The Rise of AI Storytellers: Will Humans Become Obsolete?

· 20:49

Notes & highlights

Andrew Reid and Kris Krüg outline a four-round data-storytelling experiment designed to test how survey findings can become memorable creative formats. Reid says the idea grew from the mix of creative, academic and corporate perspectives he saw at Vancouver AI, then became an open-ended challenge built around a proprietary dataset at 00:37.

Reid reports that they had agreed to run four rounds and had heard that more than 40 people or teams signed up for the first survey, built around the question of whether a hot dog is a sandwich, at 01:47. Krüg frames the challenge as a search for new ways to tell stories with data at 02:15. Reid contrasts that goal with familiar slide decks and dashboards, saying he hoped participants would produce something unexpected and inspiring at 02:57. The signup figure is his dated report, not a verified outcome.

Turning to company practice, Reid says Rival Technologies created a lab so teams could test ideas outside its usual software-development process, sometimes moving from an idea to a prototype by the next day at 04:40. He also describes a first internal AI day with no other agenda, intended to review the month's tools and learn through experiments, including failed ones, at 08:48. Those company-process claims are Reid's account in the recording.

Follow the current Build Night, the Vancouver AI community, or more talks in the video archive.

Transcript (auto-generated)The full spoken text of this recording

YouTube generated this transcript from the recording. Kris Krüg checked the names in it against the recording. The wording and punctuation are still the machine’s, so quote from the video itself rather than from this page.

0:00hey what's up internet it's Kris Krüg I'm here with Andrew Reid and uh yeah you know I've been hosting these Vancouver AI Community meetups and uh that's given me and Andrew a chance to to reconnect over the last little while we've known each other in the Vancouver Tex scene for a good long time but um I'm going to start this off by jumping right into the um storytelling hackathon the data storytelling hackathon that we're working on and then I'd love to you know like just get to know you a little bit more and what you're doing around here you're just telling me about AI day here at your company and uh even with a little bit before I cut you off I

0:31already got questions and stuff but um hey what what's up with this data story telling how Canon we're doing yeah I mean I showed up at I showed up at one of them in the fall and then went to another and I had this sort of interesting moment of maybe Clarity where uh seeing the diverse group of people that you're able to curate where you've got you know hyper creative people people from Academia you know people from the corporate world you've got this mashup of different kinds of people and it really reminded me of the mid99s when I went to film school and the internet was just getting going and so I just kind of felt like what

1:09interesting viewpoints these people would have and so what if we and I'm in the market research data collection business so I spent a lot of time helping Brands um understand their customers better and so it just sort of started with like how are we how could I get more involved in this group and what if we could um find One Singular kind of you finding piece which is data so if we can provide a proprietary set of data and then give people a challenge it's very open-ended um and I think when you and I chatted you added on I think the idea of saying well let's let's do this a few times so it's not a oneof and so

1:47we've agreed to do it four times and uh and we have kind of no expectations but it's really exciting because it's it's you know I think I heard that we have over 40 people or teams that have signed up um the data that we're providing is is a hot dog a sandwich we ran a study to uh I think somewhere between the old classic question put it to bed huh I mean I Got My Views but I want to leave the witness on let's see what the people have to say no so it's so cool so yeah we've assembled this pretty sweet community of AI hackers and Builders and experimenters and researchers and stuff and so we're launching this hackathon

2:24into this community where you're like hey what uh what are the new ways to tell stories around data so we do research in these traditional methods but there's all these new from train a model on it to make a I mean everyone's done everyone's done that there's the Deep fakes there's the make this into a podcast and I find the podcast is all the ones I've played with are like slightly too cheesy that's why I'm thinking that we'll probably get some riffs on that I don't know maybe people are going to do some spoken word thing or turn into a song I have no idea I think the the reality is what happens today when you you'll do a study you'll

2:59get your data back and and then what you build a PowerPoint so either one or two often it's a PowerPoint and it's a 50 slide PowerPoint and you're like here's we asked a th people these 10 questions that you care about a lot and here is what they said yeah and you're Crossing that data a bunch of superly cross tabs you're cross referencing other data points we'll cut this by age and income and all these other different demographic attributes that and then obviously dashboarding Tech like Tableau we have uh an integration with a piece of technology called cense that we have in our you can these beautiful custom dashboards we have some cool little mobile Topline views and then we have

