting to something that just sort of
06:32happens to you versus proactively
06:34thinking okay let's expect this let's
06:36treat it as a given let's like just say
06:38like machines are gonna be our
06:39colleagues machines you're gonna
06:41our jobs whether your knowledge worker
06:44and they automate certain parts of it or
06:46your work or where they automate huge
06:47chunks of it I think that's a really
06:49important idea and it reminds me of
06:51theory of the original notion that Doug
06:53Engelbart had of augmenting intelligence
06:56and really thinking about computers the
06:59mouse as an extension of us versus
07:02something that compete with us which i
07:03think is the reality of our lives I may
07:05think people treat their smartphones
07:06already as an appendage literally so
07:08there's a little bit of that already but
07:09can you guys talk a little bit more
07:10about the difference between
07:11augmentation and automation and why
07:14that's so important yeah you know I
07:15think there's always been this tension
07:17about AI whether it would aught
07:23fully automate us and Engelbart was
07:26certainly among the original thinkers
07:29about the augmentation idea
07:31unfortunately we can because automation
07:34is a more I don't know dramatic and
07:37attention-getting scenario we tend to
07:40keep coming back to that
07:42so probably everybody listening to this
07:44has seen that oxford university studies
07:47suggesting 47% of US jobs are
07:49automatable it doesn't say anything
07:51about when that might happen or you know
07:55which tasks will be taken over in jobs
07:57by machines because the julia said it's
08:00never whole jobs it's just tasked with
08:02an job so what we tried to do in this
08:04book is to find various ways that humans
08:07can either do things that computers
08:09don't do very well and probably won't
08:12for the foreseeable future or as I was
08:14suggesting earlier work alongside them
08:17almost as colleagues or you know instead
08:20of being a supervisor of a human or a
08:23set of human to be a supervisor of a
08:25machine and really treat them as
08:28co-workers with a certain set of skills
08:30and also a certain set of shortcomings
08:33let's talk about those ways of engaging
08:34then I'm am concretely what are ways
08:36that people can engage and adapt their
08:39readiness for jobs or existing jobs for
08:42this world the kind of core augmentation
08:45job as we see it is we call it stepping
08:48in that's really treating a smart
08:51machine as your daily colleague kind of
08:54during its performance day-to-day maybe
08:57improving it a little bit you know this
09:00has been going on in some industries for
09:02a while and insurance for example in
09:04another most exciting industry but one
09:05of the earliest that that used a I
09:07particularly rule-based systems to a
09:10substantial degree there were
09:12underwriters who would underwrite the
09:14policies that got spat out by a computer
09:17and would improve rules if it looked
09:20like they weren't working and you know
09:21this goes all the way back to the
09:23Industrial Revolution where when
09:25factories were installing new textile
09:27machinery for example somebody had to
09:30configure it fix it educate people on
09:32how to use it effectively and so on so
09:34stepping in we think is a very key job
09:37there's also stepping up and and if you
09:39think about the archetypal stepping up
09:42job it's a managerial role not unlike a
09:44hedge fund manager who even though all
09:47the trading in a hedge fund might be
09:49done by a machine the hedge fund manager
09:52sort of looks at the entire portfolio
09:53and monitors how well it's going and do
09:57we need more or less automation and do
09:59we need substantial change or has the
10:00world changed and maybe this particular
10:02set of algorithms or rules or whatever
10:05isn't really appropriate anymore and so
10:07you turn an automated system off and
10:10then finally in this you know working
10:12with machines category there's what we
10:15call stepping forward and that's
10:18developing the intelligent technologies
10:21of the president future not only writing
10:24the code for them but also in marketing
10:28them and supporting them and we know
10:31already that you know big companies like
10:33IBM we're hiring thousands of people to
10:35do this and I think there's every reason
10:37to expect that as these technologies
10:40almost every vendor will have some level
10:44of cognitive capabilities and a lot of
10:46people will need to be hired and
10:49employed for that purpose you know all
10:51these are kind of moves that you can
10:52make visa vie the machines that are now
10:55in your workplace sharing your workload
10:57so what what is it that you're doing
10:59that they're not doing a couple other
11:02ways that people can step to use our
11:04stepping kind of motif is
11:07you can step aside which would mean that
11:10you are now banking on all the stuff
11:14that is so uniquely human that it's not
11:16going to be programmed into machines so
11:18you know that may be creativity may be
11:23complex communication dealing with
11:27it may be humor it may be taste these
11:29are things that just computers are not
11:31very good at and it may be because you
11:34know it takes a sort of a human to know
11:37a human and when you have human
11:39customers a lot of times you need to you
11:41know you need to just be simpatico so
11:44there are a lot of parts of work that
11:45are going to rely on just very human
11:48strength so one example of this is in
