The Uncertain Future of AI for Marketing | Operation Automation Podcast

What parts only humans can do and what Parts the machine should do better the Best results oftentimes come from Perfectly unjustifiable actions it's Hard to ask questions about the things You don't know [Music] Hi everyone it's me Anna I'm here with Eric and we are interrupting this Technology Sprint with a very special Special because a couple of months ago All of our marketing processes have also Been interrupted by the great AI Revolution we're currently living in so We figured we deserve to talk about it We deserve to address what's happening Right now and the uncertainty we're all Experiencing and we were very lucky Because Eric was able to nail us a Perfect guest to discuss it with oh yeah I'm really excited because we are going To talk to Von tan about what it means To remain human in marketing how to Build solid marketing teams and how to Use the available AI tools to produce Meaningful outcomes so enjoy this one Thank you [Music] Hi everyone really good to have you on And before we get started it was a Really exciting topic of AI during times Of uncertainty or better said AI driven uncertainty van I have a Special question for you what's the one

Thing in life you wish you could Automate oh wow uh complicated travel Arrangements and and what do you what do You mean like complicated like uh so I think complicated I I mean sometimes You want to get from you know point A to Point B to point C to point D but you're Trying to intersect Uh you're trying to intersect sort of The dates when you could go so you you Have sort of a bit fuzzy criteria for Um what days and what routes you want And it would be very useful and I think Actually this is something I would be Very good for doing it would be very Useful to see your options Um prioritize based on parameters that You give For when you might want to travel and How you might want to travel But I think this is something that we'll End up talking about today about what Things so-called AI is good at and what Things humans are good at right so yeah So things like that I think humans are Good at traveling and AI is good at Arranging travels Yes that well it's true that humans are Good at traveling but I think humans are Good at telling AI what they want AI to Do for them or at least that is what Humans could be good at Yeah so like the second freaking work is Right now I could really disagree with

All the chat GPD prompts coming up All the tutorials and how to ask exactly What you want to written or designed Like I mean stuff like Motion AI their mid-journey It's really hard to write a good prompt I wouldn't say we're good at it just yet Yeah because like you know asking Questions first of all you need to Understand Reality a bit in order to ask the proper Questions so it's not that easy it's Like for example you you usually it's Hard to ask questions about the things You don't know or you are not aware of So I I would say that even like you know Coming with those prompts It is going to be I've heard recently Because one of my friends like you know Like a month ago he joked that the Prompt Whisperer is going to be the high Demand job in 2023 and now I've heard About like prompt Engineers so like People who can engineer the right the Correct prompts so but before we Actually you know go into the like into The conversation uh if you could just Tell a few words about yourself because You know we've met like a few months ago So I I feel that I I know you a bit but Definitely for our audience you know it Would be cool if actually they Understood Um you know yorango who you are for sure

So I'm a professor of strategy and Entrepreneurship at University College London uh what I focus on is I try and Understand how to build teams and Organizations that work well and also Can adapt and can do new things even in Situations which are highly uncertain And I I wrote a book about this a few Years ago which I think Eric was Probably how we we first met uh about How you build teams that can hire people When you don't know what jobs they need To do how you can motivate people when It's unclear uh what success looks like And also how you can set goals when There is a lot of uncertainty in the Situation that book was called the Uncertainty mindset Innovation insights From the frontiers of food because it's Actually about how Innovation teams and R d uh in food work and now I'm working On sort of broadening out our Understanding of what it means to not Know something and how to make use of That to do fun stuff basically nice and This is brilliant [Music] Right now a lot of companies like they Are unforced to do some you know layoffs Restructure and actually this year Like yeah there is like a massive Uncertainty From the very beginning of the year I Can imagine and I also Imagine That

Quite a few of our listeners they you Know start their companies they think About like you know introducing Marketing teams or actually just just Developed marketing teams so here Based on your experience and knowledge What would be you know your your tip if If anyone is like thinking about how to Build a solid theme what what should be The things they they focus on you know Yeah so I think the first thing to say About building a solid team is there are Some kinds of work which are not Actually uncertain right like if if you Know what needs to be done and you also Know what the success definition looks Like I'm not sure that you need to build a Team that uh is dealing with uncertainty In terms of what it does your company May not be dealing with the broader Business environment uh as well but That's a different question right so I Think we first need to divide the Question into two parts If what you're trying to figure out is How to build a team that deals well with Uncertainty in its own work Then you have to first make sure that The work is actually uncertain Now a lot of marketing work is not Actually uncertain a lot of marketing Workers frankly it's just going to work We don't like it but that's that's what

