Techstrong Gang – August 19, 2024

Hello, everybody. I'm Mike Bazaar. Today we're talking about, well, AI image generators and Elon Musk and all kinds of interesting juvenile behavior.

And we're gonna talk about, well, is California gonna lead the way? And maybe helping to reign in some of that behavior. And finally, we're gonna have a little chat about what is the implications of AI for managing DevOps and AI teams in general.

You're watching Textron Gang, we'll be back in a minute. All right, folks, we're back and we got a full house today. So let me introduce you to everybody who's joining us.

Um, from fu we have Paul Nati, who's the practice lead for application development in DevOps. Paul, good to see you. Hey, great to be here.

Also, we have returning as always, from Denver, Mitch Ashley, who's working both for Futurum as a, uh, chief Technology advisor for DevSecOps and Security, and, um, also works for us at Techron as our CTO. Mitch, good to see you. Good in both camps.

Good to be here. All right, finally, we also have John Schwartz joining us from California where he's our eyes and ears in the valley these days. John, how are you?

I'm good. Good to see you all. And then we have joining us, of course, John Willis, who is our resident DevOps AI expert these days, and of course, has a long history in the DevOps space.

John, thanks for coming. Hey, hold on. And then finally, Tracy Reagan, who's joining us, once again, I assume from the desert, but, uh, Tracy, I feel like if it's Friday, it must be Tracy Day.

So here we are recording. It's, I look forward to my Fridays with the gang. Here we go.

All right, well, of course most of you are watching this on Monday, but we need some time to actually record these things. So that's how it works. Um, otherwise known as the magic of television.

Yeah. Let's jump in here. And I guess I'm gonna go to John Schwartz first.

Um, we've been watching the rise of these AI image engines, I'll call 'em. Um, and people have been playing with this stuff for a while, but in the last week, uh, Eli Musk launched Rock Two and started playing around with it on Twitter. I guess you can get at it if you have the right level of Twitter account.

And suddenly, I think maybe it's just me, or maybe I'm just noticing, but there seems to be like junky content everywhere now running around because people are creating these images. And Elon Musk, of course, was at the forefront of that, creating some things that were, uh, well, depending on your point of view, tasteless, I don't know. Oh, yes.

I guess some more about tasteless stuff from from you on. Yeah. Yeah.

But yes. Um, so I was talking yesterday with a AI animation studio Called Tune Star, a couple of people who just left Warner Brothers, and they're creating all sorts of content that's incredibly popular on YouTube, TikTok, and Snapchat. And I mentioned Rock to them, GR two, and they, first of all, they said the name sounds like something out of a villain, right.

Villain movie. And they, they also mentioned the danger of kind of going unbridled with these images. And you know what?

It's the Wild West on X, right? I mean, we talked Friday on the show about Musk and X playing an increasingly large role in things like politics, what have you. And, um, I can take a, a cursory look at what people are doing in the rock's fun mode.

And among other things, they're generating images of, uh, Kamala Harris as a dominatrix standing over Joe Biden. Donald Trump is Rambo and Mickey Mouse driving a Tesla. It's kind of boomerang on Musk, because some trolls have him depicted images of him taking part in school shooting.

So it's going completely off the rails, right? And the, the concern is that this is going to play a role mainly bad across all sorts of things, not just politics. I mean, even the FBI, they have quietly resumed working with social media, which includes X by the way, uh, Facebook and YouTube about monitoring disinformation from foreign nations.

So this is all playing together, and it's gonna lead to perhaps legislation. This might have been an influence on something we're gonna talk about later, but it's, it's, it's, it's really kind of a ticklish mess. And it also plays into things like intellectual property and rights and imaging.

It's, it's just, uh, he's opened a can of worms, but I'm sure he's delighted by doing this. Yeah. John Willis, most of this stuff seems to be focused on images, but I've seen stuff where these AI models are gonna create videos and entire narratives and maybe who knows a movie one day.

Where are we on this journey here at this moment? I feel for the last year we've been talking about AI models in the context of creating text and code, but is this the next phase? Yeah, no, I mean, this phase started a while back.

And, you know, I mean, I think, um, you know, uh, you know, started with open ai, really, I mean, at least from a commer, you know, I don't wanna say commercial, but the general public understanding how this worked. It, you know, it's been going on for quite a long time. But, uh, in general, the, the, the ability to match words to, to videos and, or pictures and images.

But anyway, but, but yeah, I mean, you had sort of the era had, Dolly had stable diffusion, um, which, you know, brings on mid journey and then mid journey, and then, uh, but the, the, you know, the good news was that a lot of these, um, players are at least trying to attempt to be good citizens, right? And, but, you know, the, the, I think the alarming thing about Elon Musk in general is he doesn't seem to understand the idea of governors and stuff like that. So I think the, you know, the, the, the new sort of rock stuff is, is quite scary, to be honest with you.

I mean, they're already, I went out and I looked at sort of on x, that trash bin that used to be amazing. ai corporations references to, uh, crock too. Um, so, and the video stuff, I'll be honest you, Micah, I haven't, I hadn't gotten that deep into that.

I know there was, there was a real interesting, um, I think was in the last month or two, uh, uh, like a 32nd video that blew everybody's mind. Um, you know, which is, was quite scary for Hollywood and everybody like that, that it really was sort of a, a 32nd full length movie, you know, movie really fully generated with, uh, generative ai. But yeah, I mean, the, and the only other thing is the gr two thing's interesting because it's already come out.

