Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Microsoft Cloud IT Pro Podcast

Microsoft Cloud IT Pro Podcast artwork

Episode 440: The J Factor: Bringing Your Experience to Generative AI

Nov 5, 202638 minAudio

Welcome to Episode 440 of the Microsoft Cloud IT Pro Podcast. Recorded live at the North American Cloud and Collaboration Summit in Branson, Missouri, this MS Cloud on the Road conversation has Ben sitting down with Jay Leask of Microsoft and Microsoft MVP Norm Young. Norm turns host and asks what their work looked like before generative AI: hand-written PowerShell, manual conditional access reviews, and deliverables that took two weeks instead of 45 minutes thanks to transcripts and Copilot Cowork skills. From there it gets philosophical. They talk about whether your own experience (the “Ben factor” and the “J factor”) still shows up in AI-assisted work, why Ben feels some of the fun and creativity of the job slipping away while billing for faster work gets harder, and what automation means for people whose jobs cannot simply be reskilled. They close on whether models trained on AI-generated content will plateau or regress, whether the current spending pace can last, and why the sooner AI becomes just another feature, the better. Show Notes North American Cloud and Collaboration Summit Copilot Cowork overview Use Copilot Cowork: Cowork skills Connect with Jay Jay Leask on LinkedIn Norm Young on Sessionize Jay Leask Jay Leask is a Principal AI Workforce Engineer at the Washington DC Microsoft Innovation Hub, where he works with customers on Microsoft 365, Copilot, and modern IT practices. He brings more than 25 years of IT experience, much of it with public sector organizations, and is a familiar voice at community events like the North American Cloud and Collaboration Summit. Norm Young Norm Young is a Microsoft MVP and solution architect who focuses on Microsoft 365 and the Power Platform, with a particular interest in Power Platform governance. Based in St. Catharines, Ontario, he is a regular community speaker and blogger on how Microsoft 365 and the Power Platform work better together. Prior podcast context Ep 432: AI-Era Consulting Billing for Microsoft Cloud Delivery Ep 427: Copilot Cowork Hands-On Experiences Ep 417: Workplace Ninjas US, Zero Trust, and Conditional Access! Ep 410: Joy and Jay take over and talk Change Management About the sponsors Nasuni is a leading unstructured data platform for enterprises where file data is mission-critical for both people and AI. Nasuni powers the operational file layer where work happens - helping organizations manage, protect, and activate data so teams can work smarter, reduce costs, and operate securely without limits. Intelligink designs, secures, and modernizes Microsoft 365 and Azure environments for organizations of every size, working alongside your internal IT team or MSP. From security audits and Copilot readiness to migrations and ongoing fractional cloud services, it’s the same production-tested advice Ben shares on the show, applied to your tenant. Book a conversation

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

Welcome to Episode 440 of the Microsoft Cloud IT Pro Podcast, recorded live at the North American Collaboration Summit in Branson, Missouri. This is a show about Microsoft 365 and Azure from the perspective of IT pros and end users, where we discuss a topic or recent news and how it relates to you. In this MS Cloud on the Road conversation, Ben sits down with Jay Leask of Microsoft and Microsoft MVP Norm Young.

Norm actually turns hosts and asks what Ben and Jay's work looked like before generative AI. They also discuss whether your own experience, the Ben factor or the Jay factor, still shows up in AI-assisted work, and if so, what does that look like? Jay Leask. Ben Stegink. And Norm Young. Hey, Norm, tell us a little bit about yourself. I'm coming to you from St. Catharines, Ontario, Canada. I would like you to know that you're in Branson, Missouri.

In the U.S., not Canada. Right. I will be forgiven for my choice of words, and you too will also be forgiven for not knowing where St. Catharines, Ontario, Canada is on a map, and that's totally fine, and I like to start that way. I'd like to say that you didn't even ask if we knew before you made the assumption. No, don't ask now. I know you don't know. I know you don't know. And the last time I talked to Ben, we were on the other end of the U.S. in Seattle. I don't know exactly where it is in Ontario.

I know a few places in Ontario, but not St. Catharines. St. Catharines is maybe 10 minutes from Niagara Falls, the U.S. and the Canadian one, and about an hour and a half, two hours to downtown Toronto, which most people know. So, not in a touristy place, not in a busy downtown place, but a place where you can raise a family, you can have a nice life. I have a nice view of Lake Ontario from my front porch, and it's a blessed life. I should have stopped and said hi to you when I was at Niagara Falls in June. You brought that up in Seattle. Do you remember? I kind of do now.

