Techstrong Gang – May 7, 2025

Hey everyone. Have you ever wondered how IBM can start a technology and then have to always be playing catch up? You're watching Text On Gang.

Hi everyone, it's Alan Shimmel, and happy Wednesday to you. We've got some great stories here on Textron Gang today. As usual, there's a fair amount of ai, a little bit of open source, some intrigue, palace politics, and a bunch more.

We've got a great gang to bring it to you. Let me introduce you to them and we could dig right in. So first of all, he's, he's not up in Silicon Valley.

He's out in, uh, Las Vegas land to, in fabulous Las Vegas, as they say. He's our, uh, uh, Silicon Valley editor, John Swartz. Hey John.

How are you? I'm good, I'm good. It was pouring yesterday.

Uh, we had a little bit of delay, but, uh, on our way in, but I'm looking forward to the, uh, ServiceNow Knowledge Conference this week. Oh, very cool. Yeah, it's a good conference.

Yeah. Well, good. Yeah, we'll be looking for some good reports from there.

Enjoy moving from Vegas to San Angelo, Texas, the data center capital of the world, or soon to be in the Tech Concho Valley, I think it's called. She's our own editor, Amanda Ani. Hey, Amanda, how are you?

Hello. Good. I'm pushing it for it to be, we're getting a lot of tech companies here.

All righty. Then moving up. She's Futurum vp, uh, analyst, Mitch Ashley.

Hey Mitchell. How are you? Good.

I've returned to the, uh, guitar background, So yes, I see that. And they are real. One of these days we're gonna have to have you open up the show playing the guitar because I've had people ask me, does he really play those guitars or are they just for show?

They're they're a collection. No, I don't know how to play guitar. No.

I've been playing guitar a long time. Happy to be careful what you ask for Alan. Absolutely.

And then moving from Mitchell over to, uh, Harrison, New York. He's our chief content Officer actually on his way down what to the DC area today, right, Mike For Nutanix? Yep.

Yep. Uh, chief Content Officer, Mike Ard. Hey Mike.

Hey, Hey. It's raining here in New York and our prayers are answered 'cause the Yankees are struggling. So we're hoping for a rain out.

Pray for, what was it, Warren, someone else. And Pray for Rain. It was an old story.

I think it was the Braves Orange spawn. It was a spawn. And, uh, Johnny saying, and Spawn and insane and pray for rain.

That's it. What a, I love being surrounded by some baseball people like this. I thank you.

Um, so Mike, I I, I teased it coming out. IBM is making a habit. I mean, quite frankly, they should have been owning the ai, you know, they were doing Watson commercials when I was a kid, it seems like.

Um, and here they are now they're making a major agentic AI push. But it feels very to me. Anyway, I'm, I'm interested in your take and of course John and Mitchell's take it seems very me too ish.

It seems very catch up ish. Why aren't they dealing from a leadership position? Mike?

What, what's the story here? Well, I'll just give you the good and the bad of it. It's, the good of it is they are actually putting together an orchestration framework around Watson X for integrating and managing all of these AI agents that are out there.

And they now have 150 partners and they're making a case that says, you know, what matters here when Agen AI is access to enterprise data, and they are going to be the ones in a best position to provide that level of integration. The downside of it, and I'm gonna pass it to Mitch a minute after this, is that, you know, it's a hodgepodge of other stuff. They bought web methods last year.

They got, um, Databricks, I think as a data lake kind of platform. And they're once again, stitching it together in a way that you could say was a consultant's dream, which is not necessarily what it people are gonna love. So, Mitch, I mean, you would follow this, but what's your take?

Well, you know, we, we've all Watson has taken on many lives, right? And now of course, uh, being kind of the center of ai, or at least it has been, this is a pretty comprehensive announcement and, and I understand kind of the hodgepodge of it 'cause there are some sort of things thrown in, right? They also did a Linux one five, a distribution update for the, uh, C 70 mainframe that runs, that has the, uh, AI processing that tell 'em two processors.

But I think what, what I take from this is whether they're late to the game or not, you know, the every, every day, every week, it's hard to even write analyst papers about this. 'cause there's news every day. You, you have to go back and update what you're writing.