It is now a lot of companies that have To do marketing are in business Environments that are highly uncertain That is a separate question so if you Are in an environment that is highly Uncertain I think the the one thing that If you're building a team you need to Think about is You need to build a team that is filled With people who are able to change and Want to change what they do to adapt to Business environments as they change as Well right so for example if you have a Company filled with people who only want To do the jobs the exact jobs that they Were hired for but the environment is Changing so rapidly that the company Needs to Pivot very fast so basically All startups are in this situation uh What you want is you want employees who Are able to Pivot with you right the Company now does something completely Different the employees are willing to Figure out what it is that they need to Do next and then do it Uh to do that I I think you should my View is you should probably hire in a Different way from hiring in a Conventional company where you just want People to fit a job school and then do What is in their job school for the rest Of their careers one of the threads that I hope will come out in today's Conversation is this idea that

The things that we should focus humans On doing Are the things that only humans can do Right so we should let the machines do The things that machines can do better More cheaply more quickly and then we Can we should focus as humans on the Things that only humans can do That machines I mean ideally are things That machines cannot do at all and they Get really good at doing those things And I think that's one of the that's one Of the reasons why negotiated joining And open-ended roles kind of works as Well but I think that negotiated joining An open-ended rules will make a lot more Sense After we've talked about all these other Things that I think people are thinking About in relation to work in general and Also work in the marketing space Specifically [Music] Yeah so so maybe let's start with you Know our ideas on like okay what What aspect of marketing because we Still like you know try to focus uh in This podcast on marketing for like you Know business But about rather marketing so actually What does it mean to be human in Marketing and you know what are the the Elements what is the What do you think about this because for

Me This particular podcast episode as a Kind of context for us to to have this Meaningful talk to exchange ideas to you Know like first Network like get to know Each other and then this this is cool I'm not afraid that you know any AI or Like any technology will just feel the Need to do something like we Uh because I'm excited and I believe That this feeling is also something that You know is typically human Uh but it's also like business related In this context and I think we're making To like a really cool Polar Opposites Here because I am terrified I think Everything we're doing can be replaced By until one I would appreciate it if You like put me at ease a bit uh I don't Know if I can put you at ease but I I Can share a point of view right so I I Used to work in marketing as well so When I was at Google one of the roles I Was in was I was working in marketing in So-called big Tech Um I think that there's really sort of The I I'm going to oversimplify but it's Over simplification so that we can see Something more clearly uh in marketing I Think there is there is There are basically two different kinds Of work that you need to do and they're Both important but they're two different Kinds uh one of them is figuring out

Where you're trying to go And then the other one is actually going There right so the the analogy back to Travel planning is the same like I can Figure out where I want to go and about When I want to be there and then what I Want is I want a very very good Whole Market sourcing travel agent to do The planning actually for me and tell me What the optimal thing is given what I Want I think for marketing you've got a very Similar situation like you've got a Product and You need to do two things first you need To figure out who is the Right audience Or buyer of the product and then after That once you've figured out who the Right buyer of the product is you need To figure out how to get it to them In doing all these things I think Humans are the only ones so far who can Figure out Who can make essentially a judgment call Right like they are the only ones who Can say it may seem as though it's Equally good to go after this particular Market let's say B2B clients in North America who are uh small small and Medium-sized businesses or we can go After that market which is uh B2B Clients in North America who are like Between this this quantity and sales Every year and that quantity and sales

Uh I think a human can look at a bunch Of information And make a decision based on that but Only a human can do that right so it's Not I don't think this is yet a question That you can ask an AI maybe in the Future we can but at the moment we can't Now the other the other part the other Big part of doing marketing is a lot of The stuff that marketing teams end up Doing today like things like uh creating Assets and managing them and deploying Them in the right way in campaigns uh It's like doing a lot of the basic data Analysis to understand the effects of Campaigns and then deciding what to do With campaigns after that I think one of The one of the big challenges for us is All of the work that we have done before AI Is a combination of stuff that only Humans can do And stuff that machines can do more Quickly and that's why we're confused at The moment right it's very hard for us To tell what parts only humans can do And what parts the machine should do Better because until I would say probably Eight to ten years ago we were doing all Of it and then over time we gradually There's been more and more automation of Stuff I mean the company that you all Build is one of those things uh and so