That's the worst. I remember that too. And Norm clearly has a better memory than I do. And I bring this up, that's home for me, and things that we're famous for back home are, if you're a music fan, and you know a band called Rush, Canadian, one of the all-time great Canadian rock and roll bands. Yes. The drummer lived in my neighborhood, the late, great Neil Peart. Okay, okay. Yep, so that's kind of what we're known for, and we're also near all of these vineyards and hard stone fruit type farms because we're so close to the lake and we have an escarpment.

And so winters in my part of Canada are quite mild. What is a hearthstone fruit? A peach. Oh, okay. An apricot. Hard stone. Peaches? Yes, they have a hard stone. The pit. The pit is hard. Hard stone. Yeah, hard. I missed the D. I heard hearth. Like a hearthstone is what you put over your fireplace. Well, you know, the truth is my- It's also known as a hearthstone. My articulation of all words at the moment is being impeded by this microphone that's in a very uncomfortable position. That's that, sir. I don't know. By the way, I tried to start playing Rush just for like three to five seconds of it just to get like the drum beat that everyone in the world knows.

It's playing on my headphones. Oh. Which did you guys know good? Ah, this is true. You should have told me I could have pulled it up on Spotify and played it. You could have thought of it yourself. I was busy talking to our guest. And so now that you know where I come from, and we are in Branson, Missouri. No, that's fine. It's by Niagara Falls. By Niagara Falls. What is Branson, Missouri near? For those of us who don't know American geography. Not an airport. Not an airport. Springfield, Missouri, which is also near nothing.

There's an airport in Springfield. There is an airport. Yeah. Yeah. No, not much. There's a small river here, and everything is beautiful here. It is beautiful here. You know what? I've heard Bransoners referred to. It's like the Bible Belt version of Vegas. Yes. Which is not something I knew I needed in my life. This is my second time coming to Branson. I hear it might be the last time I'm coming to Branson, not because I did a poor job at my last session. The police have put your photo on the walls out there. They're like, do not let this guy in, eh?

Here comes security right here. Oh my God, it's true. Thankfully, my credit card was accepted this time. So that all worked out well. But that's not what I'm talking about. Was there another time it wasn't? I'm not going to disclose anything on a live broadcast. Okay. My accountant can speak to all of them. And my defense attorney, of course. But I understand that Branson is, the event that we're doing here is probably going to not happen next year, which is very sad. But I do love coming here. And I love that this event has got the right size of people who just want to learn.

And it's not in my last session. I don't know what it was like for you two. But it wasn't just sit back and receive. They were sharing. They were engaging. It was like, it's what you want when you come to these things. When you come far as from St. Catharines, Ontario, Canada, across the U.S. border to fly out of Buffalo and take multiple planes to come to St. Louis Airport. Is that where we all arrived? A Springfield. Springfield. Thank you. Well, I don't know. You may have flown into St. Louis, but that would have been a long drive. Yeah. I hope you didn't fly into St. Louis. Well, it's not that long of a drive. It was an hour drive. You don't even know which it is. That's Springfield.

An hour is Springfield. St. Louis is probably like two and a half hours. Yeah. Two hours. I flew to Tulsa, which is about two and a half hours. But I flew into Tulsa, stayed the night, and then drove to Springfield to pick you up and then came here. So bring up Branson. Yeah. Question. Yes. Are we in what the U.S. would consider the South? No, we're in the Midwest. Yeah. In the Midwest. Right. Yeah. Okay. All right. The conference food has been fantastic. And breakfast today, I had my very first experience with biscuits and gravy. That was your first experience with biscuits and gravy? I've eaten biscuits before, but that is not a thing in Canada.

I mean, you guys have poutine, so. I was going to say, that's not even a thing in Michigan. Like, biscuits and gravy is not. Biscuits is a thing anywhere there's a Denny. Biscuits and gravy. It's not a thing. And on first- I'm with Norm. I don't eat biscuits and gravy. On first bite- First of all, sir, that's all I have to say. Okay. First bite, I didn't see what all the hype was about. And then a very nice lady beside me suggested I have it with some sausage. Yes. And then it changed your world. It's a game changer. It is. It is. It's a total game changer. The biscuits and gravy here, the biscuits were super fluffy.