Uh, it's like they, they, they put out a stealth version of the model, and it, it showed up, it shows up fourth on one of the leading, uh, leaderboards for, uh, you know, you know, basically tied with, you know, all the sort of mini versions. The, you know, the, the 2, 2 4 mini the Claw three five sonet. And so right outta the gate, it's our top five Oh, the leaderboard Board.

Well, I think it would speaks really to our culture. I mean, if we wanna blame somebody, let's blame the, the, the supermarket tabloids. They've been doing this kind of stuff for how long, right?

Mm-Hmm. And now we're just seeing it on us in a social media perspective. And if, uh, I guess Elon Musk has decided that X should be like a supermarket tabloid, and that's the market he is targeting.

Um, so we, every individual have a responsibility. Do we wanna open up that tabloid and start looking at the, at the cesspool of crap that comes from X? Now, I mean, I, you know, we have made a decision, but we get off of, um, XL together.

I'm slowly migrating us off, and I'm gonna make a final announcement to say, Hey, we're done. Um, but it's a culture. And every single one of us, every single citizen in this, you know, social media world, we have to make our own decisions about what we really wanna see.

Uh, and do we regulate that? I don't know, have we rev regulated the supermarket tabloids? Never.

So how is this any different? Um, you know, I, I don't know. I don't know if it's different.

Well, well, I was watching this interview with Justine, uh, Bateman, who, uh, back in the day was a sitcom actress, but today is a director. And she was saying that we are gonna see a massive amount of what she described as throwaway content. And some of it will be created by people and others will be created by studios.

And, and from her point of view, it will be disposable. And eventually people will get bored with it sooner than later. She's predicting, because a lot of it's just gonna be maybe slightly better than the junk you're already seeing.

And, um, and her other point though is it's gonna force the people who are trying to make money in this space to produce better content that's more interesting and more compelling. So, um, John Schwartz, I know that there's a lawsuit floating around on this stuff, but what's your take? Yeah, Well, you know, it's interesting that you said that.

And, and totally bravo to John and Tracy for calling out what X is. It's, it's a trash bin. It's based, basically turned into our latter day, uh, national tabloid, you know, those, those things are kind of just gone away and for the most part, like news of the world.

But yes, this is interesting what's happening in terms of content creation. There used to be an era, remember, this is a long time ago, but Disney, Pixar, et cetera, all the studios would try to make for the, it often try to make content that was enduring that would last. And now we're in an era of AI where everything's hyper accelerated, where you create this junk, maybe it's just for 15 minutes, you know, the Warhol to, to CIP from Andy Warhol.

You create 15 minutes of, of memorable junk to influence the, the politics of the day or the hour, right? Because we can't remember what happens in this new cycle, these new cycles from day to day. It all seems like it's taken place a decade ago.

And, and so what we have is we have people just spitting out garbage influencing folks who are, may not be educated on what really is reality. And it's terrifying to me. And Justin Bateman is, is spot on.

I mean, it's gonna, maybe it'll, it'll spur others to make valuable content, but, but my fear is that it's just gonna be a huge vacuum and huge hurricane tidal wave, tsunami, whatever you wanna call it, of terrible content short term with, with an agenda. Now there's, it's not isolated to the two to social media. I am still furious, furious with Palo Alto Networks and their lampshade.

Oh yeah, we talked about that on Friday. How is that, how is that? It's not any different, right?

It's exactly the same there, except there were real people sitting in a professional environment with people who look like me walking in the door seeing that and probably turning away, going back to their hotel rooms. 'cause that's what I would've done. So, you know, this is our culture right now.

We've been doing this a long time, and I'm, I think I'm older than you most people on this call. But you remember that there was the days back in the large conferences where it, it was all about that. I thought we'd gotten rid of that stuff, right?

I thought it was like gone, right? Like even about eight years ago, some vendor did something like that, and they got hammered and they apologized, but the, the See Palo Alto do that, it was just mind blowing that An apology at this point is not enough. What needed to happen was the CMO needed to get fired, and the CEO o should have resigned.

That's how, that's how furious I am. I'm gonna, I'm gonna, I'm gonna call it into the Palo Alto conversation. We did that last week.

So let's just keep moving. Just an aside, I'm gonna meet with the CMO from Palo Alto Network supposedly soon. And I'm gonna bring it up because that, I thought we, as you said, John, I thought the arrow of the booth babes was long over, but around Last.

Hey, Mike, the last word on that, Mike, to kind of take us back to the conversation, one of the things I'm curious to see is, this reminds me of when we first started seeing synthetic characters being generated in movies, right? Mm-Hmm. Um, and then what we realized is there were people behind it to really good and authentic, you know, a character in a movie, you have to have a person who's doing, you know, wearing the, the green and the, and the, the tennis balls or ping pong balls on their, on the Mexia Act, act out.

And, uh, that shows up in video games today as well as movies. If AI can, can accurately, um, synthesize and generate video of people interacting and walking and doing things like that. Maybe that's the next generation of, of movies of film where we really do have totally synthetic characters.

Yeah, Mitch, I mean, it has the potential. Yeah. Mitch, I, I agree with you.

I think that when you look at the tools that are in front of us, I mean, I, I basically use social media for business purposes only pretty much. And, um, as last I checked, most companies, uh, you know, use very specific platforms, and X is one of them. Um, so, and because of that, um, I, you know, that's what, that's a vehicle for communication of this research and this information.