As more things get automated to Better Or Worse degrees of efficacy we are Gradually learning what things are Things that only humans can do and what Things are are things machines can do Better and I think the first thing Anna To uh assuage your fear is I think we're Worried because we're not clear about What things only humans can do and Thinking a little bit more about that And testing those ideas Against the Machines I think is a way to get uh to Get started [Music] I totally agree with you and for example The concept of creativity this is Interesting because in marketing we tend To Let's say I think about copywriting and content Creation and in content creation and Copywriting you have this you know Creativity element which means that you Can understand as a human you can Understand this idea you know that this Idea is quite complex because it needs To represent your business needs and you Know your customers information needs And something and and then And then only you as a human can come up With this like slogan or this one Sentence that actually is relatable uh But now it seems that if you have this Proper language model

You can have like you know you can just Like in a snap you can have 10 of them And Yeah and suddenly you know is is it like Real creativity or is it just like a Proper language model that you acquired It that this is like some language Proficiency so I know what is creativity in marketing It's a great question and that's Actually a great example right because Um you know earlier you were talking About prompt engineering I think And this is just my own view I I don't Know this is correct but this is so far How I think of it Uh even given that example you know you Need a tagline and you ask a language Model to give you some options You have to prompt it to tell it what World of possibilities it needs to look In in order to give you something back Right and when it gives you something Back it gives you 10 options which Uh there's there's a few possibilities For those 10 options some of those 10 Options may be things that you as the Human had not thought of before and you See them and you're like oh yeah this is Kind of a cool thing Uh or all 10 of them might be things That you've thought of before and they Are different expressions of things that

You as the human would have also done But in either case there are two things That so far only humans can do properly It may look as though machines can do Them but only humans can do properly the First one is only humans can decide what To ask the machine Right that's the problem part and more And more people are realizing that the Prompting is the really difficult part Of interacting with the AI you know you Can interact with the so-called Ai and It can give you a bunch of garbage back Because you didn't prompt it correctly Or if you prompt it correctly you can Get something really useful so clearly What this means to me is that there is Some meaning making that is going on When you're prompt when you're writing Prompts and that part so far we have not Yet been able to figure out how to get Machines to do it properly Um we'll come to the edge case where They look like they're doing it properly Later but we we haven't yet figured out How to do it properly the second part That is very human is in deciding which Of the 10 is actually good Right so you you get 10 15 20 100 Responses back and you look at all of Them and some of them will clearly be Better for what you the human want them To do than others and in the end it's Still you the human who is putting a

Layer of meaning on top of the response That comes back from the machine right And if you are not good at doing it You'll pick the you'll pick the wrong Ones as a human and if you're good at Doing it you'll pick the right ones so I Think what that example is beginning to Show us is there are some people who are Good at Using the machines to get a good set of Possibilities from which to choose those Are good prompt writers And then they're also going to be some People who are good at choosing from Among what the language model or the AI More generally is giving back to them Right they're the they're the Discriminating curators I guess you Would call it and both the prompting and The curation at the moment these are Things which if you think about what Prompting actually is and what curation Actually is in this context are things That machines can't do yet right and I Think the same is true if you talk about Writing and creativity I tend not to use The word creativity so much because There are some aspects of marketing that Are really not about being creative Right there's some aspects of marketing That are about uh for instance doing Very good quantitative marketing Begins with having good assets that You're going to test in Market against a

Particular success metric but most of Really good quantitative marketing is Making sure that you set up your Hypotheses correctly so that you're Gathering data that allows you to do the Analysis of whether or not campaigns are Working as you want and they have the Efficacy that you want uh part of it is Creativity the part where you come up With something which is new and then the Other part of it is about reliable Execution And I think we can separate at the Moment what humans seem to be good at as Being mostly connected to coming up with New ideas as well as Thinking about whether or not an idea is Good right those are two separate things One of them is creativity the other one Is what I would call more generally Meaning making and then the execution Side of things I think is where machines Tend to outperform us for good reason And I think being able to separate the Two parts is really important in Assuaging the fears of people who are Worried that they're going to be Completely replaced by machines very Very soon They have a really cool story to Accompany that because I absolutely Agree with what you just said and again I am on the side of people who are Scared about being eventually replaced

But I had a super lovely day spent with Chad GPD when I was editing the text and The curation part of it was actually Really fun it was the most fun I used I Had was the text in a very long time Because I was able to use it to Take over the procedural operational Stuff well I could actually focus on the Creative side of things so it was able To provide me with examples if I needed Some real life applications I didn't Have to spend hours on doing research And connecting with my target audience Instead you could just say hey Chad GPD What's the most common struggle Marketing agency faces and of course it Needs to be verified still because AI is Nowhere near being an absolute source of Truth but it already provides you this Framework that speeds things up a lot And it made the whole experience much More engaging for me as a writer Hopefully it also made the output much More valuable for the end user the Reader yeah for for sure and I think That example is a is a good illustration Of how we can Start to distinguish between what humans Can do and what machines can do right What machines can do at the moment this Is not a complete list but certainly What they can do is they can do things Where you can describe the procedure