3:35some of our own AI tooling but it's all very much what you would expect and so I think what we're hoping is that we'll see some things we don't expect and I'm not you know my my uh goal is just to get inspired I don't think we're looking to do any more than that I think that's what's kind of fun about it is that we'll see where this goes and whether we can um whether we learn something maybe we'll put guard rails on for phase two three and four maybe we'll leave it wide open I know we're going to have a new topic that'll be a new subject matter maybe people could suggest uh things to

4:08RI on in addition to as a hot dog a sandwich yeah um we maybe should do a poll on that when we post this you mentioned like you know we got some of our own AI tools and I know this is just something that's you've been consumed with lately just trying to understand the space and um how you can use it in your own like CEO life but also how your company can integrate like new ways of thinking and methodologies around MLA even just like storytelling stuff um talk to me a little about your AI Journey you started like a lab part of your company that's doing some AI stuff and you mentioned AI days around here we

4:40really got going in 2018 and um so we have a software company called rival Technologies and we have a services company called reach three insights um what was interesting is that we're not that old we're seven-year-old company but I felt like if we're going to drive on this AI bus we need to like be in a place where we don't have to have the same process you would traditionally have in software development meaning having to like ideate and then write stories and then groom those stories and do spikes and it's like I find in this world we'll sit down with an idea and then the next day we're looking at an actual live prototype of that idea and we've skipped

5:19all those pieces so the lab's idea part of it was just to make sure that we would not um mess ourselves up or slow ourselves down and we have that freedom it also allowed our security compliance team to be cool with us experimenting with stuff and not having to be like always in there making sure everyone's wearing their seat bels uh because this game's going so fast that I don't think that's the right way to go have you um have you thought about or experimented it all with like synthetic audiences uh yes thought about it a lot I researched it a lot and you will likely be coming up with a synthetic agent I think the

5:57first thing we're going to do is build our own that we can use uh to probably test the marketing campaigns and to test some right uh some pieces to get Mark just so that we're going a level set here make sure we're all talking about the same thing when you say synthetic audiences so you what you take like I don't know your customer stories your customer profiles from here and then build out an audience of a million or 10 million people that meet those profiles exactly and then uh develop a bunch of marketing campaigns or have it the AI itself develop the marketing campaigns test them into these audiences learn some stuff from it maybe even like

6:29iterate on the campaigns and keep testing and get like that that's where it goes exactly so I think if you think about often you have some course segments that you Market to um and so we think about like our outbound motion uh we can create these synthetic uh these synthetic personas that fit within those segments and then then we can test out what would happen if we push out a campaign and how many people respond to that yeah and we'll do that using a lot of the exhaust data that we're getting off of the past campaigns that we we run we'll do that using uh data and markers we get from customer satisfaction surveys that we send out U that kind of

7:07thing then obviously in our industry because we're in the research industry like building personas that our customers can use is is very is very cool I mean there's some cool ideas we have that we're not quite ready to talk about but that I think soon we'll be able to show you well um there's a researcher in our community I think you heard her talk her name is Mariana Papa and she uh sheach teaches at North Eastern University and she was talking about synthetic personas not just in a way to go scale like we just talked but there's audiences that one might want to survey or poll or ask questions to do research about that like either can't

7:43consent because maybe they're too young or too old or maybe they're an at risk population or something like that that doesn't want to necessarily like out themselves you know like I was at that when we did the autism when she talked about the work that she was doing with people with Autism and I thought it was really really incredible this idea that if you are in therapy or if you're trying to create programming being able to understand how um how that audience might react and like that's a without having to uh you know because there's obviously barriers to interacting with a million autistic people or something like that you know Stu so I I thought I

8:16was that that really uh really hit me in a lot of ways because I thought was like what a great cause and it also starts to make things click and I think that's probably the magic of what you're bringing I think to this community is those stories and those stories often that you know are really way adjacent to maybe what someone else in the audience is looking at but maybe they have a family member or someone else or someone dealing with cancer or whatever now knowing that this new technology can play this huge role right in making letting people live like a better life yeah let me finish that last point about I was going to make about um our AI day

8:50yeah so Friday is the first experiment and the realization I had is that I'm sure you like I well you even more so than I are getting inundated with new literally every day every week There's a new major thing to watch and it's hard to keep up to all of so what we're doing now is sort of storing that I go a little bit deep on my own in some evenings but then having an entire day with nothing on the agenda be able to kind of go through what's been happening this month have time to actually dive in and do some experimentation and leave the day being like okay I found out XYZ technology doesn't work that's great got