I, but Mitch, I agree with you a hundred percent, that if there's a way to make it more efficient to drive the better results to have, have the right content created, as long as it's sourced properly, and it says that this, there's a disclaimer that says this is what it's being used for and how it's being used, then it's, you know, we're all, uh, responsible. As I said on previous episodes, organizations are accountable for their actions. Right.

And if you use information blindly, then that's not a good way to do it. Right? But I, but I, I like where you were, where you were going, Mitch, of tying it back to there's some skill behind it.

It's not just, uh, it's not just kind of blindly just creating stuff, you know, in a vacuum, right? It's kind of like, uh, Photoshop on steroids right now, right? Yeah.

Everything's, We'll get more nuanced though. I mean, Justine Beman was pointing out, we'll have movies in the future where you can just insert your image into where Luke Skywalker is and you can be part of the movie. And that part of the thing might be entertaining for folks.

But if you're gonna rewrite the entire movie, um, yeah. And then distribute it, you're probably gonna get sued by somebody. There's also a Go Ahead.

Yeah. I gonna mentioned, uh, really quickly that Krista Wolf, who is one of the, uh, principles of MySpace as a startup that's kind of doing that idea where they're gonna start with these very, very short videos where you insert yourself in a major movie. And so it is happening.

Yes. Right? But there is this other study out there that pointed out the following and, and said, you know, just 'cause we give tools to these people doesn't mean that they're creative.

And one of the things that it surmises is that eventually people who go create content are gonna create the same boring content over and over again. And there's just gonna be stuff that is variation on a theme, because not everybody is as clever as a Hollywood script writer. And so a lot of this stuff, will it ultimately be entertaining?

I don't know. I think we're micing port pretty quickly with the, you know, the Mitch movie that we sounds just like the John Willis movie that sounds just like the Paul Nti movie. So, You know, until AI can, can judge what is good creativity or not, what, what is interesting content, what is not it?

You Yeah. That's the human in the loop until that point in time. 'cause it's gonna generate whatever it's having access to for data as well as what we instruct it to do.

Doesn't mean it's good content. But, um, I mean, in terms of it taking over and, and doing things like that, I think that's the big gap, um, filling in that real kind of human element that, that we still need today. Alright.

I will take the other side of that argument though. It is more than probable that there are very talented people out there who could create things using these tools that would never get a shot otherwise, because the cost of creating the stuff is too high and they don't know the right people, and they can't, you know, they don't have the funds to spend two years creating this stuff. So we might see, you know, more interesting nuggets appearing in the stream of crap coming our way.

Yeah. That, that studio that I, that I was talking to has created this series called Stephen and Parker, which is getting, I think on YouTube, it's getting like 30 million viewers a week by comparison. They're telling me that a good week for the Simpsons now is like three to 4 million a week.

So there is an audience that's, that's an animated series that is based on good content that children like boys and girls. So there are positives among all the morass of crap. Tracy, you got any hope for us here?

You know, what's your sense of, are you just gonna tune out all together and just write code? I think that I tune out a lot, to be honest. I, I, I don't watch a lot of, you know, I, I limit my time on these social networking channels, but I think Mitch is right.

We're gonna get bored with it. We're, you know, we're intelligent beings, right? We're gonna get tired of the sensationalism of a meme.

Um, and it's just, it's gonna get, it's just gonna get uglier and uglier, and eventually nobody's gonna be interested in looking at it. There'll be a, there'll be a section of our, our population that will continue to wanna see it, and they'll market to those people. That's just who we are.

All right, folks. I think we've gotten to the bottom of this topic as far as it's gonna go for now, you know, Ilan, he's making a case for free speech and ultimate free speech, and you can do anything you want. And there's a lot of tech bros out there that kind of agree with him.

And then there's, you know, my copyright laws and all those good things out there that enforce my, uh, rights and courts are gonna sort this all out, folks. We'll be back in a minute. All right, folks, we're back in continuing this conversation a little bit around AI and its role in, well society.

And as it applies to content, there's a bill going through, um, the legislative process in the state of California that's gotten everybody in the valley taking note of, well, this thing seems a little more stringent than anything that's been talked about so far. But John, I know you looked at this. Walk us through what's in this bill and why are the folks in the valley starting to freak out a little bit?

Sure. I'll give you just an overview first, and then I'll give you an update. So, SB 10 47, which comes from a, uh, state senator named Scott Weiner, mandates a safety protocol to prevent AI misuse.

It includes an emergency stop button to shutter an AI model. Usually it's the large model developers are also responsible for testing procedures that address risk, as well as receiving annual audits from third parties. So there's this new state agency called the, called the Frontier Model Division in California, which would oversee the rules.

And if you don't comply with the rules, you're subject to a civil action levied by the State Attorney General of up to $10 million for the first violation and $30 million for subsequent violations. The problem is that there wasn't a lot of give and take between the state senator and the tech industry. So consequently, he and I received an email from his handler this morning.

They are making amendments. Does that sound familiar? They're making amendments to the bill and, and they're trying to kind of follow the model of the EU AI acts, which I guess they got ahead of the situation by talking to tech more deeply than in this situation.

Um, in a sense, one of the people I talked to in the tech industry said, you know what? All we really need to do is we need AI to write the legislation because AI quick cha moves so quickly that by the time this bill was written, it's obsolete. So they're, they're again, advocating this idea of self, uh, uh, monitoring, which, you know, how well that worked with, with the social media companies.

So we've got this whole ballyhood the boo haha, where the large companies are most exposed to this bill, and the smaller companies and the developers are up in arms about the impact it will have on them. So eventually we'll see where this goes. I mean, we've gone through this before, privacy laws in California.