With Great precision and complete Specifiability for instance if you ask Chat GPT to replace so I I used to be a Copy editor as well and one of the Horrible things would be running into a Manuscript where for instance they use n Dashes in place of M dashes Now I used to go through this and I Would write like regular Expressions to Like fix all this stuff now you can ask Chat GPD to do it all for you right so That's the very basic it's the most Simple and I guess clear example of a Task that is fully specifiable that Really why are humans doing this it's work right so get a machine to Do it perfect now If you go on all the way to the Other Extreme and you say we want to create a New campaign And we don't simply want to go with Things that Um We don't want to go with we don't want To go with What existing competitors in our space Are trying to present this category of Products as What do we do Okay so we can ask a language model what Existing competitors in the space are Positioning the product as so that we Know what not to do

But the question that precedes it and This is sort of the analog to The Prompt First as humans we need to decide we Want to do something that is different From what everyone else does right so Far this is not something that machines Can do yet simply because of the way I Mean if we're talking just about Language models this is not the way Language models are built language Models are built essentially so that you Can understand what is more or less Common in the Corpus on which they were Trained and so if you know how to prompt You can still find out what you want to Know which in this case is what is not Common or what has not been done in the Corpus you can ask it that but in and of Itself this is not something that an llm Wants because one of the things that we Know is llms can't want things on their Own yet only humans can want things on Their own So so yeah I I think it's a very good Way to think of it Um and it's like thinking of it as a Tool That you have to decide that the Deployment of But it's a very sophisticated and very Flexible tool if you know how to use it Right which is to say it is like all Other tools that humans have ever built It's exactly the same as a knife a good

Cook who knows how to use a knife Properly can achieve things that a cook Who has never been trained in knife Skills cannot do Uh in the same way I think we should Think of right now I think we should Think of these machines as tools and we Should learn how to use the tool Properly and this is not to say that They will only ever be a tool maybe in The future maybe even soon in the future They will learn what it means to make Meaning themselves but so far as far as I can tell they haven't done that yet [Music] I'm also hoping that maybe this Interaction with chat GPT for example Like you know because there are like Some Alternatives but maybe this will Also be helpful To actually learn how to interact with Customers more than actually that you You start at some point you know to Prompt people then you wait for the Feedback for the response and then you Kind of like curate it and you just Direct the feedback to the product team To the marketing team you actually Make me I'm thinking about that because I've seen I've seen like a cool change in our team Because it used to be Constantine now it Is content in Partnerships team and I've Seen that Partnerships play bigger and

Bigger role in the whole process Also like I used to be focused on like Marketing or okay I used to be focused On on the product and looking at it you Know from the marketing perspective so We would as creative people as creative Human beings we would like invent or Like to like try to imagine what would It be like if we ran our business and we Use the product But then I just figured out it's so much Easier to actually ask customers about That and then it's like like let them Tell me what is the value of a product It's like tell me what what this product Is for you like you know how do you use It When is it good when is it bad and then I realized like wow this is this is the Angle I've I've been looking for that That suddenly everything is easier Suddenly the answers like when you said About the curating so you know I I put The prompts and then I get so much Feedback that I need to be careful about How to curate it you know because some Of it is going to be Actionable some some maybe not so still You need to be careful but but Connecting like being able to actually Reach out and and create those feedback Loops with your customers I'm hoping that maybe The the way we use chat GPT we're also

Maybe help us to Yeah like you know try to form Meaningful connections with customers What do you think about that yeah so the I I think so it's an interesting uh it's An interesting Um hypothesis right like One one possible reason why the uh so Machine learning has been around for a Really long time and in fact it's been In the infrastructure of a lot of the Services that we use and have been built That have been very successful but I Mean if you think back to before maybe Like 12 months ago that the big wave of People getting excited with AI it came I think anyway it was highly correlated With maybe not caused by but very Correlated with The emergence of a form of interaction With the so-called AI that looks a lot Like things that we already know like so Asking Google for an answer I mean essentially when you query Google For a thing that you want Uh there are some similarities between That interaction you know I asked for Something I want Uh An algorithm goes off in the back well I Mean it has already done it and Algorithm goes off in the back and Harvests a predetermined answer to come Back in some ways this is almost exactly

