Jevons Paradox & Demand for Insight – Ep. 249

Jevons Paradox is the idea that efficiency doesn’t always reduce work—it can increase demand. In Episode 249, the crew applies that concept to analytics: as Power BI (and modern BI tooling in general) lowers the barrier to build and share insights, organizations often ask for more reports, more metrics, and more customization—sometimes faster than the team can sustainably support.

News & Announcements

• Jevons Paradox: Demand for Insight — The article that sparked this episode: a great framing for why “better tools” can expand analytics consumption and expectations, and why teams need strategy and ownership alongside self-service.

Main Discussion

The big point isn’t that self-service is bad—it’s that self-service changes the shape of the workload. When it’s cheap to create a new view, people request more slices, more refreshes, and more “can you just add…” changes. Without shared definitions and clear ownership, the organization ends up with plenty of dashboards but less agreement on what the numbers mean.

Key takeaways:

• Efficiency gains in BI usually increase demand (more users, more use cases, more “one more metric” requests).
• Scale comes from shared definitions and ownership, not just faster report creation.
• A metric that isn’t defined (and governed) becomes multiple competing “truths” across reports.
• Governance is as much people and process as it is platform settings—especially around changes and communication.
• Teams should prioritize decisions: define what questions matter before optimizing for how many visuals you can ship.
• The highest leverage work is upstream: translating business questions into measurable, agreed-upon definitions.
• If demand is rising, invest in semantic models, documentation, and reusable building blocks—so growth increases clarity instead of chaos.

Looking Forward

When analytics demand spikes, use it as a trigger to strengthen definitions, ownership, and reusable models—not just to crank out more dashboards.

Episode Transcript

0:00 [Music] foreign foreign [Music] good morning welcome back to the explicit measures podcast with Tommy

0:31 explicit measures podcast with Tommy Seth and Mike good morning everyone how’s it going good morning gentlemen and happy Tuesday to you good morning everyone it is good to see your faces faces are seen phases are seen so let’s get right into our topic today we’re here a little early but we gotta also end a little early as well so this I don’t who found this article I don’t remember where this article came from you did I find the article though was it me or is it somebody else and it

1:01 was it me or is it somebody else and it was like on the article you put on the agenda we figured out what it was about [Laughter] something of I I don’t know it was I don’t know if if we are influencing the internet or if the internet is influencing us I think it’s probably a little bit of both ways here but I believe we’ve talked about this one together on the internet before and so and so we’ve I think we’ve in previous episodes I’ve alluded to or talked through this concept before with

1:31 talked through this concept before with other other people the article today which we’ll put in the chat window it’s javen’s Paradox the demand for insights is the the name of the topic or article for today in the the chat window I definitely know that you’ve brought up this Paradox before yes different in different chats but I I we’ve never talked about it in this context I don’t think I don’t think so but it it very much makes sense so I guess we should just jump into

2:02 so I guess we should just jump into the article here real quick let’s talk about what is I don’t even know how to say the name right javen’s Paradox yeah I think that’s it yeah James Paradox Japan’s Paradox I don’t think he’s French I think you I don’t think he’s French I think James Paradox oh we’ll go with it know James Paradox oh we’ll go with it so 19th century right yes and this this Paradox came out as it’s as it’s now known as the article says it’s an example of a rebound effect where the expected gains from efficiency

2:32 where the expected gains from efficiency are offset By changes in behavior in the original case as coal engines got more efficient it got cheaper to generate each watt which in turn simulated a ton more generation so there’s even an energy aspect of this right now like Technology Drive as you drive down more in efficiencies in our consumption of increases are you actually like they’re they’re conversing on how do you how do you not

3:03 conversing on how do you how do you not have the rebound of exponential growth and energy again right but we’re talking data in AI today which I man this was this was this was a fun thought provoking little article because I think where my mind’s been initially is oh yeah we’re this is going to drive efficiencies and we’re gonna we’re gonna see changes and patterns of probably less jobs in certain areas or something like this and this challenges that thought it does what’s crazy too I feel like we’ve all even