Um, I'm sure this will be altered significantly in the coming days, but it's just another thing to consider as AI pers pervades its way through our society. John Willis, I know you've been following some of these, uh, legal conversations around ai. Where does this fit in this broader conversation?

Yeah, I think it's, what's really interesting is I went back to see, you know, there's this whole, all the debates, I'm, I'm, I'm identify, I've said this before. I'm, I'm about three quarters away, down was a book on the history of ai. So I really understand all the players going back from the forties to now.

So I wanted to find out what, you know, this what they call the, they considered the godfathers. I, it's Jeffrey Hinton, uh, Gio Bengio and, and John Koon, right? And they're sort of the modern day godfathers.

Their take on it is California first is not a bad idea. Mm-Hmm. Like, it's gonna start somewhere.

It should be California. Um, and, you know, and, and Bengio is basically, we can't let the big organizations grade their home own homework. But on the other side, the sort of the negative is what the woman they call the godmother of ai, which is Fefe Lee.

And if you hadn't had read her book, it's incredible, uh, the world we see. Um, and she is, she's probably more responsible for computer vision in modern times than anybody. And you could read her book, um, but she's negative and she's got some interesting points.

You know, one, is it gonna create interesting liability? You know, like, will the model, the person who sits the model waits, can they get sued? Right?

Like, and I think of like, she didn't say this, but think about autonomous vehicles, right? I think we've done a fairly good job on like, will there be a day where a coder of an autonomous vehicle will gets sued because a car makes decision? I don't know.

But, but I think it opens up a, like a can of worms that we gotta be really careful for people much smarter than me. Could it be an open source killer, is one of her points. Uh, will it kill research?

And last but not least is if you read her book and then some follow on books, um, is that she's like, they're missing the real point, which is bias, gender bias. Um, there is, um, you know, racial bias there, and these things are real, like, you know, people are getting arrested by facial recognition systems. And according to her, there it's non-existent in the bill.

So anyway, that, that was my sort of quick take of, they call this, Yeah, a lot of those things that you mentioned, John were mentioned by some of the folks I talked to. I talked to six or seven developers. Open source killer was among their liability, was another major issue.

Actually, there's a, uh, report that, that Mike passed on to me, um, by MIT about AI risk repository. And some of those dangers that are outlined in the MIT report are trying to be addressed in this bill. And again, you're right, California usually is the model for pretty good legislation in terms of, um, tech, because there's a national legislation to speak of and it's a model for other states, and they usually fall on the path.

Yeah, I think I, I, I really agree with Dr. Lee. Um, you know, I'm not sure this is the right direction to go for legislation only because where is the accountability falling?

It's falling on the model, not on the application that's consuming the data. Um, and I almost feel like the applica, it's, that's where it should be regulated. And it shouldn't matter if you're a hundred million dollars into it or not.

It should be who's accountable and who's accountable as a consumer of that data, of that model. So I'm kind of with Dr. Lee on this one.

Yeah, a hundred percent. And I know, Mitch, you had a, you had some thoughts there, but, uh, this a hundred percent behind that statement, because if it's accountability of the application creator and the organization putting it out, that has to, that's where the buck stops. It's not just because it was available.

It has to be, it has to have ownership. Because the other side of it is, is if it's regulation that's in California that may not be enforceable outside of California, how do you, how do you enforce it? How do you move it when it, when you cross state lines or cross coun country lines?

So I think that it comes down to the, uh, the responsibility of the organization creating the code. Again, I, i, I keep, keep on this thread in, in a couple of different sessions that we've had. We've, I've said the same thing where I think that the, you know, accountability has to fall with the creator, and that's where the, that's where you kind of, uh, go back to putting the, um, putting the, the constraints.

But I think her point though is that that's danger here. If we start sort of narrowing in on the creator, maybe the, the developer or the person who did the weight training, or, uh, we, we gotta have an incredible complex 'cause it's non-deterministic. We all know that.

Anyway, I, the way I read her is not like, let's nail the person who wrote this. It's like, it's going to be a can of worms of figuring out. And if we open up legislation that can sort of narrowly point to those type of lawsuits, we got ourselves really good.

You know, John, what this makes me also think of is we, we, there are analogies to this, another arts of our, our world, our, our business world. One is protection of gun manufacturers from being sued to the AI company. Go for that section two 30 for social media.

Those are the kind of regulations I'm also concerned about, right? So the, you don't want, you don't want, I don't want anybody running unfettered without being accountable for the content, the decisions, whatever. So I'm not in favor of either those piece two pieces of legislation.

But I think that's a danger too, is that we also legislate no accountability into this. And now you have protections. Um, Let, let me throw this out to John Willis for a second here, though.

There are people who are saying that much like we have the Department of Energy that keeps track of how nuclear energy is used and consumed, and, um, so far generally doing a good job, at least, you know, there's been incidents here and there, but do we need some sort of equivalent for AI where we're gonna have an agency that's gonna attract the usage of this and kind of stay on top of it and make sure that it doesn't run away from us? Absolutely. But we need the right people.

Right? And you know, like, so if, you know, if you follow Dallas World, you probably know John Alba and Richard Cook and Dr. Woods, right?

And, and the Dr. Woods was the guy who went in after three Mile Island. That was his first gig after he got his first PhD, right?

And like all about the way that he went in into Columbia Challenger. So like, I feel really positive you're bringing those type of thinkers who are doing systems thinking, not looking for who to blame, you know, blame, you know, and, and, and, and yes, we're gonna need that type of intelligence. The scary thing is new technology, new ideas, it's sort of the gravity goes right?