They have a pretty comprehensive approach. And why I say that is they're not just introducing an agent capability, they're also including a framework called Orchestrate. Uh, they're also including an agent catalog for, uh, both agents that they built as well as third parties.

And, and you mentioned the data lake, uh, from a company that they bought that you also mentioned the web methods. You know, those are sort of the things kind of thrown in, um, if you will. But I think that's also, well, if you aren't gonna include those as part of your AI strategy, where do they fit?

So they're including it. Um, they also threw in a lot of, I say throw in. They also announced a lot of, uh, integrations with enterprise applications and enterprise services from Adobe AWS, Microsoft, you know, the, the, the notable ones, you know, serves an hour Workday folks.

So they're making a play to say, you know, you're still a safe bet working with IBM, we're your enterprise partner. And, uh, as you said, the integrators, the consultants, the people who help put put these things together, now they've got a full tool chest, if you will, to go help customers. So now again, not all of this is available.

Now, some of it is in preview, like the agent catalog, but, um, I thought it was also interesting, you know, you had a really good article out on, uh, tech Strong AI about this. CEO Arvin, uh, kna made a statement about it. I, AI is still being invested in, but only 25% of AI initiatives have achieved in ROI.

This is the year AI is, you know, people are looking for some return that it's gonna be worth doing. Um, and so I think it's interesting that, that he brought that up. I think sometimes it's helpful to be later to the game, as y'all say, because they can look and see what's working, what's not working, what problems do users have and come out with a better solution because they waited.

All right, third mover advantage, right? So, but this is so quintessentially IBM mish, right? So look, they had this Watson, and it's now called Watson XI, I remember going to IBM conferences literally 10 plus years ago.

And seeing Watson, right? And some of the cool things Watson was doing with dating apps and all kinds of things they were building with it. And then AI explodes on the world.

No one thinks of AI of IBM as an AI leader per se, but yet they're IBM just, you know, they have 500 of the Fortune 500 is as they probably have 600 of the Fortune 500 as customers. You know, they have an account rep for every one of those, and they get it in there. They're great at building ecosystems, they're great at bringing their partners in so they can stitch together this comprehensive soup to nuts kind of offering that.

But sometimes when you look under the covers, you could still see the, the threads, the stitches that stitching it all together. And they're not so tight. It's almost like we don't wanna be the first ones to market.

We want to, we wanna come in with the, the big blue solution here that, you know, is, is this big and is is everything, right? And, and there's going to be a portion of the market that loves that, right? Their, their base customer base loves that.

But there's also a portion of the market that says, you know, look up overhead dinosaur, that meteorite might just, meteor might just fall on your head. And, am I getting the cutting edge? Yeah.

Ai, ai, I'm sorry, John, right? AI was all about speed and about getting to market and about and, and, oh no, that's okay. AI is about getting to market and hitting people over the head with all these different announcements.

And the only time I've really seen IBM's name associated with AI was as a partner of all these other companies that jumped the line or moved faster than IBM The one thing, and Mike is I think totally right, is about this consultant's dream. When you stitch together so many different parts, it makes me think of somebody who's like building a crib and there's like hundreds of parts and they're supposed to figure it out. You usually bring somebody from the outside in to determine it.

And I think that that kind of hurts. IBMI also think the fact that they, they had this legacy, they had to jump on everyone and they kind of let it slip away. I mean, maybe they catch up and, you know, they're a safe bet, so they have a lot of huge customers who will be, feel safe and secure doing this.

But I, I just think they could have done more. But, you know, we'll see this, this, this whole race is so predicated on hype and on reality and ROI is Mike Rhodes. Um, maybe that helps him in the long run.

So it was interesting to me that one of the things they called down the press conference was that they have increased r and d investment, I think it was by 40 or 50% or something like that over the last four years. Which, but the issue in my mind, and, and the example of this is that for all that investment, they seem a little slow and they keep being usurped by somebody who comes along who's moving just a little bit faster. For example, Andro has this model context protocol out.

Well, you know, IBM has been working on the equivalent of that for a while, and then late last month turned around and donated that to the Linux Foundation. Now whether that was because P'S becoming a Deto standard or they're trying to have, build a community as an alternative remains to be seen. But, um, I think most people didn't even know that IBM had such a thing in place.