3:34 crazy too I feel like we’ve all even felt this even in our early stages of power bi I feel it if you feel it can you feel me even in the early night like early 1915s the 1915s the 2015-2017 when we were getting power bi I think probably all of us were some of the experts in the room about power bi and you’re like the more I know the faster things will be the more efficient but I think it made us more busy we were incred I was as busy as you can get at can be and I was the power bi Guru so

4:05 can be and I was the power bi Guru so to speak to speak and I think now it’s even more relevant as we go into this now era of AI I find it interesting though just even when the platform power bi came out our jobs got busier busier [Music] [Music] and I this is I’ve also heard this analogy used in the same way of talking about about light pollution the same way yeah so previously when you made light you would hear you would have you had to go make a candle and it took a long time to make a candle and you had to get all

4:35 make a candle and you had to get all this wax together and then as light kind this wax together and then as light evolved as the technology improved of evolved as the technology improved right so then we’re not having oil lamps and so those are changing things so now it’s easier to produce the oil and have this lamp and so now we start seeing lamps on streets so every street corner gets a lamp and people like to turn them on and cool so then it goes even cheaper than that so then the economy of electricity turns on and electricity can translate power very quickly and easily and then we invent the light bulb and then the light bulb becomes an LED light bulb and

5:07 light bulb becomes an LED light bulb and now we have literally satellite images of the entire world showing us that there is no Darkness there’s there’s just constant light on all the time in certain portions messing things up so I know we have this new consumption we’ve consumed so much of light that we now have this thing called light pollution you can’t you can’t go find a place where it’s dark enough to go watch the Stars well yes the age-old question like what do people want and again going slightly back in time towards when power bi really became

5:38 towards when power bi really became Apparent at organizations you necessarily did not know that people needed these Data Solutions or they needed the type of insights from Power bi I think from business applications such as powerapps to Citizen developers but one of those became available and became something that could be reachable by organizations very quickly all of a sudden it’s like you quickly all of a sudden it’s like what we actually could want you know what we actually could want you know what we actually could want what it would actually be a great know what it would actually be a great solution here not that those weren’t solution problems before but they became reachable but what were the let me just

6:09 reachable but what were the let me just ask this question here before we jump into this one before power bi what were the visualization tools you could use could use like Tableau Tableau which which learning services but none of those were those were all paid paid programs yeah everyone’s paid though yeah Tableau to

6:27 everyone’s paid though yeah Tableau to some degree a minimum of 35 a year or something like there’s some price that you had to pay for even just to get the program so one there’s this idea of like from Jason’s Paradox here the cheaper you make something the less the less cost you bring to something the more it becomes consumed and so we had Excel and I think Tableau are probably my initial two visualization programs and there’s probably a bunch of other open source R and python things you could go after too but I wasn’t

6:57 you could go after too but I wasn’t using them yeah and if you look on the side of the box of the Tableau gift box it said modeling not included modeling not included well you you’re you’re equating it to you’re equating costs to direct dollars but totally I think like the I think one of the biggest challenges here that might be more applicable is the time cost right like we just didn’t have systems that could handle like handle data loads let alone data volume

7:28 handle data loads let alone data volume loads like we have them because there were Hardware constraints and there are Hardware Hardware software constraints it took a lot of time to generate and build these like monolithic warehouses and that was the approach that everybody was gonna you approach that everybody was gonna get out go down and you have a lot know get out go down and you have a lot of data challenges and all of that and then so the visualization piece is certain was certainly part of it but you didn’t have systems I think that could support anything like that so yeah Tableau definitely from a visualization standpoint yes like I think was the first in in terms of providing a tool