To who do we punish? Who do we, you know? And, and, and, and then over time we get a little mature about, about how, you know, how we do sort of postmortems and analysis of sort of huge, Yeah.

This, there's like a historical perspective for me, at least from the valley point of view, is that, you know, I think back three decades to when, uh, online gambling became a big thing. And there was this idea that there was a senator in Arizona and John Kyle who came up with a bill, a national bill to address it. But the reason why he was trying to address it was because there really wasn't an effective law in the us.

There was a tell there was a 1961 wire act during the Kennedy administration that wasn't really applicable. And eventually it took seven or eight versions of the bill to finally get it right. And I'm afraid AI moves so much more faster than where we were 20, even two years ago.

Generative AI in particular, that this is gonna be really difficult to thread the needle on something like this. And there's always that this big disconnect between the political side and the technology side there. It just, I, it, it, I just, it's so frustrating because it, I've seen this for years where they just can't interact and they don't understand one another.

They have different motivations, and it's just something that will continue to be a problem or an issue for decades to come. I'm, I'm, I'm afraid, Tracy, what's your level of confidence in politicians being able to sort this out? This, uh, hello?

I'm sorry, but, you know, you know, I've been in technology all my life and I struggle with looking at these bills and thinking when, and trying to think it through. And I feel like what we often do is we rush to try to do something. And when we rush to try to do something, we literally follow the logic of failure, right?

Because we don't get the whole, we don't get all the pieces and parts that we really need to understand and be able to, to, to manage it. Uh, and, you know, laws that are pushed through at this level with this raw of a technology, um, I think we'll make some mistakes along the way. But as John pointed out, you know, it's gonna take us some time and it changes so fast.

So again, I feel like these laws have to be about accountability and where the accountability should lay. And I guess that is a discussion. Is it in the models or is it in the consumer of the models?

Uh, and, and, and at the end of the day, somebody's gonna have a lawsuit that's just all there is to it. There's gonna be a lawsuit. Oh, yeah.

That's inevitable, right? Whenever, and they don't, that'll be addressed as part of the legislation, right? It's just, it's just like this cyclical thing that goes on and on and on.

Um, oh, one, one other thing I was gonna mention is that one of the most influential senators, right, arguably the last half century is Diane Feinstein, right? From California. I remember doing a story about her and tech and this, I mean, I guess she had a lot of responsibilities.

She was looking over a lot of different things, but she had no relationship with tech. This was back in the late nineties, and they had no relationship with her. And I thought, if that happens in California among Feinstein, I mean, this is, this is a, this is a problem.

Right? Well, you bring up an interesting conversation then, and I'll toss this to Mitch, I guess for grins. Um, should there be some sort of, uh, analysis of the AI tech savviness of people running for office?

Is this gonna become a part of the political conversation about what they know? Because, um, there's so much at stake. I'm getting a little uncomfortable with electing people who have no idea what this stuff is.

I'm not sure they can use email. So I don't think we can rely on their AI skills. I would much rather, you know, pull in a John or, or a Tracy or Paul and say, okay, tell us what's, why, what's the best thing to do?

How should we handle this? Have that debate amongst, you know, One, you know, one thing, Mitch, that's interesting. So I was, obviously, I live in a congressional district that's heavily tech oriented, but the two, uh, candidates here who are running in November have a, uh, they have their policy sheets around their website, and they have a separate sheet for tech policies in addition to everything else.

And I'm wondering if we might start seeing more of that just to sh to show your bonafides. I I think that there'll be a limited number of people in limited number of states or districts who will do that. But it's something that's starting to crop up more often.

At least that's one sign of encouragement to try to be positive. I mean, the last thing is, I want some AI entrepreneur running for Congress, and we're gonna rely on that one person, right? Yeah.

They happen to know the tech. I think, I think we've got, this is, this is like security. We have to rely on the best experts.

It's evolving, it's changing. And we have to, you know, help those minds get together and figure out what we need to do. What's the right thing to do?

Um, I guess you're telling me I should put away that let's draft John Willis for Congress memo? No, I, I, I was hoping what, because that would be a terrible idea. I don't think anybody here is running, at least not yet.

Alright. Is it feasible? And this will be the last question on this topic, but, and I'll toss it to John Willis.

Um, can we, and I've seen some efforts to do this, uh, and do we need to put everybody in a room and have some great AI summit where everybody is gonna have these conversations? 'cause it is so fundamental to our future, and yet I feel like our, all these conversations are happening in the hodgepodge fashion. Well, I mean, I think I, I mean, again, I'm trying to dust my memory off, and I'm not gonna try to google fast enough to remember what it was, but it was, it was actually a, a summit that California had where they brought in AI groups and maybe John Schwarzman, you remember a little more?

Yeah. What there was, uh, remember what Schubert tried to do, right? Yeah.

They brought a whole bunch of people in and the woman, yeah, they, They, they had, I think they had, I know they had three of them. John, the first one was just the large AI players, which p****d off everyone else. So they had to bring in Mm-Hmm.

Smaller companies because the large AI rulers That's right, right. They write the rules. Yeah.

Yeah. But I think that, that, from what I understand, a lot of this, it wasn't just like the, the, you know, the, the senator just said, Hey, I'm gonna create a bill. I think this has been a road that's been traveled a little bit with a little bit what you're looking for.