And so now, you know, something is amiss with their execution, is all I can say. So some, I'm, I'm gonna disagree with you all on this, and here's my point of view on it. I'm, I'm pulling an Alan, I'm, I'm gonna disagree with the panel.

Is, is, uh, I don't, I don't think it's possible for IBM to keep up with an anthropic or a cursor or all these companies in the, I mean, innovation's just happening so fast. And maybe they'll have some areas where they do do carve out, you know, some places where they get ahead. But when you're dealing with, you know, the 600 companies of the Fortune 500, as you said, Alan, those people don't implement PO point solutions.

They implement solutions at scale that have to integrate across multiple systems, um, multiple geographical locations. It, it is a big effort to put things in. So I haven't looked at their solutions and, and say, and to say, are they enterprise grade necessarily?

But I gotta believe that that's an approach, Mike, that, that IBM has to take that the others don't. Philanthropic can can, can play in, in much different ponds and doesn't have to always deal with the very large, uh, companies and at scale, because they don't have an embedded base to deal with, they also can deal down mid and down market. So I, you know, let me be snarky for a minute.

They could be Apple that's behind. So there you go. That's my point of view.

Yeah, apple can get away with being late too. It is. What as IBM dos, you look at all these companies that are filling the gaps in all these different, uh, vertical applications or what, what have you, just sp specific types of agents and by the time IBM gets it together and their customers understand how it all to put it all together, they may have been diverted to, to a better solution that was came out faster.

I mean, again, it's all based on speed and I just, IBM just moves slowly. That's just the way they are. But, But here's Mitch, did you just call, I Wanna clarify a point, Mitch, did you just call IBM the apple of ai?

Is that how you went with That? No, I was saying a Apple is even later to the game than IBM m That was my point. They can get away with it though.

I mean, they're like, they're polling and surveys that are just consistently show that people are wi willing because of all else, because it's only because it's Apple. They're just willing to yeah, to wait to wait, you know, and hey, No Apple to be so, But here's the paradox of this. Apple didn't invent ai.

Apple didn't have an AI product out there when they had Newton, uh, Dell didn't IBM invest a ton of money in pure research, pure science. And as a result, they make breakthroughs, whether it's in ai, what they did with Watson or Quantum computing, they owned the whole quantum computing thing. Now you're starting to see all these quantum companies pop up and they're gonna run circles around it a bit.

Um, think back to the, remember the IBM, what was it? The I-B-M-P-C processor, it was really sort of an early sort of risk based processor. You know, ARM did pretty well with those.

IBM got out the processor business basically, right? So their MO is, they spend billions and of, and a lot of time those billions are well spent in basic research that come up with groundbreaking inventing new technologies, and then they sort of abdicate that leading horse spot and let this whole industry grow up around it. And then they'll jump back in because they don't like to be first, they don't like to, to do it, I guess.

And then they jump in a little late and play catch up. Now, apple and Dell specifically, they have very different philosophies. Apple and D are sort of the old bull and the young bull at the top of the hill, remember that story?

They're not looking to race down there and be the first one down. They don't like to jump into a market until the market is well established, it's multi-billion dollars and that they have a plan that's gonna capture a substantial share. They're not gonna jump into a market prematurely, and they're not gonna jump into a market that they don't feel they can carve out a substantial share.

And that's their philosophy and right, wrong or indifferent, it served them well. Um, I think that has been their philosophy in Apple, Alan until ai, they were late to announce and then they over promised and underdelivered and you know, as you said, you know, apple's not gonna be the first to necessarily introduce something, but when they do it, it's done well. That that hasn't been the case on ai.

I think to your point on IBM, what IBM has done in the past is they may not necessarily invent something. They didn't invent invent SQL databases, but they legitimized it when DB two came out, right? 'cause I remember people were like, nah, nah, you would never do relational what IBM has a a relational product.

They do that kind of thing was they'll do that in ai. I don't necessarily know, they're a much different company now, but I don't, I don't look to IBM to lead in every category in ai. I think they're better off, they're a services and product company.

They're better off serving customers of implementing successful AI projects with an ROI as opposed to leading with the next flashy squirrel. Yeah, to Mitch's point, they're kind of counting on the fact that AI is hard and a lot of organizations that are going down the path are gonna stumble because it takes a lot of coordination between different departments to make this work. And eventually somebody in that organization's gonna go, Hey, we're being left behind.