Loved that. The gravy could have used a little bit more peppercorn. The heat from a peppercorn does change it. And then the flavor from the sausage absolutely made it. I don't even, did you try it? I've tried it before. Maybe I need to try it again. Did you try it here? I haven't tried it here. Okay. Well, tomorrow morning for breakfast, biscuit. You take a biscuit, you break it in half. You take a sausage, you cut that into pieces, and then you pour the gravy on top of it. And if you can find some good peppercorn, throw that on top. Throw that on there. I'm learning something about you today, Jay, that I don't know that I do.

That the fat guy likes food? No. Between lunch and this, you seem like you are borderline a bit of a foodie. Oh, yeah. Yeah. I don't know that I really... I cooked in a kitchen for a while. Did you? Did you really? Yeah. When I was a teenager. What kind of cooking would you have done? So I worked for Pizzeria Unos, but I worked the whole line. So I started at the pizza side, and then I worked my way to the chef side, and I spent about six months on the chef side. Yep. And then I went to college. Learned your lesson. That if I wanted to make money, it wasn't going to be in the kitchen of a Pizzeria Unos.

I'm glad you brought that up. Yeah. Because all of us have done something in the past. You were a line chef at a pizzeria doing all the different things. Now, we'll start with you, Ben. Whose show is this? Apparently not ours. I'm helping out. I'm helping out. This is an audition, I hope. What was your work focus like two or three years ago before Copilot hit? What were you doing? You know what I was doing. I'm going to let Ben go, but... What was I doing before I was spending a lot more time... It has definitely been a time saver.

I was spending a lot more time writing PowerShell scripts by hand that were not commented nearly as well, nor had nearly as much error checking and fault protection in them. They were purpose-built to do one thing and to only account for the scenarios I encountered. They were not nearly as robust as they are now that I have Copilot. Yep. The other thing I was doing before Copilot was I was manually looking through a lot more data. Conditional access policies.

I was manually reviewing each conditional access policy, jotting notes, trying to figure out where there are gaps, where there are discrepancies, health checks, audits of Microsoft 365 environments. I was clicking through the UI, I was running some PowerShell scripts, but I was just getting a lot of data that I still had to manually look through and look for misconfigurations, that type of thing. Now that I have Copilot, I'm getting a lot more data and letting Copilot do a lot of the hard work of analyzing the data. Oh, you're letting Copilot do it.

Now that there's a normie in front of us, you're going to call it Copilot. Well, he said before Copilot came, what if I sent Copilot stuff to the anthropic models? You could have just said generative AI. I use Claude. Yeah, you're right. I should have said generative AI. Yeah, before generative AI, because I do a lot more of that with Claude, because I will end up with like a thousand. Well, that's for a different scenario. I will end up with a lot of files in my file system, and it's a lot easier to use like the command line in Claude to work with those than to upload a bunch of them to a SharePoint site. Well, what about you, Jay?

What was life like for you professionally before generative AI? I spent a lot more time doing boring parts of my job than I do now. I spent a lot more time reading emails and trying to find content, like you said, trying to search for stuff. I considered myself pretty good at searching for things, but it was a significant part of my job. I spent a lot more time creating documents. When I worked for AvePoint, I did a workshop for the U.S. Army and for Lon Lane, the sales guy.

And it took me two weeks to get my deliverable from that engagement out to him and the customer because I was just so busy. I didn't have time to spend six or eight hours building this deliverable. And now that I've got Copilot and Cowork and generative AI, I can knock that out in under 45 minutes. And it's a better deliverable than it ever was when I was doing it by hand because I wasn't relying on my memory to write it. And I wasn't relying on the hours of search to find the documentation for it. I now take a transcript and I give that transcript to a Cowork skill, which knows what the document should be formatted like.

It knows to search learn.microsoft.com to find documentation on the major topics. It pulls all of that together in between the transcript, which replaces the need for my memory to be so, well, I've got a terrible memory. It's really completely changed how I do the menial parts of my job so I can spend more time either tinkering, which I had no time to do when I was at AvePoint because that's when Copilot became a thing for me is when I switched to Microsoft. But I now have time to tinker, like so much time. I'm writing skills and generative AI things that I just didn't even have time to think about before.