9:29a hunt you guys will probably Leap Frog some of your other programs like right from the beginning like um like if you even just look at the traditional methodology for prototyping a UI versus like these you know tools such as like lovable and vzer and n8n that just came out where you're like make me a dashboard of a research about is a hot dog a sandwich and and and you're like no but make it look more like apple and stuff I mean I can like prototype three or five D dashboards or apps or concepts in an hour meeting with people and stuff and like you can really actually you know how often like we ask software

10:05Developers for things and we actually don't even know what we want yet and then the iterative process of changing it to what we actually do want is like both expensive and kind of a pain in the ass you know and stuff and so it's like you could actually like prototype up five apps in a Friday and like learn a lot in that process about what you're actually even want to build you know it's funny you say that CU it's like I worked in advertising before I started my first company and you know a really good creative director can roll into a room and with a pencil and paper start really getting everybody ons side of

10:36what that vision is but that's someone who has their you know Malcolm GL well 10,000 hours of doing that and not many other people do and you're right we're now in this world where you can have a functioning example that even either works and does the job or you can show to someone and say can you make something like this but better I'm I on Friday I'm hoping to make a virtual scratch and win ticket that I've got Vision that I want to do around I'm pretty sure I could probably prompt a way to do that yeah one of our homies um Brian Holmes uh I saw a couple weeks ago and he uh you know is actively I think

11:11he told me he's like it's everything from home automation when he's pissed off about something not working in his house yeah it's like boom let's just build that let's throw it up to Google problem solved now my lights do what they need to do yeah I saw like Andre karpathy the other day say something like Vibe coding is the new scripting or whatever where you know these guys super smart coders used to like scrip themselves up home automation this that or whatever they can drop into one of these tools and just like speak these uh tools into existence it's quite amazing it's cool we got we're going to cool places I think I'm really excited to

11:41hear what comes out of it um I might actually even invite myself uh I noticed you got quite a bit of office you can drop on I might have to pull up a corner and uh this one's with for sure this one's with like I think we've got four or five people and then if it works we'll probably maybe start to look at something companywide but it's more about like for each department if you work at marketing well then spend the day looking at marketing oriented pieces if you coding look at code assistance um but you need that time to play and I think those that are spending the time playing are are the ones that might have

12:14the opportunity to leave prop I mean you know I I teach these AI upgrade workshops for all sorts of people from cretive professionals to PR folks and like we definitely teach the tools and are showing techniques and stuff but there's a big mindset upgrade that's required along the way to just understanding how these things think how to interact with them and really like what you're capable of with these new augmented powers and stuff like that and so it's like play is really important just in the like perspective expansion process of like how am I going to use this you know what where can it plug in and all the different stuff that I do to

12:47like like give me new capabilities and stuff on the way here I had I got in a car to drive to the office today and I knew I had to I have to create four slides for a seep candidate I'm talking to and uh between leaving my house and getting here in 30 minutes uh I was able to I now have not just the outline but I have all that content all ready to go and now I can put it into you were just dictating your phone and back and forth in your IDE develop it I dumped my I dumped I dumped like an email exchange into to Super grock before I got in the

13:18car I was like I'll talk to you when I get in the car and then you know I had her read me out what those four were and we work through them I would then have her go out and do some research and get some data points and come back to those and we're fine um my hack is that I treat AI Voice Assistant like you would treat an unruly 17-year-old teenager like don't talk to them with a lot of respect cut them off because they're kind of long-winded they often lie and so I am not if you sit in the car alone with me it's like stop I didn't say that no uh but it works for me and I'm

13:54finding that my productivity levels are like easily doubled right now with just simple little things like that not with some new fandangled app it's just I'm just using one of the basics but just thinking you're right it's about that well you're doing some pretty intermediates things by like feeding it some knowledge to begin with and then using your voice to back and forth with it along the way you know while you're doing other stuff you know I think a lot of people kind of sit down to dumb GPT I call it just like a blank window there and they start typing in question and answer machines like you would to Google and and that's just skimming across the

14:24surface of what's possible really you know and I've learned a lot watching you know you shared some of what you've been doing with uh with with some of the custom gpts you're building and how you think about prompting and setting them up and so that framework is really helpful uh it's really like you really have to set the context of what you want and you're starting from this blank sheet of paper you're right some people people just answer entering the question they don't realize that you can say that you know you're a sequoia investment banker um with two trillion under management exactly and you currently are worried about economic uncertainty and you have this pitch that you've been