11:51financial advising where now you have
11:53the Robo advisors and these are great
11:56algorithms for figuring out how your
11:59investments should be allocated across a
12:02portfolio of different things with
12:04different risks and different returns
12:06and you know at any point in your life
12:09whatever your goals are there's a
12:11there's an optimal allocation of that
12:13while computers are really good at
12:15figuring out what that is and it's very
12:17hard for the human mind to keep up with
12:19them on that so we talked to a financial
12:22advisor and asked him does this worry
12:24you about you know the future of your
12:27profession he said definitely I hear the
12:29footsteps behind me and already I feel
12:31like I'm spending more of my time being
12:34almost like a psychiatrist my clients
12:36telling them what the machine says but
12:39our point is that's that is extremely
12:42valuable providing that hand-holding
12:45that the client needs and that is not
12:47the part that the computer can do so
12:50that would be stepping aside I mean
12:52you're not leaving your job you're just
12:54focusing on the parts of the job that
12:56require the human touch and leaving the
12:59parts of the job to the computer that it
13:01can excel at and then the last kind of
13:04way of stepping would be to step
13:06narrowly which is to to focus on an area
13:11we you're really in such a niche area
13:15there is no compelling economic case to
13:18be made for putting it in silico form
13:20for creating an algorithm to do it in
13:23silica I've never heard that it's
13:25probably because it's an area of new
13:28discovery and where you know there's not
13:31so much demand for it that a human can't
13:36serve much of the demand for it or a
13:38small set of humans so for instance in
13:41scientific inquiry you know you're
13:43always looking for the thing that hasn't
13:45yet been discovered and you're you're
13:48kind of going into narrower and narrower
13:50and narrower niches and that is a very
13:53viable strategy for human work because
13:57only after that new territory has been
13:59discovered will it eventually move into
14:02the realm of automation a few weeks ago
14:04I included this article in our
14:05newsletter about how there's this
14:08machine learning to discover drugs out
14:11of lab notebook notes and it's super
14:13fascinating because it's an example of
14:15discovery where humans would have
14:16actually ignored it and I think it's
14:18really interesting because I used to
14:19embrace this idea that because computers
14:22automate more and more of our lives
14:23humans can actually be more creative and
14:25we can do more of these narrow and
14:27interesting things and our skill sets a
14:29psychological the emotionally
14:31intelligent things and I definitely
14:33but I also think that there's just this
14:35concurrent move where we're seeing a new
14:38kind of creativity coming out of
14:39machines that we haven't even begun to
14:41explore yet I mean so far they're just
14:42doing things that are versions of human
14:44activities like you know you know
14:45algorithms that are painting like like
14:47van Gogh or you know that's an obvious
14:48case or deep learning art I just went to
14:51an art show a few weeks ago where there
14:53was art being generated out of deep
14:54learning algorithms and it looked very
14:56cliche but it was still an interesting
14:58beginning I think it's not so black and
15:00white anymore I'd love to hear your
15:01thoughts on what happens as these worlds
15:04become muddy because I think there's a
15:05potential that computers can actually
15:07get more creative as well I was talking
15:09about this issue last night with my son
15:12who's visiting me from LA he's a TV
15:15comedy writer and we had put an example
15:18in the book of a joke that a computer
15:21had created and he said this is so lame
15:24it's so obviously programmed even the
15:28has obviously been programmed in so you
15:32know I agree that we both agree that
15:34there are more of these kinds of
15:36creative things starting to happen but
15:39they're pretty far behind humans so far
15:42and I think at some point there will
15:44probably be just a human preference for
15:47human created art and humor and someone
15:50just because it's human humans as the
15:52artisan that's taking artisan it's like
15:54oh it's artisan not AI it's actually
15:56human crafted I mean it's a relief to
15:58hear you say that I mean I'm only
15:59bringing it up because even the example
16:01of hedge-fund that you brought up
16:02earlier Tom I mean I was thinking of the
16:04show billions but one of the funniest
16:05things about it is that beyond all the
16:07things that can quantitive liyan
16:09computationally happen that a computer
16:11could easily do there's a huge broker
16:14network of information where the people
16:17are actually hubs of information flows
16:18I'm not even thinking the way a computer
16:20could do it where it would bind signal
16:22in the noise it's like you're saying
16:23Julia where there's this human
16:24interaction with information like
16:25there's a funny scene where one of the
16:27characters get some insight or trading
16:29knowledge which he didn't get busted for
16:32it because technically he made it sound
16:34like he was just like helping a farmer's
16:36daughter I mean if you haven't seen the
16:37show this puppy means nothing but the
16:39moral of the story that for this purpose
16:40of this podcast is that there is
16:42information and it's got layers and