That's amazing. I want to hear Norm's answer and then I have some interesting thoughts around what you said and some of the stuff you said versus some of the things I'm feeling. He's good. He's feeling. What? He's what? It is his turn. I agree. Thank you. We actually have an audience here. And if I had the camera, I would show you. I think most of my time pre-generative AI would have been focused on thought organization, collating content in a way that even though it made sense in my head, it was never going to make sense to anyone else.

Yeah. And it has become, I don't want to use the word game changer, but it has significantly made my content more consumable for people. I am able to just brain dump at all times and have something else organize it. So the quality of my work has probably improved. The value chain has, I've definitely moved up as a result of it. The but part, and it was the thing that I wanted to ask both of you is, Ben, when you first started talking about your pre-generative AI work, a lot of it sounded like it was based on your experience, your trained knowledge, and all of these things that would make it possible for you to go through those autolog files to perform that health check and so on.

You can't assume that generative AI knows those things. It knows a lot of things, but it doesn't know those things. And I was going to tie it back to what you said about the two weeks it took you to generate a deliverable for a very important customer. Now you can do it in the span of hours instead of weeks, but it's still not the same. And I'm not suggesting that generative AI is better or worse or whatever. Do you feel that the Ben factor and the J factor is still representative in your current day work and your outputs? Yes. And I would say because I will generate those a lot quicker.

That being said, I will spend a lot of time reviewing it yet. Okay. I will generate, I will have it analyze the data, but I will still read it line by line. And there are times, to your point, normal, I will read it and I'm like, that doesn't feel right. That doesn't look right. Like there's something that because of those 20 years of experience, there's something off here. And I need to go dig into it further or validate the source or something to that effect.

Or even PowerShell scripts, I will go review them and look for certain things. Like if I'm writing a PowerShell script to go get a bunch of information, I will go through it and make sure there are no update dashes or add dashes. Or those types of things. I want to make sure everything is a get, a read, something to that effect. So I think there are certain things where it's quicker, but it's not exponentially quicker because I'm still spending a lot of time on the validation of what it generated.

Got it. What about you, Jay? Like you're, you're spending all of this new free time building things like skills and having time to tinker. Do you think there's a way for someone like myself and others who are, I don't feel like I'm losing my voice or my, I'm not losing my voice to AI, but I'm, the lines are blurring. And I still want to make sure that my authentic self, my approach to problem solving or articulating solutions to users, customers and so on is still there. What does the future look like for us?

Yeah. And I ask you in particular, because you're like, you're at this table. I think you're the one who's probably the most informed at where the Microsoft technology side is going. I think there's still a lot of unknown in that. Part of that is society and government are still trying to figure out how to influence where the technology is going. My eight-year-old stays up until 10 p.m. after being sent to bed talking to Alexa about life. Literally, he had a conversation one night about how unique am I as an individual.

And Alexa would ask questions about his eye color, his hair color, his height, his weight. Are you skinny? Are you a little bit bigger? And Alexa helped him figure out, like, he is one in one blah, blah, blah, blah, blah, billionth unique. Like, it was really cool. Like, it was also kind of creepy, but it was really cool. My 14-year-old hates AI, which is understandable. My 14-year-old is a cosplayer. He considers himself an artist. And those are areas that AI is really encroaching negatively on.

It's using publicly accessible art to be trained. It's pulling. People are using AI videos to, or AI to create videos of themselves in cosplay rather than putting the work into doing the cosplay and then claiming they did. And so my 14-year-old is very against it. The 8-year-old is growing up with it, which is really interesting. It kind of feels like phones were for, mostly for all three of us, where, like, at one point, it was take a quarter so you can use the pay phone.

And then it was take a quarter so you can use the pay phone when I paid you. And then it was, here's a phone. And then it was, here's a phone, text me. And then it was, here's location tracking. And, like, we lived through this change. And the 14-year-old is living through that change. Meanwhile, the 8-year-old's like, what? No, AI's just in everything. So I think we're still, as a society, and then through the government, trying to figure out what the technology should look like. Yes. So it's hard to say what it will look like. If you're managing file storage today, you've probably felt the pressure.

Refresh cycles are harder to plan. Costs and complexity are increasing, especially as file workloads scale across regions. That's why Microsoft customers trust Nasuni. As an Azure-certified software, Nasuni provides a global file system designed for enterprise-scale environments, keeping file data secure, governed, and accessible across Azure environments without added infrastructure complexity. It's the foundation to keep your unstructured data stable, even as your infrastructure evolves.