8:00 first in in terms of providing a tool for end users to engage with data and and pull insights out of larger data sets sets so so like Kudos there but what power bi did I think for us and for businesses in general is it it reduced the time or the cost of pulling all of this data together in such a way that was understandable for business users right so you’re lowering the the level of

8:30 the the level of knowledge that somebody has to have about all these back-end systems and that’s where my thoughts today around Paradox are because I’m like holy cow like we’ve been talking about AI Expediting right yes I don’t as a as a code generator helper right I don’t need to know all the code right I don’t need to know all these things as it becomes more efficient well aren’t we doing the same thing in and in reality like the thought today was like wow I

9:00 like the thought today was like wow I wonder did we just Elevate all of data analysts into a realm and this is where fabric comes in into a realm that they otherwise would have to spend a lot of time to engage in and I’m not saying we’re at that point right now there’s still a lot of like the technology pieces but eventually did we become so efficient at at reducing the complexity of a lot of the elt ETL processes and programs and

9:32 elt ETL processes and programs and compiling data and applying business logic to make data meaningful that we’re going you you put businesses in a position of hey what we become so efficient we don’t need all these people or we’re going to reduce the time that we need from people like hey why don’t you work less or or are we going to capitalize that and hire more people because we don’t want our competitors to beat us out right well in the middle paragraph towards the end here it talks about the demand for data

10:03 here it talks about the demand for data isn’t fixed so we’re talking about a supply and demand curve right yeah so increasing Supply also increases demand or increase if you if you in if you decreased the cost if you decrease the time to deliver these insights if you leak decrease the effort it takes to make to your point Seth of report a thing like that the more you decrease that load the more people are going to demand more things so and and this Rings very true to me in my career because it was a lot around we

10:34 career because it was a lot around we need to build a data warehouse okay well let’s let’s spend some time deciding what the data warehouse was okay let’s figure what that thing looks like months go by and now you have a universe a thing created that you can go read well now you don’t have to do that as much now we’re talking weeks or months a month or less to be able to build some of these of these collection of data places and I won’t call them a warehouse yet because I think there’s more things that go along with a warehousing technology that you want to put in place but I’m saying we can now collect data faster than ever

11:05 can now collect data faster than ever cheaper than ever on local machines like my laptop for me it was it was the experience of I made an Excel sheet that used power query and I couldn’t load anything more than a million rows of data and if you’ve even if you did load a million rows of data in Excel it would be so stinking slow it would just be so stinking slow yeah but you don’t want to you want to move to power query and then I can get tens of millions of rows I remember having the moment of hey we’ll build this little interim solution

11:36 little interim solution while we wait for the data warehouse to be built a year later we’re not done and a year later we have 10 million rows in Excel and I’m floored I’m like oh wow this just changed the game for me I’m like this is different and then it prompted then it prompted you to be like well I’m just gonna go build this myself and that’s and that’s why I was like wow if if I’m given tools like power query and Excel where I can do things at 10 million rows in data and I’m just I get

12:06 million rows in data and I’m just I get I can figure it out myself I can get stuff done I’m just gonna do it myself and so I won’t wait on other teams just to do this anyways I’ll pause there and there’s so many more things to think today because I think all of us here have a very good Insight in terms of business and also technology and I think part of the adoption I’ve seen with power bi at a new organization is they still treat it as the box of like it’s bar charts and line charts and that’s all you can do yes and that’s why I usually ask at the beginning of a discovery saying imagine

12:37 beginning of a discovery saying imagine we’re not talking about the report you got a million dollars you had a thousand people working for you what would you have and try to get them out of the box of what they perceive the technology can do and usually that and it offers insights or solutions that are possible in the platform and that’s still I think

12:54 in the platform and that’s still I think we’re getting I I still think we’re getting there with power bi much less Fabric or AI where people are understanding the capabilities of power bi and now we’re introducing with fabric we’re introducing with some of the AI tools now from data modeling and all from both the business and the developer on what’s possible which is that I think that that Crux point in when it comes to the javens Paradox on the more consumption the more consumption because there’s more demand for something what that demand is right