What is going on with things like chat GPT right So I I wonder whether Using chat GPT Will do what you're suggesting or if we Will realize that actually I think the Where I probably end up with this I'm Just thinking in real time Uh where I end up with it is I think a Thing like chat GPT is different from Google because it makes it more obvious That when you ask the right questions You get better answers right this is the Discovery that prompt engineering is an Important thing you can't just like Randomly ask it a question and hope to Get a good answer you can design the Question well and get a better answer Back and you know I I'm an ethnographer By training and I teach interview Methods right so one of the things which I I know from my previous work which has Nothing to do with asking an AI for a Bunch of stuff is that there are ways of Asking questions that get you more Informative answers and there are ways Of asking questions that get you answers That you expect to get and the ways of Asking questions that don't get you Answers at all right so knowing how to Ask a question I think is something Which good questioners have always had In the past and now maybe watch things Like chat GPT will do is it will show

More and more people that this is the Case which is which is a good thing um So human centered design is basically uh Is a way of approaching design that Takes this approach right like it it Says you want to understand how to do a Thing and design a product design a Service you can do some market research But the most important kind of market Research is asking the humans who might Actually use it what it is that is the Problem that they're trying to solve and Figuring out what what kinds of Solutions they might be interested in Using which is basically what we could Do if we were interacting with chat GPT In a particular way or any other kind of Like conversational front-end to a Back-end large language model Thank you [Music] I mean here I'm also kind of curious Because prompts are usually more than Just questions right especially if we're Talking about image generational Problems so there is a certain level of Coding event involved where you have to Create all the parameters in order to Provide those parameters you need to Understand them Um for example like the way I approach Um mid-journey would be just to say Draw me this which can be I don't know a Husky on a snowboard

I would end there like I wouldn't use Any professional you know artistic Lingual I wouldn't use specific Styles Just because I don't know them and in The end I would probably be satisfied With my husky because I wanted it to be More Modern or not and that's why like or Impressionist or like whatever right Exactly So you need to train not only to ask the Right question I think which is Undoubtedly highly important but still Learn the field you're asking the Question about I hope I got my point Across here for sure yeah no no and and Actually I I think those two are In a way they're both the same thing Right so you know to use the mid Journey Example if you're trying to prompt it to Create an image that you want in a Particular visual style or with Particular elements inside it It's I think what a I guess a conversational Interface with a language model does is It shows that you actually need to Understand the context in order to get What you really want because for Instance if what you actually want is You want like an impressionist Style If you want an impressionist Style with A anime style dog on a you know like Impression style whole image within with

A anime style dog on top of a brutalist Style snowboard if you don't know that Those are if you don't know that those Are the labels that we have in the past Attached to those visual styles of Representation you will actually find it Quite difficult to generate the kind of Output with a prompt right whereas if You did know it if for instance you were An art historian of modern you know World art I suppose You would know those things and you Would be able to attach a word To the visual effect that you want And in fact text to image is Meant to reverse that kind of Association because the way it would Work by the way if you ask for an Impressionist style husky on a snowboard Is You would essentially have to rely on The fact that in previous times in the Corpus that this model was trained on People use the word impressionist or Stick or words close to impressionist to Describe images that looked Impressionist Because without that the model would not Have the connection between the two that Was necessary for your word Impressionist to create an impressionist Seeming image on Um as the result so I think writing a good prompt is not

Only knowing what you want In this case I think what we're saying Is writing a good prompt is knowing that You want something which in your mind Looks like an impressionist husky on a Snowboard for example and also knowing The context enough that you can use the Correct label So that the prompt does what you have in Your head if you want to think of it That way that that's exactly the same Problem that uh like for a very good Artist who knows how to use their tools Really well They have an image for instance they've Got an idea of what they want to create As a sculpture out of marble And if you are very sophisticated with Using your chisel you can make it look Like what you want I mean the analogy Back to for instance just using an Example that everybody knows about Because most people have coat I suppose Is if you know how to use a knife really Well you can do things that people who Don't know how to use knives can't do Right like um the example that I love is Uh katsuramuki it's uh it's where you Take a cylindrical vegetable and with a Knife that is usually quite straight on One on its Edge you peel the cylindrical Vegetables into a very long sheet and Then you cut it into like needle fine Like almost like Juliet

