Episode Transcript
Speaker 1 00:00:05 Today I want to talk about the rise of the co-pilots.
Speaker 1 00:00:10 There was a massive series of announcements this week from Microsoft all about AI and Windows 11 and many, many other interesting things. And that was of course, following the big announcements by Google last week, Google io about its new AI services, Bard Palm and others. But I think rather than talk about the whole thing, which I'll write about in an article, I wanna talk about the co-pilots. So the idea of a co-pilot, as Microsoft calls it, and I love the name, is that we have this AI agent, you can think of it as a chatbot, but there's many types of chatbots that is context aware of certain information that we're using at a particular point in time and gives us recommendations, advice, search and analysis of that information. So if you are a history of fan of Microsoft like I happen to be, because I've been around for a long time, Microsoft has been working on this for a long time.
Speaker 1 00:01:10 There was this funky tool called Clippy that used to appear in Windows and get in your way and try to tell you what to do. It was the right idea. And there was this entire system called Microsoft Bob that was actually a hilarious flop that tried to do the same thing. But I think the folks at Microsoft understand this because they've been building individual productivity tools forever, and they know that as our economy becomes more productive, and by the way it is, I'm gonna be talking about that at our conference in June. The tools we use have to get easier and easier to easier while they become more powerful. In fact, I think there's a really fascinating quote from Steve Jobs where he said computers are bicycles for the mind. In other words, what the computer is doing for us as individuals, and it really doesn't matter what you're doing in your job is it's giving you the opportunity to go where you want to go, do what you want to do and maybe have some fun and creativity in the process.
Speaker 1 00:02:09 And so if you're bogged down with a spreadsheet or a PowerPoint tool or something else and you can't figure out how to use it, you're not having any fun and you're not going anywhere. So that's what the co-pilots do. Now, the first co-pilot that Microsoft built was the GitHub co-pilot for software engineers. And if you've ever heard of pair programming, there's actually a pretty interesting history of software engineers helping each other to code. So what the GitHub co-pilot does is it shows you where you're making mistakes and gives you recommendations to code better. And there have been a lot of studies by Microsoft and others that show that tho that particular tool increases software programmer productivity by anywhere from 35 to 58% because coding's kind of a technical domain and there's all sorts of different ways to do things and it takes a lot of energy to write it.
Speaker 1 00:03:04 So the co-pilot is helping you. And so what Microsoft has done in a really fairly brilliant way is they've launched a family of co-pilots all based on the same large language models from open ai, which I'll talk about in a minute for all sorts of interfaces. There's going to be a co-pilot for Windows, which means that billions of people will have a co-pilot built into Windows 11. There's a co-pilot for Power bi, which allows you to literally look at a large source of data and analyze it with conversations, which is a massively powerful idea. There's the Microsoft 365 business chat and co-pilot for Excel and PowerPoint and Word, and there will be co-pilots for other parts of the Microsoft infrastructure. By the way, there's also a Microsoft co-pilot development studio now called the Azure AI Development Studio. They opened up the stack of the co-pilot and they show you how it works level by level.
Speaker 1 00:04:06 So you can add your own data, you can create your own context, and you can even create plugins. A plugin is an application interface so that the third party application can send and retrieve data from the co-pilot. So if you're in the co-pilot and you wanna book a hotel room, it could take you to Travelocity or go out and look for hotel rooms or even Airbnb without you having to log into Airbnb and find it. And that includes plugins already from Adobe, Atlassian, ServiceNow, OpenTable, and others. So these co-pilots are going to be really, really useful interfaces. And it's interesting that since Microsoft is so far down the learning curve here, and obviously so embedded into it that some of the best HR applications are gonna do this. In fact, last week there was another announcement by Success Factors that they are going to be shipping co-pilots embedded into success factors.
Speaker 1 00:05:05 I don't remember when it comes out, but they basically demoed it that will be able to automatically create job descriptions, candidate communications, candidate emails, and other recruiting tools from the corpus of data in your Success Factors instance. So in other words, if you have a bunch of jobs or skills or roles in success factors, the copilot in success factors will know that information and will be able to produce this generative AI content for you. Now, I see many, many, many implications of this. By the way, we are building our own copilot. There is a Josh Buren co-pilot that I've demoed for some of you that you're gonna see at our conference, which is really, really cool. It's basically an interface to access all the research we've ever done through a conversational interface and it's just spectacular. Now, there's many implications of this and I want to talk through them from the standpoint of what it will mean to you in HR and then what it'll mean to you in it.