So maybe we're getting a little smarter to you about, you know, I, I don't have an opinion either way. I mean, I, I, I agree with Tracy more. I fall on the pay Lee side, but, but, um, but I, you know, I think there is, they, they're not just throwing darts on the board.

They, they spent some time trying to figure this out. So it's a decent model of what we're doing. So, And it does work to some extent.

I mean, the, um, open SSF did a lot of, and still do, they do a lot of work with Washington. They have conferences there. But the irony of that is, you know, I went to one of those and I sat in the room with Google and Apple and IBM and Microsoft thinking, these are the same companies that buried zero day vulnerabilities for years and years and years.

So, you know, who is, you know, is the fox guarding the, the, the hin house, so to speak? Good question. Alright.

So who's gonna be in that room? We gotta have the right people. I do think, Mike, back to your energy question, know, do, do you need a regulatory agency?

Well, in the energy industry, we have nuclear ES that get together at conferences and talk about the challenges, right? Uh, we have the National Institute or Energy Institute organizations like that, that help these conversations to, to happen. So I think there's some good models to, to look to of what we can do around ai.

All right, in the AI space, we have Elon Musk, freedom of speech, and boy, democracy is still messy. Hey, we'll be back in a minute. com is the number one online destination for DevOps education and community building.

com covers all aspects of DevOps, including DevOps, best practices and tools, DevOps culture, DevSecOps, business impact, continuous testing, continuous delivery, and more. com has the largest collection of original DevOps content featuring breaking news, blog posts, podcasts, and more. com to learn more.

com where the world meets DevOps. All right, folks, we're back and we're gonna shift a little bit of a gear here on this AI conversation, but it's really about how should it be structured in the age of ai? And it's an ongoing conversation.

com talking about, um, now is not the time to embrace no ops and let the processes run well, because, well, AI perhaps, as you've seen in our earlier segments, isn't everything it might be cracked up to be. And that then starts a whole conversation about whether or not we should have Chief AI officers. So we're gonna jump into that in a minute.

But Paul, what is your assessment right now of what can AI do in a, in an IT workflow? What, what can we rely on and what is, you know, aspirational? Yeah, that's a great question, Mike.

When we look at ai, there's this, uh, kind of, uh, I don't know, utopian view of saying how it's gonna solve the world problems. But, but the, uh, but the challenge that a lot of organizations are running into with regards to, um, operations and DevOps and developers and such, is there's really a maturity that, that organizations are, are kind of struggling with. Honestly.

What, what we see is, uh, in our research, we see from, you know, whether it's an SMB or an enterprise company, you have organizations that have IT lines of business, it DevOps, lines of business, IT DevOps, sre lines of business, IT platform engineering lines of business, and anything in between, right? And so there's just kind of a, a plethora of different kind of structures and methodologies that people are using. But what we find in our research is, uh, and I have done some trending data about this is year over year, we've seen a growth of organizations that responded that, um, that, that employed DevOps either in a limited or an extensive fashion, um, 71%, um, two years ago to 75% of respondents, uh, are increasing their use of, of DevOps within their organizations.

Now, the reason why that's important is because we talk about AI kind of being a way to, uh, you know, take over the operational side of the house. But, you know, we have to remember that many of these organizations, whether like as SMBs or or large enterprises, are still struggling with just general methodologies, not nevermind what tool stack to use, right? So, uh, that 71 to 75% growth, yeah, we see that there's a lift in and the adoption of using DevOps a good lift of using adoption.

I think it's necessary. We also see in our research that nine months ago, uh, production applications only 18% of production applications we use in ai, uh, in production. But we ran that study nine months later, that number jumped to 54%.

And, uh, what we also found in a recent study that, you know, 90 plus percent, 94% of, of, of production applications are starting to use AI in their, in their workflows. So there's definitely a, a pendulum swing in very fast to use AI there. And to, you know, echo what Mitch was talking about earlier, there's definitely a need to have a human in the loop.

So having a no ops environment is, is not the answer. Um, although what we do find is those organizations that effectively can adopt AI solutions are seeing a roughly a 250% increase in development speed. So they are seeing a, a significant lift in, uh, in pushing code out the door.

And again, because I'm an analyst, I keep throwing stats out there. The other stat that kind of comes to mind is many of these organizations, actually, 24% of these organizations are looking to push code out the door on an hourly basis, yet only 8% can do so. So if there is that desire to move towards a rapid cadence of releasing code out the door, there's, you know, there may be a desire to have a no ops solution and have a text X deal with it, but we're nowhere near that level of complete automation.

Uh, at least I wouldn't feel comfortable recommending that to organizations. And I don't think most, uh, CIOs would say, that's fine. We're just gonna let AI do its job.

And I don't even know what those tools would look like at this point with the, the DevOps platforms we have now. Um, we're talking about platforms that are very long in the tooth and they're not, um, they're not agile themselves. We can't easily, uh, build AI into the, into some of these workflows.

It's gonna take a real shift in the way that we approach DevOps for to doing two things. One, building AI into the DevOps process. We have to have the data.

We don't even collect it. It's in logs everywhere. And two, being able to bring in tooling to support AI code because it's gonna look different, we're, there's gonna be new testing tools and new ways of, of deploying.

So we've got a long way to go before this is gonna happen. And right now we have, so we can't even put an sbo om in a, in a, in a workflow. I mean, let's be honest, we can't, there's millions of work scripts that we can't update.

So it's gonna, Yeah, to respond to that though, that, that makes a lot of that actually ut you hit on a couple of key data points. One, you know, the SBO M is, uh, an executive order now with mo most companies. So you have to follow that compliance, right?