If you know how to do it You can totally do it if you don't know How to do it you can have exactly the Same cylindrical vegetable and exactly The same knife and you won't be able to Do it like in replacement of the knife Have language model In replacement of what you're trying to Do have a prompt and then if you are Good at prompting the tool you can get Customer if you want you can't I love it very good analogy so it when I Was like listening to I So now I just think about myself and then if I Thought that in order to be able to Produce better prompts I should actually Think about experiencing more so like You know the goal will be to understand The impressionism the brutalism the you Know the anime the anime or basically Being out there and experiencing life Converts into better prompts and and Also like because before um our meeting I read your article uh like you know on Being human and I even like wrote down This this because it's a really cool Cool sentence like what makes us human Is not just how we interact with humans But also how we interact with ourselves And this is also like like you know it Just blew my mind because it's true that Depending on who you're talking to

The conversation is different the Questions people ask each other are Different because we focus on different Things because we experience life in a Different way So yeah it's it's I wanted to to ask you About this like Did you Um yeah how do we actually make better Prompts and where do better prompts come From yeah okay so I I'm glad I'm glad we Got to this question I I really wanted To talk about it So that article that um that you just Mentioned Iraq is really trying to ask People to think about the problem of So-called AI in a different way right But so a lot of people think about AI Today and Um there's two major sort of themes one Theme is AI will solve everything And the other theme which is sort of the The mirror image of that is AI is going To create all these problems for us Because it is going to take away jobs Things like that Uh it feels to me like actually there's A different way to think about it that Helps us understand how it can help us Do things better without also Everything up and that way is to ask What it is that and this goes back to The question I posed at the very Beginning what is it at the moment that

Only humans can do that the machines Cannot do and at the moment it seems That making meaning and we'll talk about What that means in a little bit making Meaning is still one of the things that Only humans can do that machines cannot Do okay so what is making meaning and This goes I think to your point about uh Interacting with others and interacting With ourselves When we are interacting with others we We have to understand what their context Is so that we can say things to them That make sense for both of our contexts Right like if for instance you um So so there is another good example of Of giving giving gifts is a social Interaction you know you you give a gift So that you as an expression of regard Um to show that you care all those Things but giving the wrong kind of gift Can be really bad right so for instance If you went to let's say you go to Japan And you give someone a clock That it's it's not a good gift uh in Some in some respects because it's um it It's a an indication of impending death For instance so understanding context Allows you to take an action in a social Interaction that either makes sense or Doesn't make sense or has the wrong Effect But when we're thinking about Interactions with ourselves

That's when we start to think about The sort of meaning making in its purest Form When you're interacting with yourself You don't have to you don't have to Think about whether or not someone else Cares about something that you've Decided you could in theory just care About what you think about it so for Instance I I've just been drinking some Coffee And I thought today that this coffee Doesn't taste very nice Now even that simple decision that I Don't think it tastes very nice today it Tastes different from yesterday On the one hand it's a it's a sort of It's an assessment of a fat of some sort Right like the coffee today does taste Different from yesterday But I've put some meaning on it I've Just decided I like it I don't like it And that is the kind of meaning making I'm talking about at its most basic Right it is deciding on something even Though other people might decide on Something a different way that's just Meaning making now it can go all the way From deciding that coffee is good or bad That's meaning making to deciding that It's worth it for my company to invest In building a phone that does more than Any phone on the market does at the Moment which is what Steve Jobs decided

To do when he did when he decided to Invest a lot of time and money in the IPhone Those two acts the decision on the Coffee and then the decision to make a Phone that is a real smartphone the very First one they both they look very Different but they're both decisions to Do a thing because you have put some Kind of value on it that's what meaning Making is And in our interactions with ourselves That's the first time that we sort of Have a A clear and unpolluted view of what it Means to make meaning So right now as I I keep on saying I Don't think machines can do this yet Maybe in the future they will be able to Do it but our ability to decide that Something is good or is bad or is better Than this other thing that is the source Of our ability to make a decision about Whether to ask for a husky on a school Board in impressionist style because That's what we think is better than a Husky on a snowboard in anime style Which we think is not as good because Otherwise what is the reason for you to Ask for it in one Style versus another Style And let's not even talk about the style Why are we asking for Husky on a Snowboard we like it more than a cat on