13:24 for something what that demand is right now now well we’ve we’ve we’ve talked about this I I think on the periphery as well right but it’s it’s I think the same concept where when we talk about self-service there’s there are challenges to either a way in which we’ve done it in the past or or from an A not and I don’t like the word administrative perspective here but at the same time like if we if we talk about data governance or quality of data and ensuring certain

13:55 or quality of data and ensuring certain levels of accuracy and certification like there’s still holes in the self-service realm right like people can build things inefficiently they can spend tons of money that not knowing what they’re doing like there are a lot of inefficiencies in bringing all of those people into this ecosystem where I don’t think we’ve solved it but I think there has to be that adoption like the road map for okay like we can get immediate business value but how do we transform that into permanent business

14:27 transform that into permanent business value and I think that’s the difference because that we haven’t solved yet and maybe maybe we become more efficient at in both ways right in terms of allowing not allowing but having more people be able to do more even more than they do not right now right like what is five minutes to wow we’re not talking about a complex Dax statement we’re talking about it some but I think you take one of the examples I think Mike you brought this up all the time like people people go from like sum

14:57 time like people people go from like sum to time intelligence I want to see how things change over time and that’s it was hugely complex from the standpoint that like you just jumped into the deep end of Dax well in the future did they right if we have code generators that automatically allow a business eater like hey I want to do this and this and this and then it just happens well that’s a huge efficiency thing and now they’re much more powerful right so in some ways I think maybe I’m

15:28 right so in some ways I think maybe I’m starting to think about this in in a okay well if I can if I can explode the usership of all of these reporting tools because we’ve driven down the efficiency so much can I do it in the other direction can I can I simplify the the process by which we get into certified data sets and governance and a way in which the data systems work and operate in an

15:58 the data systems work and operate in an expected fashion across the board and hey maybe maybe Microsoft comes out with the the auto here’s everything tool but I we’ll see when that happens maybe that’s the next gen fabric V2 I I think but but I really think you’re hitting on something that is becoming increasingly important for organizations while while the power bi ecosystem and I’m talking more about the power bi side at this point I’m not talking about the other fabric elements right now

16:29 other fabric elements right now there’s another I had other comments around this right now that we’ll we’ll come to those in a minute but if you think about just the power bi side side with this javon’s Paradox with the reduced cost to build create you’re going to get a lot more so the the for lack of a better term the pile of trash can can continue to increase in size so right it’s it’s the challenge for organizations now is is quickly changing into what is really adding value because we can create every

17:01 adding value because we can create every if you have one Creator versus a hundred creators creators right you’re going to get a lot of stuff made right so okay solution exactly there’s a lot of there’s now pollution occurring in your analytics so how do we have a consistent Story the truth how many Enterprise data models should we be maintaining and what do those look like so I I firmly believe I’ve worked with some organizations where we’ve we’ve basically taken all of their production data and wrapped it up into two or three

17:33 data and wrapped it up into two or three solid models and every aspect of their business is being touched by a couple different aspects I Engineer the data to get into two or three solid models now granted there’s a lot of edge cases organizations are big everyone every department has their own unique needs but at the core of what you’re doing there your the business is simple enough or the business can be thought of in a way that says okay look we need to highlight and from this again lack of a

18:03 highlight and from this again lack of a better term here I don’t have a better way of saying it there’s a pile of trash let’s highlight the gems where’s the Diamonds in the Rough in our analytics space and how do we one identify them mature them so they become elevated above everything else and get people to understand that that is the story of the truth truth and how do we highlight those and this is where I think when I look at what Microsoft’s building for like one like data hub data hub it is part of that right here’s a whole bunch of data sets you have access to here’s the ones that are

