 Good morning, live from Chicago. It's theCUBE on the floor at Ansible Fast 2022. This is day two of our wall-to-wall coverage. Lisa Martin here with John Furrier. John, we're going to be talking next in this segment with two alumni about what Red Hat and IBM are doing to give Ansible users AI superpowers. As one of our alumni guests said, just off the keynote stage, we're nearing an inflection point in AI. The power of AI with Ansible is really going to be an innovative, I think an inflection point for a long time because Ansible does such great things. This segment is going to explore that innovation, bringing AI and making people more productive and, more importantly, this whole low-code, no-code, kind of right in the sweet spot of the skills gap. So it should be a great segment. Great segment. Please welcome back two of our alumni, Richie Peary is here, the Chief Scientist, IBM Research and IBM Fellow, and Tom Anderson joins us once again, VP and General Manager at Red Hat. Gentlemen, great to have you on the program. We're going to have you back. Thank you. And thanks for joining us. Fresh off the keynote stage. Really enjoyed your keynote this morning. Very exciting news. You have a project called Project Wisdom. We're talking about this inflection point in AI. Tell the audience, the viewers, what is Project Wisdom? And Wisdom differs from intelligence. How? Project Wisdom is really about, as I said, sort of combining two major forces that are in many ways disrupting and really constructing many aspects of our society, which are software and AI, together. And I truly believe it's going to result in a seismic shift on how not just enterprises, but society carries forward. And as I said, intelligence is, I would argue at least artificial intelligence is more, in some ways, mechanical, if I may say it. It's about algorithms, it's about data, it's about compute. Wisdom is all about what is truly important to bring out. It's not just about when you bring out a insight, when you bring out a decision to be able to explain that decision as well. It's almost like humans have wisdom. Machines have intelligence. And it's about Project Wisdom, that's why we call it Wisdom, because it is about being a assistant, augmenting humans, just like be there with the humans and almost think of it as behave and interact with them as another colleague will. Versus intelligence, which is, as I said, more mechanical, is about data, compute, algorithms, crunch together. And we want to bring the power of Project Wisdom and artificial intelligence to developers to, as you said, close the skills gap, to be able to really make them more productive and have Wisdom for Ansible be their assistant. To be able to get things for them that they would find many ways mundane, many ways hard to find, and again, be an assistant and augmented part of it. You know what's interesting, I want to get into the origin, how it all happened, but interesting IBM research, well known for the deep tech, big engineering, and you guys have been doing this for a long time, so congratulations. But this interesting here at this event, even on stage here, event, you're starting to see the automation come in. So the question comes up scale, so what happens, IBM buys Red Hat, you go raid the IP, treasure trove of AI, because this is kind of like bringing two killer apps together, the Ansible configuration automation layer with AI. It's just kind of a... Yeah, it's an amazing relationship. I was going to say marriage, but I don't want to say marriage because that may be the last thing you watch. I didn't mean to say raid the treasure trove, but to kind of like, oh yeah. Amazing relationship where we bring all this expertise around automation, obviously around IT and application infrastructure automation, and IBM research, Richier and his team bring this amazing capacity and experience around AI, bring those two things together, and applying AI to automation for our teams is so incredibly fantastic. I just can't contain my enthusiasm about it, and you can feel it in the keynote this morning that Richier was doing, the energy in the room, and when folks saw that, it's just amazing. The geeks are going to love it for sure. Richier, I want to get into the whole evolution computers on computers. I remember the old days, Thinking Machines was a company generations ago that I think they've sold or went out of business, but self-learning, learning machines, computers programming, computers was actually on your slide. You kind of piece out this next wave of AI and machine learning, starting with expert systems, really kind of, I don't want to say static, but like, okay, programs. And then now with machine learning and that big debate was unsupervised, supervised, which is not really perfect. Deep learning, which now explores some things, but now we're at another wave. Take us through the thought there, explain what this transition looks like and why. I think we are, as I said, we are really at an inflection point in the journey of AI. And if AI, I think it's fair to say, data is the bane of AI. Without data, AI doesn't exist. But if I were to train AI with what is known as supervised learning or data that is labeled, you are almost sort of limited because there are only so many people who have that expertise. And interestingly, they all have day jobs. So they are just going to sit around and label this for you. Some people may be available, but this is not, again, as Tom said, we are really trying to apply