So you, so that's a requirement for a lot of the SDLC. But to your point, earlier point about complexity and tools, we see that 75% of respondents in our, uh, survey of over 850 respondents are using six to 15 different tools to do, uh, you know, just gather information about what's going on within the CICD pipeline. And, and, and not even actionable insights, but this is, this is like just understanding the data.

So there's so much data being thrown in it. There's a great amount of opportunity for AI to come in and clean, clean that up. But it, see, I agree with you.

Not, it's not there. Not even close. Not even close.

So Here's the thing, right? It's not, and I'm not just, I agree with everything both you just said, but the, um, and I love data too, but the, it look, too many people think this is a binary thing. Either it will, and this is sort of AI in general, but certainly in the DevOps space, I just got an argument with, you know, Andrew Clay Scheffer, one of my best friends, right?

We spent way too much time on the jet arguing on this point, and, and I, I think I finally convinced him it's not an either or, it's not like, is it gonna do everything or not? And so when you start thinking that way, you literally start thinking about the incremental, um, advancements and, you know, optimizations you get. And, and so like for example, um, you know, some of the stuff going on, you know, in gentech sort of automation tools, right?

Uh, Microsoft has, they haven't publicly released it yet, but it's, uh, co-pilot workspaces. You literally can grab an issue on a GitHub and it will walk through, analyze the repo, give you a proposal, allow you to mutate the proposal, then create the code changes. And then basically if you hit it, it'll create a pull request.

And there's about 15 other tools like open Devon, um, Amazon Q workflow. I mean, there's, this is an aggressive space. Now, I'll tell you, I've used this stuff aggressively.

I, 'cause I'm DevOps, right? So I, I really wanna try to understand what it works. A lot of it, you know it for the easy stuff, like I'm adding a parameter or somebody left the code parameter off, it fixes it brilliantly, give it a hard problem, and it, it falls off the cliff.

And, and I would say the Microsoft workspace is probably the mo I don't have the actual data. I'm gonna try to get the data to say it probably is the most advanced tool for modern enterprise stuff. 'cause a lot of stuff's going on right now.

It's like a bunch of, like, it's, it's, it's sort of a game and I don't want to diminish all the incredible work people are putting into it, but it's like, here's all the Python open source tool problems. There are. We already know what those are in the models and we're gonna fix 'em, right?

Like, and, but, but the, what I'm going after is like real live banking Java code, right? And I'm finding that the, um, you know, that the, the, the, the Microsoft, uh, co-pilot workspace is pretty sophisticated and what you would expect, but even then on the hard problems. But here's the beauty of it.

What I'm not thinking is, oh, it didn't create the, it didn't create the perfect fix, therefore I can't use it. Yeah. What I, it takes me 60% there, it finds me exactly in a code base.

I know nothing about it literally pinpoints the code I need to look at. So, so long as we stop thinking about like, it's gonna be everything or nothing, and you know, how these tools can just create all these advancements and, and I agree, Tracy, it's a mess. But if we, you know, and I'm not saying you think like this, but if we stop thinking that it can or it can't, and we start thinking about, okay, you're right.

If we can't even map the dependency map with SBOs, like how we go, well, there are things we can do. We can use AI to discover dependencies from a repo, right? Like, and so there, I mean, I think, I guess I'm rambling now, but I think all these things are, uh, the glass is definitely half full if you look at it.

Certain John, I can share some, some data to back up what you're stating though, right? I mean, you know, I, I think it is an evolution, it's a maturity. And if you look at it, depending on the organization, what we find in our research that, you know, GI UPS get GitHub and such can provides its critical part of the CICD pipeline.

But like, just to kind of back up some of the points you were talking about, 53% of respondents say that, that they use GI ops to do better collaboration. We see that 43% are using it as a single source of truth, right? Um, you know, 40% are using it as for tool independence, and then 38% of respondents are using it to harmonize and, and standardized code repositories.

All of these things that you're talking about are steps in order to get to that kind of view of where they want to go to. But if you take that response from the, from the respondents of the survey, and then somehow roll that into a, a platform, an AI platform that can actually use that data to say, this is how it's going to mature and help organizations, that'll get us closer to the quote, no ops, uh, space. You know, and just to take that a little bit further to Paul, uh, we just released a study called AI augmented DevOps, and very intentionally called it AI augmented because to your point, John, it's not, it's not, uh, autonomous software development like autonomous driving.

It's, it's an augmented tool to help you do development, do testing, whatever part part of that that, that it is, you know, we, we asked how much do you, how do you much do you trust the, uh, output of something that's been created by ai, whether it's a test plan or new code or a change to code. And, you know, a fraction 14% said, you know, they don't, they take it as is and move on, which I think is incredibly high, but most people realize that AI needs adult supervision. So you have to look at the output.

I think one of the challenges with it is, even in our own testing of software, we don't know always what is an open source library versus what Mitch wrote or what, uh, John contributed from pulling something down from OpenStack or who knows where that stuff came in. So we have to really think comprehensively about our testing strategies and not rely on knowing this was generated. I wanna get this point in before, um, is that, I was talking to somebody the other day and they, they made a good point.

Like if you think about a place like Capital One, right? At one point I think they had 20,000 Java developers, right? They follow design patterns.

There's gang of four patterns, right? They, and the design patterns of that, right? So those places are gonna have a much better chance of having highly optimized systems if the models can be trained, you know, like, you know, they probably, I mean, most of the common frontier models have those patterns in them.