18:34 access to here’s the ones that are certified right they’re trying and here’s what domain takeover domains help you group things together like we’re already we’re already trying to group this files together so we can add value well there’s there’s a kink in the armor there and in honor of week one and now Jordan loved being the next day Aaron on hot takes thing I feel like the the honestly the true like the which my gosh it’s like wait I guess we have to wait till Jordan Love’s gone now for the Bears

19:04 Love’s gone now for the Bears but honestly I really think thinking about where power bi is now introducing fabric I I think a lot of those problems Mike are because that it’s twofold organizations are not willing to invest at the time the technology is

19:19 invest at the time the technology is changing in with the governance and what the quality is whether they’re not aware of what the Technology’s capabilities are at the Times introduced much less then adapt and react to it and I think the technology is moving so rapidly I I would I would true or false statement or if you’d say this is an overreaction the statement that most organizations are still behind in data governance just in power bi alone I’m going to say true but I’m going to refine your statement okay

19:50 refine your statement okay the data culture or organizations okay is being far outstripped by the technology at this point the technology can do what it needs to to provide a strong and solid data culture and and the reason I’m saying this is because the number of organizations I walk into and they’re like we have Professional Engineers that know how to do power bi then I’m like well great show me what you do and they’re like we tell everyone go to the web and build reports reports I’m not quite sure building reports in the web is like what I would call a power bi professional you’re getting

20:21 power bi professional you’re getting stuff done I agree but to me that’s the very low end of things you need to know about power bi it’s a very limiting feature set of what all of power bi can do you’re skipping a lot of modeling and custom building a desktop you’re not you’re not even giving people a desktop like what’s going on here so like organizations are overrating how capable their people are because they don’t even have a standard to measure their people against and this is where Seth you made the skills Matrix a while ago talking about like more

20:51 a while ago talking about like more about the report and modeling experience this was super helpful because people can actually evaluate where they’re at in the scale and what skills should they know or what skills do they know and where they can place themselves as far as how mature they are in that space your point here is I think further ahead than I would be right I you’re you’re speaking or at least the way I heard what you said said it came off that we’re dealing with the pollution and you’re already navigating some of the the governance and

21:22 some of the the governance and strategies conversations and I’m not saying that doesn’t happen sure but if AI is the driver for efficiencies and data and data that I would argue that the challenges that we’re talking about haven’t even remotely started like if if you think we have problems now with pollution pollution right that’s true I would drive this so far down that art that the user base for

21:54 far down that art that the user base for reporting tools and accessing working with data is going to explode like we’ve never seen before that’s so I wasn’t and think I wasn’t thinking about where AI fits into this but you’re really right I would agree with that statement the AI is going to continue to throw fuel gas on the Fire and you’re now going to have even more proliferation of this because right where this leads me into thinking is is longer term strategy right like if

22:25 longer term strategy right like if we’re already having these challenges and conversations like what would happen if if 25 more of my organization is now polluting that environment and I don’t have a plan or a strategy on how to attack that from the AI Limelight are you saying this from the Builder the the bi team or the the author point of view are you saying this from the consumer point of view how that AI in a sense has that effect because I can see it both ways of

22:55 effect because I can see it both ways of having like implications we’re not aware of I I’m saying you would transform all of your consumers into builders I see like like the capability skill set the skill sets are so far reduced because you’ve you’ve interjected tool sets or systems and in this case it’s AI help me help me build this report I want to sit down in an interface and like in the the pictures right that even Microsoft has shown like generate me

23:25 Microsoft has shown like generate me a report that shows me this and this and this and this and this bam a little magical right like I’m not saying we’re ever gonna get there nor do I know but at the same time like if that is the case or those capabilities do arise in some way shape or form I don’t I don’t I’m a consumer I don’t need you as a developer what I need to do is go build my own reports the artifacts and how that report gets generated on the back end will certainly still be the same and you need to Val you need people to