To see how much you could save while keeping your initiatives on track, visit nasuni.com slash TCO. That's N-A-S-U-N-I dot com slash TCO. And your question is interesting, Norm, because I'm still bringing my experience to it in my history. But I do feel, and that's what, similar to even what you started alluding to, Jay, where you were like, it takes away the boring parts of mine. In some respects, I feel like it took away some of the fun parts of mine.

Because I enjoyed, similar to your 14-year-old, I enjoyed writing the PowerShell script. I enjoyed some of the creativity aspects of writing a script, of problem-solving. How do I write a script? How do I account for this error? There was a lot more thought that went into what I was creating in a PowerShell script. Now I'm to the point where I legitimately, and to your other point, I don't get to free up my time in Tinker because I'm billable.

And this is a conversation Scott and I have had. Now it's like, I feel bad charging a client for spending eight hours writing a PowerShell script when AI can write the exact same PowerShell script in 30 minutes. I can't now go bill eight hours in Tinker for seven and a half. I'm trying to do more and more and more work because the work I'm getting done happens faster and faster. So it's also brought about an interesting aspect of how do I bill for some of this stuff?

Can I still do a pure T&M? So I have a whole bunch of those thoughts where there are days where I do struggle with it because I feel like some of the fun, some of the creativity, some of the exploration of an audit log, the mystery of how did this happen? Let me dig through and problem solve is going away. And I've watched some other people talk about that's going to be a danger of the younger generation. I can still spot stuff because I've done that.

How does somebody growing up now that's eight, nine, ten, AI isn't everything, are they going to be able to think as logically through problems and problem solve? Is it going to matter? Like, what is that going to do to the whole, even the brain development of younger generations? Absolutely. And I think the pendulum is still swinging. We don't know where the line should be where you know the human must be in the decision-making process.

When you know where that point is, then you'll know that, yes, you still need those diagnostic skills that would let you spend hours writing a PowerShell script or going through an audit log to do forensic work. That still has to be there because right now, AI is just creating velocity in our work. And things at high speed will not always have the same quality. And I hesitated there and I corrected myself because knowing that it's going to be based on a large corpus of reference material as it's doing these things. But when we loop all the way back to what you did prior to generative AI coming out,

your experiences may not be representative in that learning base. Your customers may want Ben J and Norm and not vendor of choice you're using for AI. Right. And I didn't ask these questions to go down like this. Existential crisis. Exactly. Philosophical. This is very much a philosophical IT discussion. It is. No, it's wonderful. Because we must concede that we are all pro-technology being at an event like this in Branson, Missouri, advocating for technology and how it can help.

But at the same time, realizing that it's still early days. Oh, yeah. No one is an expert. Some people just have a little more experience than others. And we're still trying to figure out like, what does this look like for the next generation to your point? And are we going to be able to read an audit log anymore? Will audit logs not exist? And I don't believe that's the case. I think it's going to be right-sized at some point when it's just like too many mistakes are going to be made because the models are just, the human in the loop is being too lazy or something like that. But I don't feel like my job's in jeopardy. It's only in jeopardy if I don't adapt to this new tool set.

So I love that you brought up the analogy, Jay, of the quarter to the payphone, to moving to the pager world, to the cell phone world, and so on. This is where we are. It's just a new tool. I have a buddy of mine who is an Amazon fulfillment center operations manager. He manages the 6,000 person fulfillment center that loads the trucks to be able to go out and do deliveries. And he also is a local political leader in his community.

And the conversation I have had with him is really unexpected to me. I'm a knowledge worker. It's where I live. He said, the fulfillment center they're building three towns over is bigger than mine. And because of technological enhancements, it will only have 1,000 people. And he goes, and if you think about that, where do the other 5,000 people go when that job is automated? Because these are not all people who can go just skill up. They're not all people who have, they don't necessarily speak the local language very well.

They may have had skills when they came to the country, but have not worked those skills in a long time. They can't all go become mechanics or they can't all go work construction or in a kitchen because eventually those industries don't need more people. That's right. And so that's a big part of the philosophical side of this discussion. is how are we supporting the community? Because I've always said to my customers, yes, AI may replace your specific job, but it's

up to your company and you to make sure you're staying ahead and that you are learning new things and that you are putting yourself in a position to take a new or enhanced role. So AI doesn't replace you. It helps you go somewhere new. And his response to that was, that's great, but what does the cleaning person at my facility go do? What is the person who literally stands in front of a conveyor belt and checks a box? That's right. Like, that's not something I can just quick skill up. And I don't have an answer for that.