16:44layers of intuition built into it and I
16:47think it's tough to tease apart
16:48sometimes which parts are human and
16:51which parts are machine some of the
16:54episodes but you know there's obviously
16:56a lot of psychological
16:59calculation going on it's this person
17:01telling the truth is their confidence
17:04you know just posturing or is it is it
17:07really real and I think you know think
17:09about a poker game you could obviously
17:11get a computer to easily figure out the
17:14different hands of poker and what to bid
17:16under certain circumstances but you know
17:18the whole thing about looking in your
17:20opponents eyes and figuring out whether
17:21they're bluffing or not I think that's
17:24gonna be tough for machine to do for a
17:26while let's talk now then about
17:28something that I think a lot of people
17:29linked to the AI discussion for better
17:31or worse I don't fully subscribe to the
17:33link but a lot of people make this link
17:35between the future of automation and the
17:36future and this this conversation about
17:38this need for universal basic income
17:40that people should have a fundamental
17:42basic amount of money to live on and the
17:45ideas behind it are complex and varied I
17:47personally think I mean I'm more into
17:49the idea of thinking about insurance for
17:51people versus subsidizing a baseline but
17:55that's a whole other conversation the
17:57conversation that I want to focus on
17:58today about universal basic income is
18:00it's linked to automation and that some
18:02people argue that having a universal
18:05basic income can actually change the way
18:06we think about work because if you don't
18:08have to worry about basic necessities
18:11you can then think about work as this
18:12creative act and that's why they tend to
18:14link it sometimes to this notion of
18:16automation and created creativity and
18:18I'd love to hear your guys thoughts
18:19especially having written this book
18:21write universal basic income I mean the
18:24link that people make with it is a
18:26really I guess depressing link so I
18:29would say first of all it's kind of an
18:31unnecessary link and in and it's sort of
18:34a damaging way of thinking about things
18:36why do you say that because I think
18:38that'll be really controversial for some
18:39of our listeners who are very Pro that
18:42that link well of course the reason
18:44people make the link is because they
18:45believe that automation means that there
18:49will be much much less for humans to do
18:52and that somehow we have to provide for
18:54human human livelihood because jobs
18:58won't do it anymore and so we just
19:00fundamentally reject that premise and
19:03that's kind of what our entire book is
19:05about it's showing that in fact there
19:06are lots of ways that humans are still
19:09going to be able to and be required to
19:12add value to what computers do in the
19:16workplace so we don't at all see no need
19:20in the future for human employment that
19:22every you know all work can be given
19:24over to smart machines so if you reject
19:27that premise then you don't have to to
19:30think so hard about well then how do you
19:32provide for people's livelihoods so
19:34that's why it's unnecessary the reason
19:36that I think it's actually a damaging
19:38way to think about things is that it
19:40denies the fact that actually work is
19:42really important to people it's really
19:46part of the human condition
19:47I mean being part of something bigger
19:49than yourself allowing your efforts with
19:52other people's being compensated for
19:56this is all really important to identity
19:59formation and a sense of being
20:03worthwhile and you know it's easy to say
20:05oh well we would all just have the level
20:07of self discipline required to to do all
20:11that even if our pay wasn't contingent
20:14on it or even if there was no link
20:16between effort and reward in the world
20:18but that's simply not what we ever have
20:21seen and what we know is that we
20:24actually appreciate having you know not
20:27not just the source of income but the
20:29source of structure in our lives this
20:31sort of structure that the meaning that
20:34it gives to us and you know what work is
20:37a good thing to have even dogs like work
20:40I mean that I've I've had for my dogs
20:45you know put these little saddles on the
20:47dogs because they like to feel like they
20:49have something important to do and carry
20:52around so I mean we do believe that it's
20:55certainly possible that there will be
20:57some job loss on the margins but because
21:00of this belief in the importance of work
21:03for you know meaning and and life
21:06satisfaction we'd argue for guaranteed
21:10work which would be compensated rather
21:13than guaranteed income alone and as
21:15Julia suggests there have been a whole
21:17lot of experiments around the world
21:19we're probably going to see one in
21:21Switzerland within a couple of weeks
21:23when there's a vote on a guaranteed
21:25basic income only of about two $2,500
21:28but in the experiments thus far it
21:31appears that instead of doing you know
21:33highly meaningful activities people just
21:35watch more TV so clearly I don't think
21:39that would make for a terribly
21:40satisfying life this finger finger day's
21:42watching television totally throw off
21:44our productivity numbers too which is
21:46apparently a big deal yeah no I think
21:48that's right with the key point to me
21:50and the ubi topic universal basic income
21:52topic is incentives are just completely