23:55 and you need to Val you need people to validate it understand it ensure that it’s accurate like where does the data come from like et cetera et cetera et cetera I’m just saying like that’s a lot of pollution right or potentially could be that would have to be somehow fit into a some sort be somehow fit into a some review process I think or how do you of review process I think or how do you ensure that people don’t make things on their own anymore and it only comes from curated data sets it’d be interesting to see if like today the professional realm changes from doing report building to

24:27 changes from doing report building to almost always and only doing curated data sets right because I think that’s where it’s going to go technically as long as these things should fit together I could I could see AI systems understanding how they bolt in or you understanding how they bolt in or us us telling them how to so that know us us telling them how to so that exponentially there’s a lot more usage of that data in many different ways and and you can’t say we’re not we’re not close to that yet because I did another test with that Cruiser AI tool yeah but so so but that that’s the interesting

24:57 so so but that that’s the interesting other part of this yeah right where technology is already way out front of business right like Mike said right so the technology may like get there instantly but at the same time like it’ll be interesting to see whether or not because of the AI Buzz organizations dive into that immediately but do they we’re seven eight years into Power bi and we we know organizations that are just starting to to take that Journey so the the

25:28 to take that Journey so the the adoption of new technologies is has in the past not been so instantaneous and we’ll it’ll be interesting for me to see whether or not AI is different in in that regard than other technology technological advancements so I guess with with the

25:47 advancements so I guess with with the introduction of AI and these easy tools and I want to see this calculation and AI can generate it how do we make sure then or maybe not how do we make sure but how can a business ensure that like the human element remains still like integral with both the that interaction with data and and the business and I want to answer your question Tommy but I also want to go I also want to make another note here as well around this is going I feel like AI can do two things well it does more than two things but I’m just saying there’s two areas where I think improvements will be seen from AI

26:18 improvements will be seen from AI one is I think AO will help in the creation of models AI will help in the in the engineering of the data right Tommy you were saying like look I use this Cruiser AI it helps me write Jack statements I can literally point it at a a page from you literally point it at a a page from here’s here’s Dax dot guide right know here’s here’s Dax dot guide right so now I can I can find the best sources of information around data and then it knows now okay here’s all the Articles from SQL bi here’s all the Articles around Dax dot guide it uses them as

26:48 around Dax dot guide it uses them as reference points to then scan that and produce to me answers from there which is amazing so I see this that’s creepy how good it was and then we see things like coming out from from Marco Russo and SQL bi team there’s now the Dax stop some the Dax Optimizer right there’s they’re probably doing some maths or AI pieces behind the scenes of that to you behind the scenes of that to figure out what’s going on there so know figure out what’s going on there so really cool very neat technology but that’s all about the creation of the

27:18 that’s all about the creation of the data side of things Seth I think what you were alluding to was another part of AI that is it’s around how do I ins provide the insights like I’m just going to throw AI a cube here’s my data model AI should be able to go through that and say here’s things that are interesting you may want to know about it and yeah I’m not sure if we’re there yet I don’t think I’ve seen a very solid example of throwing a bunch of information and data at an AI and having it produce lots of

27:48 at an AI and having it produce lots of charts graphs or insightful pieces of data but I don’t imagine we’re too far away from throwing a data model at an AI and saying why are my sales down I think there’s some things that like data robot and there’s maybe a couple other ones that are doing similar things but I question when I see these things I question the validity of what we’re actually getting out of it right and is it done in an inefficient way so I don’t I don’t know if I can really vouch for these other tools because I haven’t spent enough

28:18 tools because I haven’t spent enough time to really know are they going to be able to create the insightful pieces but what I will say is there are patterns in how you want to represent data like I want to talk about position and Direction I want to talk about sum of sales broken down by various categories so where I do think AI could help out is I could I could give it some prompts around I care about these kind prompts around I care about these columns I care about these

28:48 of columns I care about these calculations and then what the AI does is it just runs away and it builds 15 20 30 different visuals for you and comes back and says here’s a whole bunch of visuals that may answer certain kinds of questions now if the AI can start matching let’s say look it’s it’s part of the features in desktop where you say explain the increase explain the decrease on bar charts that that is probably one of the more relevant features I see from AI that Microsoft has produced in desktop that seems