Neither do I. And I bet if you talk to people and their guard comes down a bit and ask them, are you having like an identity crisis with all of these effects of generative AI and how they're affecting your current job? Like even in my own space, like I like to focus on power platform and I can see a time when some of these apps and services, you're probably not going to need someone to do it. You just say what the outcome is going to be and the rest just kind of takes care of itself. Yep. So do I continue to invest or do I embrace generative AI? Do I start? Like, how do you?

I heard your big sigh. So I looked at you. Oh. Did you not have a big sigh? I don't know that I had a big sigh. Are you having an identity crisis? Or I did. I am having an identity. I feel like there are certain days I absolutely feel like I'm having an identity crisis of where, again, going to, I feel like some of the fun things of what I used to do are disappearing. I'm not experiencing that problem, but I can totally see that it exists. And I love building skills. Like, I will not lie. I have had a ton of fun building skills for Claude to go analyze audit logs, to do in-tune

security reviews, like coming up with prompts to say, go look at how this tenant is configured and figure out how you would compromise it based on current known vulnerabilities out there, based on attacks you're seeing. Like, building skills to go do some of that work that would literally take me hours or days to do is fun. Yeah. But I can only build so many skills and clients are not paying me right now to build skills.

They're paying me for the outcomes. So it's, yeah, I have fun doing some things, but I'm losing the fun in other things. And the things I'm having fun doing are not as repeatable. Sure. As building PowerShell scripts or digging through audit logs or trying to find some error buried deep within a log somewhere. Yeah, that's fair. So you brought up tinkering with skills. Now I know Ben's building them when he can. If you're one of our fellow IT professionals out there and you were thinking that you're suffering

from this impending identity crisis about what your previous work was like and trying to reconcile it to what it's looking like today and in the future, would you say like creating generative AI skills is probably like this, one of the safest first steps you can make to start future-proofing yourself? So for me, creating an AI generated skill is not more than documenting the process which I use to do something. It's not necessarily, and this is where things will get interesting, is every ounce of my

25 plus year career is going into the definition of those skills because it's the process I use to do things. That's the J. That's the J factor. The kid coming out of college can't build that skill, can build a skill, can't build that skill. And so it is a really interesting time where the people with experience, if we can latch onto AI, this AI wave, we have the opportunity to make sure it's doing things really well.

And then if we are actually concerned about that younger generation, we have a responsibility to share with them, how are we doing this? Because they have to see what it takes to be able to figure this out. And the response might be, well, by the time you retire, old man, AI is going to know what you know. And to a degree, that's not wrong. But as I like to tell my customers, AI does have the potential for infinite knowledge, but it has no concept of how to pull that knowledge together into a thought, into a process.

It really doesn't. And so that's where this experience comes in. So yeah, youngster, you're right. AI will have lots of my knowledge, if not all of it. Maybe one day we'll figure out how to data dump my brain. But the things that connect that knowledge, that's stuff that it takes experience. So it's not a check yourself moment. It's a recognize that you as a person still matter because you as a person, that inference you bring to the table is something AI cannot do. The technology is not there yet.

And I don't know when or if it will be there. So yeah, you matter. And you got to put an effort in to be able to then have the J or the Ben or the norm to put into the AI so it can do cool things. So that could bring about a whole nother conversation of, I'm also curious, do you think AI will continue to improve at this rate or do you think at some point in time will actually start regressing due to what you said, a lot of the AI that we have now, and I've heard this argument a few

times now, is because the knowledge that was used to build AI is the J knowledge, the norm knowledge, the Ben knowledge, from blog posts we wrote, from videos that we created, all of this human thought created content. And Claude just came out and said they're starting to watermark their content now so that it can be identified. And this is a conversation Scott and I had is one thought behind that or one of the reasons they do that is if Anthropic trains its new models on content that Anthropic created, the

models start regressing. So as, and I've seen stats about the amount of content online that is generated by AI, are we going to create so much AI generated content that is watermarked that AI is going to continue to improve models or it's going to start training models on what's available, which is AI created content, causing it to actually start regressing to the point where we're going