21:54misaligned and I think that's why do you
21:56mean the insurance angle is really
21:57interesting because it's similar to your
21:59notion of guaranteed work because the
22:02incentives are more aligned with how
22:04people are driven I will say however
22:05that this notion that that's how we've
22:08always done it that people find meaning
22:10there's a requirement in this drive for
22:11structure and work but I do think that
22:13technology and and the way things change
22:15that surprised us and we don't know how
22:18our lives will be as more even more of
22:20it as automated we could actually find
22:22ourselves very surprised by how the
22:24nature of work changes well I think
22:26it'll be very interesting if Switzerland
22:27votes this in so we'll have the one of
22:29the world's largest experiments on what
22:32people do when they can live at least at
22:35a very low level without having to work
22:37it'll be interesting to see does it
22:39lower the desire to pursue work and to
22:43see what what people do and their quote
22:45leisure time will be very interesting it
22:48will be interesting and it's always
22:49gonna be an early test study but just
22:51like with all studies when we talk about
22:53countries like Sweden and Switzerland
22:54where they're smaller and less diverse
22:56and I'm thinking of extremely
22:58heterogeneous place like the US or India
23:00or even China and you think about just
23:02layers of culture and tradition and
23:05people's relationships so be super
23:07fascinating to see how these play out as
23:09it cascades into other regions too quick
23:11last questions and in the context of
23:12this world of augmentation versus
23:15automation and for this world where only
23:16humans need apply what happens to the
23:19nature of the firm the organization and
23:21how does this change management science
23:23I mean you guys are both affiliated with
23:25MIT and Harvard Business and so I'm
23:28curious for the management school
23:30thinking of what happens on this that's
23:32right so we know all the answers on
23:33these management topics because we we
23:36hang out with the really smart people in
23:38Cambridge well we believe that
23:41organizations will continue to exist and
23:45in fact may bring more of their work
23:48back in-house than they have had over
23:51the past couple of decades that
23:53automation may take back a fair amount
23:56of the work that was distributed through
23:57outsourcing and we think that it's very
24:01important for an organization from the
24:02beginning to say augmentation is our you
24:07know our preference and our objective
24:08here in part because pursuing an
24:11automation tends to be kind of a race to
24:13the bottom in many ways that sort of
24:15lowers everybody's cost but it also
24:18lowers everybody's margins and everybody
24:20ends up doing kind of similar
24:22so innovative things letting people know
24:24that augmentation is what we're doing
24:26and I'm ideally saying you know we're
24:28not going to layoff people just because
24:30of automation that frees up everybody in
24:34an organization to think about what they
24:36could do with these new technologies and
24:38and potentially liberate them from the
24:40tedious work that we all still have to
24:43do well it's funny cuz you talked about
24:44you know the scholars in Cambridge and
24:46and one of the things that we notice a
24:48lot with startups and entrepreneurs is
24:50that the management there's basic
24:52fundamentals that are true of any
24:54company you know that have to do with
24:56profit and loss and running you know
24:58good financials and running a good
24:59organization and culture and HR they're
25:01just best practices but there are these
25:03things where I wonder if even the
25:06emotional component this reference is
25:07the point you made earlier Julia of
25:08management will change management
25:10science because will it focus more on
25:12emotional intelligence versus like you
25:14said the outsourcing model and thinking
25:16of efficiencies and cost-benefit
25:18analyses like does it change management
25:20science in that way where are we in that
25:23evolution is it still too early to see
25:24yeah no I I think you're absolutely on
25:27to something there I mean we've always
25:28known that management is kind of a
25:29synthetic discipline it draws on
25:31engineering it draws on psychology it
25:34draws on sociology and I guess you might
25:37say it has drawn on engineering much
25:39more in the past in terms of you know
25:41what is the optimal workflow and how
25:43could we possibly do this more
25:44efficiently which probably means
25:46bringing in more automation and getting
25:48rid of some of this expensive wetware
25:50but maybe now we'll see it move much
25:52more to drawing on psychology and really
25:56thinking about okay now if my true
25:59source of competitive advantage is the
26:01human element here and how well
26:03leveraged it is by these augmenting
26:06technologies then what does it mean to
26:09make somebody more human than they were
26:11before how could I enable the people who
26:14work in my organization to draw on the
26:16part of them that's really gonna give me
26:19a competitive advantage because we're
26:21now working with some stuff that isn't
26:24easy to put into code and therefore
26:26won't be in our competitors hands by
26:28next week I just want to say thank you
26:31for joining the eight 6nz podcast and I
26:33think people should just read your book
26:36Thank You Thomas and Julia thank you oh