Let's just call one 800 IBM. Well, like Mitch pointed out that 25% of people aren't seeing the return on investment. So, um, IBM spoke to that in this article, and they think that their solution will.

So, And the deployment has also been slow among enterprises, right? In terms of, uh, governance compliance, just try and piece it together. There's been a Hess hesitancy and a fear of getting it wrong.

So things aren't being deployed as quickly. So maybe IBM benefits kind of in a small, in a small way from that, I would look to who IBM's gonna buy next. I think that's where they had, you know, big companies don't often innovate themselves.

They buy or they innovate through acquisition. Cisco does that. A lot of companies do.

Fair. Hey, we gotta take a break and come back for our next segment, our friends at Redis. You know, the old saying, you can hammer a pair of pants, but if you make it too short, you can't make it longer.

Again, once you leave open source, can you go back? We're gonna discuss that you're watching Textron Gang. All right, folks, we're back and we're moving on to Redis and they say, you can never go home again.

But here comes Redis trying to be an open source company one more time, because they kinda led the charge with moving away from open source, uh, when there was a lot of, uh, investors were unhappy with the returns that were being generated for these companies. And yet that led to a lot of forks of Redis, and a lot of people jumped on those forks. And we've seen this play out before in open source environments.

But Alan, let's start with you. I know you tracked this whole area pretty closely. Can you go back once you've left the farm?

Sure, you can. All you gotta do is put on a pair of Ruby slippers and click your heels three times. It's no place like home.

There's no place like home. Um, look, first of all, I'll give credit to Redis. It sometimes it takes a big database company to admit you made a mistake and a mistake was made.

Um, you know, the, the the the open source world can be forgiving though, right? If, and, and the beauty of it, back to my analogy with the hemming of the pants, is if you had something that's open sourced, right? It stays open sourced, then you come out with a new version that's no longer open sourced, that new version, everything from that point forward is not open.

But then at some point here, you open it back up. Well, assuming you still own the IP for that non-open source stuff, you could open it up and it's like nothing ever happened, right? And, um, so I think in this case it's a relatively easy thing.

But here's the real question. Will the community trust them that they don't pull this again? Or have they already moved over to the forks in their comfortable with them, or, you know, burn me once, shame on you, burn me twice, shame on me, and I'm not gonna give you a chance to burn me twice.

And so I think red is in addition to just returning to this open source model, you know, needs to do a confessional, a couple Hail Marys and whatever, Mike's Smiley, he remembers those. Um, you know, Forgive me father, it's been six months since my last open source confession. Yeah, well Act, so actually it's been 2020 since the last confession for Redis, because that's when, uh, Salvato spo I think is his name is, you know, who was the original developer of Redis.

He came back, he left in 2020. He's come back. And that's a lot of what they're crediting for, why they're returning to their open source route.

So maybe some holy water and some Hail Marys, you know, with him on board. Well, I mean, and, and he may be the personality they need for that community to, To embrace them. It's got the credibility, right?

It's got the Credibility, right? That's the credibility because that's going to be the issue. I mean, changing the license here and going back to an open source license is a relatively trivial thing for them.

Um, regaining the respect and, and commitment from the, and trust of the community is gonna be the hard thing. But I I will also say, you know, nature pours a vacuum and there have been competitors that have popped up in the open source space there. And it's gonna be interesting to see if just moving back to open source is enough to put Redis at the top of the hill again, Or they could know there's a path down the middle.

They could support Velcade and their own open source core database as two separate forks that are, then they're delivering a bunch enterprise services on top of that and extensions. And whether you're running Valki or Redis Open source, maybe they don't care. I was reading this article, um, and what you were talking about Valki, how they created that.

And I was thinking that it just shows the, um, the stronghold open source has, and kind of the power I guess, that developers have when it comes to breaking away from open source. I look what they may wanna do with the, you know, with the founder back and everything is try to merge it back in, into, into Redis. 'cause having two versions of, of in essence the same thing, right?