A snowboard right as in when you say I Asked for this because The thing that comes after because is Your reason for doing it and if you Really dig down deeply Eventually all humans have a subjective Reason for asking or wanting to take an Action in the world and that's the Meaning making part Right we often disguise it with a lot of Justifications but in the end we only do Things because we've decided that they Are worth doing or that they are worth Doing more than other things that we Could also do And that is the meaning making part of Human of Being Human right now machines Take instructions from us because we Decide what is worth doing and the Machines do it for us which is fine That's actually at the moment anyway Where I think machines should be because Machines are still tools they have not Yet become agents like you know agents With the ability to decide what to do They may be agents that get instructions And do things on behalf of real agents Like us And maybe in the future they will uh So-called wake up and become truly Agentic right like we get to AGI that Might be cool but also would be Terrifying I'm thinking this is such a Philosophical direction that most of

Your actions need to be guided by this Why am I doing this like this higher Purpose because coming back to the Husky On a skateboard example Because it's fun like that everything Needs to have a reason and I think this Is my problem with approach to marketing We try to over justify things when Sometimes it's just about trial and Error and seeing like some of the best Marvel projects became viral not because Of Um extensive research supporting them But just because someone was having fun And ended up creating a thing that other People found fun and I think this is the Human Excellence bit when the best results Oftentimes come from Perfectly unjustifiable actions when you Just feel like something good Well I mean if you if you think about That example you know like why a husky On a snowboard because it's fun uh you Would have to go like deeper back into Understanding why for any particular Person the things that they find fun are Fun right because a huskyo is no more For some people who like huskies and Snowboards I suppose that's fun for Other people who don't like Huskies or Don't like snowboards or don't like Huskies on snowboards it's not fun and You know one big part of marketing is uh

Identifying a coherent population of People who all want the same thing even Though you may not Even though it may not be obvious yet so That you can activate all of them with a Campaign or with an asset that gets to All of them at the same time I mean this Is a even the ability to interpret what Is fun or relevant or useful Uh for a group of people that is a that Is a gift that really good marketers Have which is a human gift like you Can't ask an llm that at the moment Uh what you can do is you can ask it to Do a segmentation based on for instance If you have a hypothesis as a marketer That the way people use words to Describe a problem is how you can Identify a segment That is something you can ask a language Model to help you do right because in The past you might have to do very Complicated field exercises in order to Get that information out today maybe if The Corpus exists and the Corpus is what The language model has been trained on You can ask a language model Um show me all the words associated with Describing a particular product in this Way and you will have done the research In a way that is probably more cost Effective probably quicker but Ultimately you are still you the human Are still the person that has decided

That this is what is worth doing and Then the language model is giving you is Is your tool in doing it [Music] Because we're running up close to the End of the episode I wanted to ask About the uncertainty we are all Currently in so as we've kind of touched Upon in the beginning you know I'm more On the scared Edge Eric is more on the Optimized excited edge of it Um do you have any Predictions Maybe Based on your experience with uncertain Times as to how this General sentiment Towards AI application and marketing Could evolve I I always say that Prediction is I I usually don't predict Because Things never work out like that Um I I think I I will I will say some Things about what seems to work best What's what things seem to work in times That are highly uncertain so so maybe Before I say that I'll say something Very quick Uh there is a difference between what we Think of as risky and what we think of As insert right like we all we often use The two words more or less Interchangeably but A risky situation is one in which you Know all the actions you can take All the possible outcomes that could

Result and you know the probabilities of Any Given outcome given and given any Action I think that's Almost universally not the case for us Now like we're actually facing true Uncertainty where we may not know all The actions that we can take you know we May not know all the outcomes that are Possible and we certainly don't know all The causal connections the probabilities Between given actions and given outcomes So In this kind of situation The first thing that seems to work Better Is People and teams and organizations that Are clear about Facing a situation so I now call them Situations I'm not knowing and there's a Reason for that If the person or the team or the Organization is Clearly identifying that they're not in A risky scenario They tend to act in different ways right So if you're in a risky scenario what You do is you risk you risk manage and You try and mitigate against expected Outcomes so that you you think you do Things like buy insurance against Expected outcomes that are bad If a team or a person or an organization

Is facing a situation that they Recognize as being not risk but also Filled with not knowing of some sort They tend to behave differently in Several different ways they are useful For them They tend not to take very big bets they Tend to experiment in small ways Right because you you don't take a big Bet because you don't know what to bet Big on you think there may be lots of Small actions you can take and you want To figure out what those small actions Will do so you take lots of small Actions that are very different from Each other and you learn a lot along the Way so In a sense experimentation On lots of small things is something That Especially teams and organizations and People facing non-risk not knowing they Do and that helps them figure out what To keep on doing Another thing that seems to work for These kinds of organizations is that They hire people who are able to change They are more comfortable with being in A situation where they don't know what To do and it turns out these people are Actually quite rare right most people Are trained to be good at doing a Particular thing they become comfortable Doing that particular thing