to need humans to start creating real content again. Well, what a great question. And the way you frame it up, it does seem that we can expect a plateau of performance. It's going to have a bias based on its own generated content. And so maybe this becomes the new era, the transition point where human generated original content is going to be the most important factor. And it's less of just creating velocity and outcomes and more about making sure that the

information that it's based on is still going to be relevant for today and the changes in technology. Sorry guys, I'm... It's okay. It's a heavy conversation. Yes. It is. It's fascinating. It's your point. It's so new. We don't know. Yeah. We don't know. We don't. I'm intrigued by the idea of the knowledge plateau. I'm intrigued by the idea of regression of the AI just starting to learn from itself. I don't know that... I don't necessarily think we'll go backwards so much as slow down a little bit.

I think that a couple factors will cause that. One, the recognition of the content being its own generated content. As we start to see that, we'll start to see the patterns that come out of it. But I do think that regulatory controls may start to slow things down a little bit, at least for a little while. But also, we aren't going to stop generating our human content. We may generate much less. I don't see it ceasing. The trick is going to be separating the wheat from the chaff, the human content versus the

AI content. And if companies like Claude are going to start putting watermarks on it, that's going to help us be able to do that. I can see the slowdown coming. I don't know about a full regret. I think we also might start seeing a slowdown with costs. We've talked about this already. The fewer costs and the amount of money people are spending on it. Yeah. Can that continue at its current pace? I mean, at this point, we're... I won't lie. My stock market investments have loved the pace of this. Oh, sure. Who hasn't? But it's... Waiting for that bubble to pop.

I have definitely been interested in the conversation of whether or not this is a bubble or as we saw in the earlier 2000s and such, the late 90s. But yeah, Norm, I'd love your thoughts on this one. I don't play the stocks. I just know that these announcements can be very volatile to the actions on Wall Street. Every new announcement seems to be the next thing that everyone must have if they want to be on the generative AI frontier. I pause because I'm still thinking about the state that we're in today, what it looks like

in two to three years from now. Yeah. Is it going to keep costing as much? I don't think so. I think that because there's going to be... There's possibly going to be this plateau regression. I don't know. But also in a plateau, maybe there's an optimization of that current knowledge base. That means it doesn't need the same compute. That means that it can be more distributable. It means that it's more trustworthy, more applicable to daily operations. And people actually know how to use it and aren't always going to the most expensive, the most complex option inside of those generative AI tools to always try and get the best when

they don't need the best. Because I still don't know, am I using option one through option five when I'm using that model picker and whatever AI tool I'm using? I still don't know the difference. But conceivably, there's a time when that happens or it's obscured from the user. And there's another layer that is making that right decision for us in a consistent way, in as consistent as we can have in generative AI. And then all of a sudden, maybe things are more consistent. And as we get consistency with how people use AI, then the spend becomes more consistent,

predictable. And it's not less of this thing where everyone's racing to save their business, improve their skills, become the next great thing, jump on the bubble if it were in a bubble. And it's just another tool in the toolbox. Yeah. I definitely think we're going to stop talking about AI soon as a capability where we're very close to everything having AI inside of it. Like every technology having AI as a significant portion of what it is. Part of it as a tool within that product. Yeah. So we'll stop selling AI is what I'm trying to say.

And at that point, yeah, I think there's going to have to be an optimization for it. Because I'm a big believer that if you want to build something a certain way, you should have the ability to do so. And that's one of the things I love about AI Foundry. You have that full control. You can pick the model. You can pick the machine learning algorithm. You can do all of that stuff. But I'm a bigger believer that it should be abstracted from the user. And that's what I love about how Copilot works is it's on auto. It selects the right language model based on the way it thinks it should.

And it's trying to optimize. Frankly, it's trying to optimize because I'm only charging you $30. Hey, Hat. I'm only charging you $30 a month to use this tool, whether it costs you $1 or $50. And so it's in our best interest to build the facilitator to figure out what is the best tool for it. And if you're a power user, if you have a preference, go ahead. Select you want to think deeper or select Claude 5.1, whatever. I think we're going to stop talking about AI as a capability and we're just going to keep talking about the business that we're trying to run.