I gotta believe Salvador's got some pretty strong Yeah. He'll, he'll get them. I think you're gonna see that merge back in, because over time they just separate more and more and it becomes harder to, all right, So Reconcile, So then donate the two of them to your favorite consortium du jour, and then have the consortium pull 'em all together.

You could do that too. Some consortiums don't like to pull it together. They, you know, they like to see how many projects they're managing.

Just saying, just saying. Um, but look, I, I think this is a good thing for the community and for the user base. Um, I don't, I don't think Greis will have a disruption in innovation or anything like that.

Um, if anything, maybe it'll be more innovative. So I, I actually think it's one of the, maybe the few credible ways of coming back to open source, having the original creator, author. I mean, it seems to me you've got, you know, provided that person's, you know, got a good, still got a good reputation in the community, I assume so they can help, they, they can help pull it back.

'cause that's who put, you know, it's like Linus, right? Putting trust in Linux and all of that. So yeah, no, I, I, I agree a hundred percent Marty.

Sure. Right? You know, but we'll see.

Let, let's see how, you know, how the community embraces them and do we see a emerge of, of the forks and so forth. But it's a great open source story, right? It's part of what makes open source great.

So I say congrats to Redis. Good luck. Fair.

All right, let's take a quick break and come back here with our next block. Who would've thought AI is biased in its hiring? You're watching Text on Gang 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 Techron Group.

Hey folks, we're back. And this is the subject of a lot of interest these days. 'cause well, folks are looking for work in all kinds of fields, and there's always been this perception that somehow or other, uh, the process has been biased.

And John has a story now on, uh, tech Strong a talking about how, well, I guess if you look at it, the data and the processes used to train the AI model, to apply it to HR seems to be favoring men. And I guess, John, let me ask you this. Is that the fault of the model or is it the fault of the processes and the data you use to train the model, which are clearly biased to begin with?

Yeah, it's probably a combination of, probably a combination of both. But what's disturbing to me is that there are more, uh, AI hiring, uh, models being used within organizations. In other words, when they wanna winnow down a number of candidates and then recommend a a handful to a person in the process, it's, especially for high paying jobs, it is shifting decidedly towards males.

Uh, given this test of, uh, open source AI models, that was done by a couple professors. The title of the paper was called Who Gets the Callback Generative AI and Gender Bias. So they basically looked at, uh, job candidates and they analyzed mid-sized open source large language models for signs of gender bias and hiring recommendations.

They looked at a more than, uh, 330,000 English language jobs, job ads from, uh, India's national Career Services online portal. And they sized up various models. And what they found with the models, with supplied, with job descriptions and asked to choose between two equally qualified male and female candidates, the callback rates for females was significantly lower.

And according to the professors, and I'm quoting them, these biases stem from entrenched gender patterns in the training data as well as from agreeableness bias induced during the reinforcement learning from human feedback stage. So in other words, Mike, the first two things that you mentioned, those two factors weigh heavily in this. And I don't know how you address something like this, but it doesn't surprise me in the least.

And, and the other thing that's interesting to me is that while this is happening, the Atlantic did a survey, it is a study of college graduates and the unemployment rate, and it's up to 6%, which is significantly high. And AI is playing a part in that, not just in terms of taking jobs or low level jobs, but I think, um, in the hiring process, it's, it's skewing dramatically towards, towards men. So the job market, if it wasn't already difficult enough because of the, of the, uh, prospect of competing with ai, it's not being helped by the AI hiring process.

That's in effect a lot of companies, Well, first of all, I'm glad this is being pointed out, but my reaction to it is, duh, we can't fix the biases in our human processes of hiring people. Why would we believe that AI would be suddenly better of it? Maybe it's the hype of ai, but to the, to your point, John, it's all it's trained on the data.

Guess what's in the data, you know, the information that the models are trained on, unless you take explicit measures and change the weights to try to try to change that model. Yeah, of course it is. Not that it's a good thing, but okay, let's address it now.

Yeah, that was my Thought too. Like, well, the data being used to train would be representative of that because, um, men still do overall tend to have the higher paying jobs from a percentage standpoint. So if that's the data you have to train AI on, The question is do we fix the models or do we fix HR first?