And when you take away the certainty of This is what you have to do this is your Success metric most people are not very Good at it the people who are good at it Are very valuable because they can Change what they do they can adapt They're a little bit like a Swiss army Knife rather than a knife And teams and organizations that Specifically look out for these kinds of People they also tend to be more Successful now looking out for these Kinds of people people who can change What they do as the needs change that's Actually what negotiated joining and Open-ended roles do like to go back to What we started this whole conversation With and that's why I said maybe it Would make more sense to talk about at The end because if an organization is Facing a situation where they don't know What is going to happen next but you're Not in a risky scenario really the best Resource that you have is having an Organization that is full of people that Can figure things out along the way and Change right and the way to do that is To hire in a way that I think of as Being fundamentally different from Conventional hiring And the way I've given it a name just For convenience uh it's about Continually changing roles that you Negotiate with your organization and

With your other co-workers because the Needs that you are trying to fulfill With the work that you do are also Changing over time and you know to come Back to what's happening now in the World of marketing and more generally in The world of work if you see that There's a tool that's taking away a lot Of the work that people in this industry Traditionally do like a lot of the work Of asset management in marketing is Frankly something which I think a Machine can do better but what this Means is all the people who spend a lot Of their time managing assets they need To figure out what it is that they can Do better than the machine so that they Continue to add value and in fact they Can increase the value that the person Plus machine combination can bring to The organization or to the industry Right and in fact we've talked about a Lot of that um today if Asset management is just about you know Creating 45 colorway variants of an Image of a shoe or a shirt A machine can do that better Now what the what the asset management Team needs to do is they need to figure Out Uh what are the color variations that You want this you know set of assets to Have they need to figure out questions Like what are the hypotheses we want to

Test with the campaigns that we're Building with these assets these are Things that right now you can't you Cannot ask the machine to do it for you You can ask the machine to help you do Them I mean you can ask it for instance Um I want to run 10 different campaigns Give me 10 audience segmentations based On the following criteria this is Something that machine can probably do Better than a human in the sense that it Can do it more quickly but the human Still needs to ask it the question so That the machine can do it totally I Feel it like you know now it's it's more About like experiencing a certain Challenge and then coming with different Ways to even like you know explore it And and test it and just figure out you Know how to deal with it so definitely As you said about this negotiated Joining like like you you certainly Search for people with Autonomy and actually you know people Who are able to figure things out and Actually even now with The the number of like tools that we Could test you know this is Extraordinary huge number so actually we Need to be able to choose like like the Kind of like let's say top five or top Three or like top two and then just like You know run with it for the for a short Period of time learn share like spread

The the outcomes and then like yeah go From there and and then explore further So definitely I I feel would have what you've just Said yeah just one thing to say by the Way I I don't think it's just about finding People who are able to adapt that's Incredibly important as well but even The way you hire people also needs to Change right like inside the company if You think about the traditional way you Hire people you write a job description You go out you look for people who match The job description and then you try and Hire people who are the best match you Use interviews and like aptitude tests And things like that to do it Um I think what I'm proposing is Something fundamentally different it Requires changing the way you think of What it means to join a company it it Means saying part of your role is not Yet set in stone you figure it out along The way and you show us what benefit That brings to us as a company and That's the negotiated part of it it Starts with the assumption that your Role is not fixed and what that means is It forces both the team or the Organization that is hiring and the Person that is joining To rethink what it means to work right Like it means always having a little bit

Of not knowing about what it is that You're doing so that you're always Figuring it out it's like this very Dynamic sense in which there is no Stability but that lack of stability Causes things to happen that are good For the organization and the person Involved sure and I guess that it also Means that it's Ah very important to understand like Yeah what what we do and why we do it And then how it just just we will figure It out but I think that It's it helps uh I mean Especially in in times of uncertainty if You at least know why you do certain Things That helps a lot for sure and and the Only people so far who can say why you Know like doing this is good doing that Is bad doing this is better than doing That are humans because answering the Why question is always a meaning making Exercise right It isn't a statistical exercise It can be that you say if 51 of people Want to do this then I will also think It's good But even when you say something like That that's still a meaning making Exercise that is on top of a statistical Exercise that you have chosen because You've decided that that is the way you Will decide whether or not something is

Good [Music] Operation automation is backed up by get Response the marketing automation Platform that's been on the market for Over 20 years subscribe for more juicy Insights and remember automate don't Complicate

You May Also Like

About the Author: freeautoresponder

Leave a Reply

Your email address will not be published.