I would not mind that. I am ready to get done with every announcement. And not just from Microsoft. I think this is across the board. Every announcement being, AI can now do this. AI can now do that. I'm ready to get back to, here's new features. Yeah. And to your point, AI just becomes a natural part of it. Yeah. It's a natural component or a natural feature. I think kind of like identity. Identity is a natural feature, a natural component of just about any solution. Yeah. AI gets to that point and we start getting back to what are tools we can build, what are features we can build for the business.

How can we improve SharePoint where it's not adding AI to another aspect of it? How can we improve the power platform without AI being the thing to build my latest flow or power app or all of that? But what are some of those, getting back to what are the features we can build to make these products better for the business that it's not a focus on AI. AI is just there. Yeah. That's right. All of this should just be an enabler of an outcome, usually a business outcome. Yeah. The real person signing the checks, like the CIO, they don't care what feature's there.

They want the outcome. Yes. 100%. We do. Our trusted advisors at companies like Microsoft, they care about the features, but we just want outcomes. Or the business users really just want outcomes. And having a, the more abstracted the features are, the better it is for them. Yep. Yep. Absolutely. Wow. We went deep, boys. Yeah, we did. And Jay needs another drink. And Jay's all choked up about it. Yeah, Jay is all choked up. I think Jay probably needs to take a break here. Yes. Norm, we should have you on again tomorrow. What a pleasure. All right, boys.

My brain's going to have to go recover before we have Norm out again. Yeah. That was... Too deep? No, it was beautiful. It was a great conversation. It was not too deep. It was refreshingly deep. Yeah. Yeah. If there were still an audience here, I'm sure they would cheer loudly for us. Oh, I'm sure we'll get at least one like and subscribe. Yeah. I mean, actually, we need to start saying that. I forgot about that. Oh, yeah. Like, don't forget to like and subscribe. If we start saying that, I might mute your mic because... You don't want me to like things? I would love people to like and subscribe. I am so tired of hearing that in every single YouTube video that I watch.

Because, frankly, even for the podcast, I do it because I want to do it, not because I want another like and subscribe. All right. Well, don't like and subscribe. Go tell your friends how amazing this was and catch us up tomorrow. Yes. Thank you both. Later, y'all. If you enjoyed the podcast, go leave us a five-star rating in iTunes. It helps to get the word out so more IT pros can learn about Office 365 and Azure. If you have any questions you want us to address on the show or feedback about the show, feel free to reach out via our website, Twitter, or Facebook.

Thanks again for listening and have a great day.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Episode 441: Microsoft Ignite 2026: Autopilot, Workspaces, and Spot Plus

    Nov 19, 202648 minAudio

    Welcome to Episode 441 of the Microsoft Cloud IT Pro Podcast. In this episode, Ben and Scott take a first pass at Microsoft Ignite 2026, starting with a question a lot of IT pros are asking: is the pace of change...

    Credit the people on this episode

  2. Episode 439: Azure SRE Agent vs. the Azure Copilot Troubleshooting Agent

    Oct 22, 202638 minAudio

    Welcome to Episode 439 of the Microsoft Cloud IT Pro Podcast. In this episode, Ben and Scott dig into two Azure agents with confusingly different names: the Azure Copilot Troubleshooting Agent and the Azure SRE Agent....

    Credit the people on this episode

  3. Episode 438: Taking your WorkOS and LLM Wiki to the Next Level

    Oct 8, 202647 minAudio

    Welcome to Episode 438 of the Microsoft Cloud IT Pro Podcast. Back in Episode 429, Ben and Scott introduced LLM wikis. In this episode, they come back a few months later to compare what those wikis turned into after...

    Credit the people on this episode

  4. Episode 437: Nasuni and AI-Ready Unstructured Data with Nick Burling

    Sep 24, 202640 minAudio

    Welcome to Episode 437 of the Microsoft Cloud IT Pro Podcast. In this episode, Nick Burling, Chief Product Officer at Nasuni, joins Ben and Scott to talk about getting unstructured file data ready for AI. Nick digs into...

    Credit the people on this episode

  5. Episode 436: Finding the Entra Tenants You Didn't Know You Had

    Sep 11, 202638 minAudio

    Welcome to Episode 436 of the Microsoft Cloud IT Pro Podcast. In this episode, Ben and Scott dive into Microsoft Entra Tenant Governance. Almost nobody runs one Microsoft Entra tenant. Mergers, divestitures, dev and...

    Credit the people on this episode