Well, this has been broken forever, right? In my mind, we've been playing, you know, keyword bingo on resumes forever and running them through these HR apps to kind of find somebody who's quote unquote qualified. I also think that a lot of the people who are doing the hiring kind of, you know, they kind of tangentially work with HR and they go out and post a job.

But most of the people I know who hire folks, they are people they already know. So they build their own little network, and they're not kind of, um, shall we say, very serious about what HR might bring in the door. It's not what you know, who you know.

Well, I mean, I, I'm gonna, I admit, I, I benefited from that. And, and I, one, the one thing that I was, I, I was hoping that they might even look at these males and they tend to be probably whites. Maybe that's gonna be a follow up, but it just, it's just the same old, same old.

And, uh, it's, my fear is that as these companies and hr HR departments in particular depend more on AI for hiring, is this gonna maybe even exacerbate the problem? So in my, that's MAGA voice. This is DEI nonsense.

Move along. I was gonna say de I was gonna say that, but I was waiting for you to say it, Alan. It, it's so true though.

You're gonna defund it. Alan defund it. Not only that, I'm, I've asked the Justice Department to look into it investigate, Right?

If we, if We, I'm say a truth, If we stop testing, it'll go away, right? Right. Yeah, exactly.

It didn't exist. I'm gonna say, I know a lot of people really struggling to find work, and it's come to a point where people are trying to use chat, GPT or other forms of AI tools to figure out how can they break through the AI filtering process. What's gonna be the key words the AI picks up that's gonna get theirs ahead of the thousands of other applications Or power to 'em, weaponize AI to either A lot of them.

Yeah. And, and a lot of people are using AI to write their resume, resume letters now, a Lot, most people, so They all, they all tend to be the same. So most people are.

Yeah. Well, But you know, for specific jobs, let's say in terms of coding, right? I, I remember interviewing a company that were based in the uk where what they would do is you would hire them if you were looking to hire coders, and they would prepare a prompt, you know, code, code, something like this.

And then it would, using ai, look at the code you generated, the code you wrote without having your name, your background, your sex identity, whatever, and then just pick who wrote the best code. And that was a way of kind leveling the playing field right now with today, AI writing the code, I think that screws the whole thing up. But, but there are ways where I think we could use AI to level the playing field and maybe try to get some of that bias out of there.

Um, I don't know, Alan, because if the model was trained on the, all the code that's been developed, and most of that code was developed by males, and if you assume there's a difference in that, or maybe culturally India versus us, absolutely it's gonna have those biases. Even in the code, it's gonna favor what it thinks is Even English is a second language. Coders are at a disadvantage, perhaps.

Not saying we shouldn't try to use it, but I think we have to recognize it. It has a natural bias into it. And that's what we have to figure out.

I thought that's why we had a DEI program. No, we defunded that. Well, I said had in the past tense.

Oh, okay. But that was the reason for it, right? Was because this is, it's built into the system.

It's built into the SATs and the L SATs and everything else. You know, when you have basically white men making the rules, you're gonna have rules, the favorite white bed, and no matter how much you stamp up and down and whine and threaten and huff and puff, it doesn't change the facts on the ground. Mm-hmm.

And I would, I would like to see the AI models vetted by some third party and maybe, you know, some sort, And that third party not made up of white males. But part of the, the, I think part of the reason why they're turning to AI in the first place is because so many people have access to just apply online now. There's just thousands of people randomly applying to hundreds of jobs at a time.

It makes it very difficult to filter Through. Yeah. It's not one of my favorite things is yeah, trying to hunt through resumes for a person to hire.

Anyway, let's hope it gets better. I think this is an area where we can get better with AI and we can level the playing field and, and we owe it to all of us to, to do that. Guys, I think that's gonna wrap up today's, uh, text, text Drunk Gang.

John, enjoy Las Vegas. Stay away from those tables. Yes, thank you.

The house always wins. I don't gamble. I know I don't gamble.

All right. No, Amanda, Mitch will see you soon, Mike. We'll see you soon.

Well, you'll be at Nutanix this week. Looking forward to some reports from there. Until next time though, until tomorrow, we have a full text Drunk TV lineup following today, including I believe we have a Tech Field Day going live today.

So you'll be able to see that here on Techstrong tv. Until next time, this is Alan Shimel. Have a great day.

We're out.