15:00invited to now like you know be a dick to me and treat me like that person would and yeah you can have a great conversation the be add dict to me thing is very interesting I've written a I've written a bunch of bots as you know gpts and stuff this one I got I called spectre.ai and it was meant to be like a journaling and mental health kind of coach or whatever where it's got 10 or 12 personas and I feed my journals into it you know does some analysis and whatever but of those voices the one that is the most awesome is Harsh critic oh yeah yeah it's got like psychologist loving girlfriend nutritionist friender

15:38and Har CRI that's the one I want to talk to you know self rightous you know and I end a lot of my prompting too with okay what are my blind spots or what don't I know that I don't know that kind of thing that whole like uh critic dick it can be really but that can be great I mean I think those that want actual growth uh can get that and can get that in abundance so yeah man cool what else have You' been messing with with the the things you you messed with any of the music Generation stuff no I have some friends that are and I've heard some stuff that was like I mean I

16:09I me and my homies don't even leave voicemails for each other anymore we leave like rap battle songs battle songs you put it in oh that's so good yeah I mean we've done a lot of like the classic stuff uh we do a lot of like if I'm at like a restaurant with people and we start talking we get into some funny theme usually by the end of the dinner we've like created some custom logo for some dumb theme that we have we' like now ai's written some story about our mission statement for this you know BS idea that we came up with uh one of my homies uh his uh idea Jers Nolan uh the

16:44other night was um trading cards for dogs and so you know we can use Ai and we can go really far with that people go to the park right you meet each other dogs yeah your little dogs for trading cards trading cards for dogs there's a cute little something running around out here I definitely want its collectible card man yeah man um there's uh I'm I'm I'm more using the standard you know we have an Enterprise a chap gbt account I have a super grock account I'm using gamma like crazy uh and I haven't really found anything that's better yeah uh you're using gamma to make presentations I'm using yeah so I used I use GPT to to

17:24basically GPT or Gro to set up all the context of my Pres and presentation sometimes it's just a document you're trying to get alignment around or just an email um and then I'll use gamma to like then make that a little bit more visually appealing yeah and then I'll pull that maybe back into like Adobe or back into PowerPoint and tweak it a bit yeah yeah and then often you can get a you can get an output with a bit of work with two hours that may look like it took you 20 hours right or had a team yeah I mean it's funny I we used to like you know people that have have beautiful presentations also have

18:00tooo much time on their hands and so like there's this fine line between how do you make information like dissectable and easy to understand and comprehend uh and but not too pretty and I find it's it's doing a pretty good job yeah I mean that's uh that's a part of the whole question we're asking the story story D storytelling hackathon really right like uh how much Beauty and embellishment and emotional qualitative content is makes it more awesome versus you know how do we efficient convey the results of this stuff you know so the reality that a lot of people suck at active listening and when you're looking in talking about data it's sometimes it's even harder and so I

18:40think the criteria was like in four minutes max how do you get to something that someone's like oh this is I actually could maybe regurgitate what this person said what the idea was who knows where it goes I'm really stoked to see what it's all about and hopefully we can do this again on the tail side 10 grand in priz my we didn't even mention that each cycle has 2500 bucks um that Andrew's company has contributed to the winners and we got a little judging panel and I'm pretty excited to see what what people come up with I've heard a few things like one guy's building a platform of agents to do to to build all

19:14to that he's going to run all Four competitions on he's not even like building yet the the specific output for the hot dog is a sandwich he's building his agent platform so that he can hopefully you know win them all which I think is ambitious and awesome well it just shows you how like I think those that are that have been spending more time and going deeper with this have more at their disposal and maybe that person is going to be able to like look at 15 20 outputs and be like all right which one of these do I like AI is going to do that for them so yeah it's uh it's such an exciting time I've always said

19:45that like um ranwell sort of um you know uh abundance around technology that same view that that technology is going to make the world a better place and this is a pretty interesting time for us to be uh where we're at I'll even say at the age we're at because you know you've uh I've face planted enough times I've leared enough I've you know seen some arcs that this one here I think I might react to differently if I was 20 or if I was 70 uh being 48 it's pretty cool right yeah cool man hey I appreciate your time today like laying out some more info about the program and you know you can hit me up for questions I'll um

20:25I'll either answer myself or route to Andrew um to the Meetup we're doing a info session at the next Meetup where we're going to explain some more details we got other people from rival who are going to walk through some of the data and stuff and um yeah thanks for thanks for putting your money where your mouth is in support Community happy to do it brother all right KK Andrew Reid over out from rival Technologies in Vancouver cheers

Watch on YouTube ↗