it to some very sort of key domains which require subject matter expertise. This is not like labeling cats and dogs but everybody else in the world knows. There are, the community is very large, but still the skills to go around are not that many. And I truly believe to apply AI to the word of enterprises, information technology, automation, you have to have unsupervised learning and that's the only way to escape. And these two trends really about information technology percolating across every enterprise. And unsupervised learning, which is learning on this very large amount of data with, of course, very large compute with some very powerful algorithms like transformer architectures and others which have been disrupting the domain of natural language as well are coming together with what I described as foundation models. Which anybody who plays with it, you'll be blown away, literally blown away. You call that self supervision at scale which is kind of the foundation. So I have to ask you, because this comes up a lot with cloud, cloud scale. Everyone tells horizontally scalable cloud but vertically specialized applications where domain expertise and data play. So the better the data, the better the self supervision and better the learning, but if it's horizontally scalable, it's a lot to learn. So how do you create that data ops where the machines are going to be peaked to maximize what's addressable but what's also in the domain too. You got to have that kind of diversity. Can you share your thoughts on that? Absolutely, so in the domain of foundation models, there are two main stages I would say. One is what I'll describe as pre-training which is think of it as the machine in this particular case is knowledgeable about the domain of code in general. It knows syntax of Python, Javascript, I don't know, Go, C, Java and so on actually and also YAML as well which is obviously, one would argue is the domain of information technology. And once you get to that level, it's almost like having a developer who knows all of this but may not be an expert at Ansible just yet. He or she can be an expert at Ansible but is not there yet, that's what I'll call background knowledge. And also in the case of foundation models, they are very adept at natural language as well. So they can connect natural language to code but they are not yet expert at the domain of Ansible. Now there's something called, the second stage of learning is called fine-tuning which is about this data ops where I take data which is sort of the SME data in this particular case and it's curated so this is not just generic data you pick off GitHub, you don't know what exists out there. This is the data which is governed which we know is of high quality as well and you think of it as you specialize the generic AI with pre-trained AI with that data and those two stages including the governance of that data that goes into it, results in this really breakthrough technology that we've been calling project wisdom for our first application is Ansible but just watch out that area, there are many more to come and we're going to really, I'm really excited about this partnership with Red Hat because across IBM and research, I think wherever we, if there is one place where we can find excited open source, open developer community, it is Red Hat. Tom, talk about the role of open source and project wisdom, the involvement of the community and maybe, Richard, any feedback that you've gotten since coming off stage? I'm sure you were mobbed. Yeah, so for us, it's called project wisdom, not product wisdom, right? Sorry. No, you didn't say that but I want to just emphasize that it is a project and for us, that is a key word in the upstream community that this is where we're inviting the community to jump on board with us and bring their expertise. All these people that are here will start to participate, they're excited in it, they'll bring their expertise and experience and that fine tuning of the model will just get better and better so we're really excited about introducing this now and involving the community because it's super smart. Everything that Red Hat does is around the community and this is no different and so we're really excited about project wisdom. That's interesting, the project piece because if you see in today's world the innovation strategy before, where we are now, go back to say 15 years ago, it was all standard, it's got to have standard bodies. You can still innovate and differentiate but yet with open source and community, it's a blending of research and practitioners. I think that to me is a big story here is that what you guys are demonstrating is the combination of research and practitioners in the project. So how does this play out? Because this is kind of like how things are going to get done in the cloud because Amazon's not going to just standardize their stack at higher level services, nor is Azure. And they might get some plumbing commonalities below but for project wisdom to be successful, it doesn't need to have standards if I get this right. Mike, on point here, what do you guys think about that, react to that? Yeah, so definitely, I think standardization in terms of what we will call ML Ops pipeline. For models to be deployed and managed and operated, it's like models like any other code. There's standardization on DevOps pipeline, there's standardization on machine learning pipeline and these models will be deployed in the cloud because they need to scale. The only