But if they're not, you know, again, it, it, I think it goes back to what Tracy said. Like, and I know you didn't say it exactly like this, but it's garbage in, garbage out. If you're terrible at doing your infrastructure, your pipelines, you, you have no sort of pattern designs.

You do this, AI stuff's gonna make your life a lot worse. But Let me ask this question then. 'cause you know, we alluded to it at the start of the thing, but, um, CEOs have a tendency when they hear these kinds of issues, they'll go, wow, that sounds hard.

And then the first thing that comes to mind is, let's make this somebody else's responsibility. And so we're gonna appoint somebody to be the chief AI officer. Um, John, I know you have some opinions on this.

Is that the way to go or, um, do we need it to be more like something that just permeates through the organization naturally over time? You know, I, I mean, I, I don't know what the right answer is. I think my instinct is the wrong answer to bifurcate C level from a CIO and the chief, I, Mark Schwartz wrote a book called Seat at the Table, and he was the head of NI, uh, uh, border of security, um, you know, for, uh, Homeland Security, right?

And, and he said that like a chief data officer was a cop out. And I'm summarizing, right? I mean, like, why was the chief data officer on the same level of, of as a CIO isn't like it's network computing storage folks, right?

And so I, I think there's a push to do a Chief AI office right now because, you know, like there's a competitive advantage of bringing somebody at that level. You know, maybe you get the kind of person that really has the chops to do innovation, understands the technology, the business doesn't really understand. The negative is if you see that as a proxy for, um, understanding supporting the brand basically, which is your sort of the CIO responsibility, then, uh, from an IT perspective, I think it's a huge mistake.

So, uh, the quickly, the positives are you need a disruptor as somebody who probably wouldn't go to work for your company unless you made a role called Chief AI Officer. And there's all the positives, the negatives are if you sort of blindly say, well that's, you know, that's not the CIO's responsibility. You are gonna wind up in a world of hurt down the road.

I I say I talked to a large company the other day and that they, they're not a chief AI officer, but the person who is running a lot of like Wall Street Journal stuff that they're gonna pronounce, um, I asked the, the, the chief, the VP of engineering ai, might as well have been Chief aos 'cause clearly on an island. I asked if they had had any discussions with the seesaw and the answer was no. Um, I was like, you can take your aspirin now.

You can take your aspir later, but you're gonna be taking aspirin, my friend. So, you know, the devil's in the details on this too, because if you're, if you're doing the, let's hire an expert and they're gonna make all those decisions and all the knowledge and skills you can rely on that chief whatever officer, you know, bad mistake. I think of it more of like a digital transformation.

Someone who's leading this effort, whatever title you give them, right? Look at AI to say, how can this technology help us in our products and our services, in our operational efficiencies? Where should we be investing in and looking at where we can apply the technology working with the ciso, the CIO, the head of product, all of the organization saying, here's an opportunity, we want to make sure we maximize it to our, uh, benefit.

Maybe look at some defensive measures. But to your point, John, isolating is that, uh, I'm the data expert, so I'll make the data decisions is, like we were saying before, AI systems don't operate standalone. Very few of them mm-hmm, all of them tie into other systems and their value is through what those other systems and the people who use those applications get out of it.

So think of it as a, how are we gonna use this? Not who is responsible for making all the decisions about it. Yeah.

Mitch, one thing that, to add here, anecdotal information is great and having, having like a come a couple of conversations with different customers is, is actually boots on the street is really a good kind of validation point. I as, again, as an analyst, I look at it in the context of data and such. I have a study of 378 respondents to your point of spend.

And John also to your point where they're investing, 91% of respondents are looking to invest, uh, in their, uh, strategies around not just resources, but where they're looking to invest in 2024 is acquiring new skills around AI and having those resources available that have somebody that's responsible for those AI strategies. Hire more personnel. Is, is what they're looking to do.

Good luck with that. You'll never find them, but if you do, cool. Uh, the other big factor 40% are of, uh, respondents indicated that they're looking at working with service delivery partners to get their job done.

So they're working with experts in the space. Going back to our accountability conversation earlier, if they pass the buck to somebody else to say, okay, if you know how to do this, you're the accountable one. The other thing that, uh, that was, that was interesting is, uh, implementing, uh, cross kind of platform ecosystem, uh, solutions across, uh, an observability 42%, but 52% of respondents for top spending intentions, uh, for 2024, was to improve collaboration across DevOps application developers and IT operations exactly to what Mitch was talking about, right?

Which is, you know, it can't be done in that silo. And John, you were mentioned that silo of, okay, I'm gonna look in my four walls and that's it. That's not what the, these respondents are saying.

They're saying, we want to invest 52% actually saying that they want to invest in that collaboration across those organizations. All right, last word on this, Tracy, you gonna hire AI officer or what? Uh, no.

This is the responsibility of the architect. Uh, if we have a chief architect, it is an architectural decision, and that's where it belongs. All right, folks.

You heard it here. All I know is never take a job where you're responsible for something, but you have no authority. So keep that in mind when you're actually thinking about what title to take on.

Folks, thanks for sharing your thoughts and insights today. As always, you guys were awesome. I want to thank everybody for watching this episode and stay tuned because we got an awesome lineup of content coming up on text Drive TV right behind us.

See you next time. Discover Textron Group, the epicenter of tech innovation. We are your go-to for reaching IT leaders and practitioners worldwide.

Our secret, impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients.

Let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Textron Group.