29:18 has produced in desktop that seems useful to me right that that helps me so doing more things like that where businesses will be able to incorporate these technology pieces into their their workflow will be important so let me let me answer your question directly Tommy how do you prepare a business to use these new more modern AI based Technologies in what they do every day I think it has to be a strategic and slowly thought out thing to some degree because because with the introduction of AI you have

29:48 with the introduction of AI you have a harder time Discerning what is truth what is really real what is I was I was using the other day I said I had a blob of text it was awesome I had a blob of text I threw a blob of text at the AI and said add these numbers up and it was just text and numbers and all this random stuff it was it was like from a statement of work I was writing and did a great job it said here’s a way you can do this and it literally found all the numbers in the article and it and it added a mob here’s how you could do this and it added the numbers up and the my engineer who’s working with this Amigos

30:18 engineer who’s working with this Amigos check that math I’m like what do you mean check the math he goes check the math and literally it pulled out all the right numbers it pulled them out it literally listed them 73 plus 5 plus 12 plus it did the it did the math added all the numbers up and gave me an answer the answer was wrong so it was it wrote out the entire statement of the math but the answer was incorrect so I literally copied the text from okay great it parsed out all the numbers for me no problem I parsed out all the additional the the math the math

30:50 all the additional the the math the math part and put it in Google and hit go and I got the right number I did it again on his second set of data and it did it correctly the math actually worked out correctly so it’s subtle things like this if you’re not really in tune if you’re not really paying attention like where do you check the AI to make sure that it didn’t just make something up arbitrarily it could have it it may have found something in the article and then misrepresented the addition at the end of it like it did so good and got the very last step and then fell over

31:20 fell over what for my science projects yeah yeah Tommy’s volcano it was it looked so good and then the volcano just fizzled out the side or had a leak at the bottom and it never came out interesting point there and I I don’t want to divert too much so Seth did you have something to say before okay so the interesting thing with AI now is not necessarily and to your point not necessarily getting right the first time but one of the most powerful statements you can ask any prompt or whatever tool you’re using is like

31:51 whatever tool you’re using is like that’s not what I wanted I want to focus on this yeah and then it usually adapts and that’s actually correct the strongest point and I’m I’m thinking about this in the context of like from the business to the data to AI to the data or or the business to Ai and I think that the point of that is like the dialogue right where it’s not just

32:11 dialogue right where it’s not just saying I want this point out to spit out this and I think our strongest point from from the human element or at least from you think about where does AI fit where does data and outfit or where does fabric fit is also being able to have a dialogue because I’m seeing too much right now either still this top-down approach of here’s how we’re rolling everything out business adapt to the new the new governance here’s how the day you’re going to get your data

32:41 the day you’re going to get your data adapt to how it’s going to be and there’s a big part here on open dialogue and being more open both from the business to to the business intelligence team or however that framework is because not that we have to copy what AI is doing in terms of what it works but if we’re going to have this dialogue with AI I think that needs to be a big part of us moving forward too I I would agree with that but at the same time I almost think that like business always wants outputs right whatever they’re

33:11 wants outputs right whatever they’re using the output is something they pass on to somebody else or they they use it to make a decision I think what will be interesting is with the proliferation of AI tools allowing business users to be like hey I want this and an output being generated that rather than developers building outputs for people they’ll probably review like the compilation how did you go about getting this information does this make sense right so rather than so you’re verifying and double checking my tierpoint that yes calculation adds up to what you would

33:43 calculation adds up to what you would assume it would but in the context of business it would be well I know these things about the business and how the data is stored and the other thing in here is a at that where a is going to get tripped up is how how much innate just business logic garbage do we as employees know that isn’t represented in the data anywhere yes like imagine brain dumping all of the things and why systems were set up the way they were and which flags mean this thing and these filter sets all have to be in this place except for