way to scale to thousands of users is through cloud and there are standard pipelines that we are working and architecting together with the Red Hat community, leveraging open source packages is really to help scale out the AI models of wisdom together. And another point I wanted to pick up on just what Tom said. I've been sort of in the area of productizing AI for long now, I've been experienced with Watson as well. The only scenario where I've seen AI being successful is in the scenario where what I describe as it meets the criteria of flywheel of AI. What do I mean by flywheel of AI? It cannot be some research people build a model, it may be wowing, but you roll it out and there's no feedback. Yeah, exactly. Okay, yeah, so what actually? The only way, the more people use these models, the more they give you feedback, the better it gets because it knows what is right and what is not right. It will never be right the first time actually. The data it is trained on is a depiction of reality. It is not a reality in itself. Reality is a constantly moving target. And the only way to make AI successful is to close that loop with the community and that's why I just wanted to re-emphasize the point on why community is that important actually. And what's interesting, Tom, is this is different between standards bodies, old school and communities because developers are very efficient in their feedback. They jump to patterns that serve their needs, whether it's self-service or whatever. You can kind of see what's going on. It's either working or not. Yeah, yeah, yeah. We get immediate feedback from the community and we know real fast when something isn't working, when something is working. There are no problems with the flow of data between the members of the community and the developers themselves. So yeah, it's great. It's going to be fantastic. The energy around Project Wisdom already, I bet we're going to go down to the Project Wisdom session, the breakout session, and I bet you the room will be overflowed. How do people get involved real quick? Take a minute to explain how I would get involved. I'm a community member. I'm watching this video. I'm intrigued. This has gotten me enthusiastic. How do I get more confident with this opportunity? So you go to, first of all, you go to redhat.com slash project wisdom and you register your interest and you want to participate. We're going to start growing this process, bringing people in, getting ready to make the service available to people to start using and to experiment with, start getting their feedback. So this is the beginning of a journey. This isn't the midpoint of a journey. This is the beginning, even though the work has been going on for a year, this is the beginning of the community journey now. And so we're going to start working together through channels like Discord and what not to be able to exchange information and bring people in. What are some of the key use cases? Maybe we're cheers starting with you, that you think, maybe dream use cases that you think the community will help to really uncover as we're looking at Project Wisdom and really helping in this transformation of AI. So if I focus on, let's say, Ansible itself, there are much wider use cases, but Ansible itself and I would say, I had not realized, I've been working on AI for a long, but I had not realized the excitement and the power of Ansible community itself. It's very large. It's very bottom self, which I love actually. But as I went to a lot of CTOs and CIOs of a lot of our customers as well, it was becoming clear the use cases of, I've got thousand Ansible developers or IT automation experts. They write code all the time. I don't know what all of this code is about. So the system administrators, managers, they're trying to figure out sort of how to organize all of this together. And think of it as Google for finding all of these automation code, automation content. And I'm very excited about not just the use cases that we demonstrated today, that is beginning of the journey, but to be able to help enterprises in finding the right code through natural language interfaces, generating the code, helping Darpers debug their code as well, giving them predictive insights into, this may happen, just watch out for it. When you deploy this, something like that happened before, just watch out for it as well. So I'm excited about the entire lifecycle of IT automation, not just about at the build time, but also at the time of deployment, at the time of management. This is just a start of a journey, but there are many exciting use cases abound for Ansible and beyond. It's going to be great to watch this as it unfolds. Obviously just announcing this today, we thank you both so much for joining us on the program, talking about project wisdom and sharing how the community can get involved. So you're going to have to come back next year. We're going to have to talk about what's going on, because I imagine with the excitement of the community and the volume of the community, this is just the tip of the iceberg. Absolutely. This is absolutely exact. You're not excited about it. Excellent. And you should be. Congratulations. Thank you. Thanks again for joining us. We really appreciate your insights. Thank you for having us. For our guests and John Furrier, I'm Lisa Martin and you're watching theCUBE live from Chicago at Ansible Fest 22. This is day two of wall-to-wall coverage on theCUBE. Stick around. Our next guest joins us in just a minute.