34:14 all have to be in this place except for this value and that’s what that it all of that won’t live at least initially in the back end for Asia to understand or provide instantaneously like magic right so I think there’s a long way we’re going to go before like everything’s just so instantaneous and AI is going to solve all our world’s problems because innately a ton of that isn’t in a computer computer probably for the detriment of the business business future I like this idea this idea is

34:45 future I like this idea this idea is amazing because what Tommy said around being conversational and what you said Seth around like there’s this huge wealth of business knowledge that’s producing introducing the business knowledge back into the AI engine yeah so I I think there’s like a and then you get you get the sales of marketing what sales mean very true like well and that’s maybe where the AI can prompt things a bit more right hey describe the calculation you’re trying to produce well I want to calculate this thing that adds all the sales for this region excluding

35:16 all the sales for this region excluding these things great but as I think about that scenario right why doesn’t the AI have conversations with more business users and what you start sifting out is the AI should be able to collect multiple requirements from multiple people all the same time and start trying to figure out okay based on these 10 people and 10 people’s worth of different requirements the AI should be trying to figure out okay okay where’s this where’s the center point of

35:48 where’s this where’s the center point of all these requirements how like hey there is this hey I see there’s this key column that has a weird ID and and the businesses were oh yeah that key column is this and the business user describes the data and then the AI says great I’ll keep that for later on the data then the the the knowledge of the user gets put back into the AI the AI stores it somewhere and then the next person who asks about something the AI brings up hey did there’s a column over here that does exactly what you want based on this flagging like because it

36:18 based on this flagging like because it learned from the one conversation that it can now extrapolate new data for the next user like that’s the golden ticket here why aren’t we having the conversations Seth you just said to me like the Golden Nugget here where it’s not just the output where no one and I I will still say this no one will spit out 15 000 in a in a report and make a decision it’s how did we get there and it’s always about whatever that final result is well why did it get there and that’s my last Point thinking

36:48 that’s my last Point thinking about being open with the business let’s talk about not just about your final number but how did we get there like that and a I should do that they should tell us how to get there your your thought is another whole podcast for me which is it like you just maestroed me man like where where do I want to use AI yeah I want to use it as the requirements Gathering yeah I want to go build this here are all the people in the business units go figure out what they know like 100 and like generate these summarizes everything back to me

37:18 these summarizes everything back to me anyway yeah love the conversation I wanna I wanna continue having I think I think we’re gonna have a lot more conversations around this I I love this article because it it Twisted me on my access and gave me a different perspective on the whole AI discussion so it was a fun one I like that a lot as well with that we got about thank you all very much for listening and watching the podcast we really appreciate you thank you very much chat audience a lot of good thoughts there in the chat as well well so thank you very much for participating and jumping in with the

37:48 participating and jumping in with the conversation if you like what we have here if you like this information if you like thinking about AI or javen’s Paradox or figuring out where does the intersection of data culture and data modeling fit in your organization this is probably the podcast for you we talk a lot about those things here so if you like this please share it with somebody else let somebody else know you had some value from this episode and we’d love you to show someone else Tommy where else can you find the broadcast and we’re on a Time Crouch if you like this podcast make sure to subscribe on Apple Spotify

38:18 make sure to subscribe on Apple Spotify wherever podcasts are sold make sure to join us live on Tuesdays and Thursdays and if you have a topic a question or an idea you want us to talk about go to powerbi. tips slash podcast and you can submit a mailbag is FYI the podcast is free so you don’t have to go buy it you free so you don’t have to go buy it there’s nothing there’s nothing but know there’s nothing there’s nothing but podcasters sold for free podcasts or not apps apps enjoy we’ll see you all next time

Thank You

Thanks for listening to the Explicit Measures Podcast. If you enjoyed this episode, share it with a coworker and drop a question or topic suggestion for a future show.