Speaker 1 00:06:04 And relative to other AI stuff. The first is that the underlying infrastructure under the co-pilot from Microsoft is based on open AI with lots of security and other services added. For example, there's a technology that all of us, Microsoft people have called the Microsoft Graph and that defines what content and information each individual has access to. So if I'm developing or using a copilot on my desktop, I'm not gonna see somebody else's documents unless the IT department made that available. So in the Microsoft infrastructure, security is built in. The second thing is if you have lots of information inside of your company, which I know you do, everybody does, that you wanna make available to your employees, this is the way to do it. Not only can you now use Microsoft's new data system called the Microsoft Fabric to do this, but you can take any documents you have, whether they be in Word and PowerPoint or just sitting around in various IT systems and you can import them into your co-pilot and your company, just like we have done that for the Josh person co-pilot.
Speaker 1 00:07:17 You will see that in your company, and I think your IT people are gonna go nuts with this. And in fact, most of the sessions at the Microsoft build launch this week were targeted towards IT developers. And there are also features in the Azure services to protect you against images or documents or words that are inflammatory in some way. So you can keep these things safe. So if somebody loads something inflammatory or porn or something into your system, that will not make it into your infrastructure. The third reason that I think this is huge is this completely changes the way we use corporate applications. Now, I've had this conversation with a lot of vendors and they're starting to get it very, very quickly that no matter how hard you work to create a great website, it's never as good as the Google search bar <laugh>.
Speaker 1 00:08:13 I mean, imagine if you remember what Yahoo used to be like before Google existed, it was a webpage with all sorts of links and tabs and it was really kind of chaos cuz they couldn't keep up with it. Well, the whole idea of the search bar from Google was to simplify the user experience because Google's products are really focused on the user. Well, this is what a chatbot or a co-pilot does. It eliminates the need for you to build a complex portal for employee benefits or time off or payroll or whatever the employee experience issue you're dealing with and replace it with something that's very intelligent, easy to talk to, and can produce many forms of output. For example, in the co-pilot that we're building, you can ask questions of our research and it will answer them in a narrative form and then it will give you a list of links to the research studies, the podcasts, the articles, and soon the courses in the JBA that you can take to learn more information about the question that you asked.
Speaker 1 00:09:19 That is basically a portal. And in my particular case, I've now spent 20 years building websites <laugh> for various implementations of our research, and they've never really worked perfectly. They work okay, but they're not that great and this is a radical new way of solving that problem. So you're gonna have lots and lots of opportunities to do that. If your workday, if you're adp, if you are a software company that's selling to hr, you're gonna have exactly the same opportunity. All of the interfaces and tabs and various UI designs that you've built to make your system easy to use are likely to become unnecessary or at least mostly unnecessary. But there's more. Think about the applications we do in HR for training, and let me just take a couple minutes and touch on this. The process of building a training application generally goes as follows.
Speaker 1 00:10:13 We have a performance consulting process where we diagnose the problem or the educational and and informational challenge we're trying to address, and maybe it's onboarding, maybe it's sales training, maybe it's compliance training, maybe it's d e i training, maybe it's management development and so forth. We then look at the corpus of information we want to build the training from. That may be documents, documentation, compliance documents, frameworks, models, discussions we've had with different people, interviews we've had with different people. Then we create an instructional design. Is this gonna be a two minute course, a five minute course, a three hour course? Is it gonna be blended? Is it gonna be online? Is there gonna be simulations? Are there gonna be drag and drop exercises? Are we gonna do testing? Are we gonna do certification? Are we gonna do credentialing? Are we going to have a capability or competency model behind it?
Speaker 1 00:11:02 Et cetera, et cetera, et cetera. Then we set a around and build it, which takes time because we have to build content, we have to write content, we have to do graphics, we have to do editing of video and editing of images. Then we put it into a platform, an LMS or something related, then we test it, then we launch it. It's a big process and most of you who don't work in l and d probably don't appreciate how much craft there really is to building amazing training. Well, I'm not saying we're gonna replace all of it, but a very significant amount of that can be done with a co-pilot. Co-pilots can build instructional plans, they can obviously access content, they can structure quizzes and tests, and I'm sure there will be plugins from many learning vendors to use co-pilots for the development of complex or simple instructional programs.
Speaker 1 00:11:58 And I'm not saying you won't be watching videos and taking courses, but you'll be a lot smarter and quicker at getting through that material. And maybe you won't need it with a co-pilot because a lot of times the reason we browse through a course is to find something we forgot and just learn it and go back to work. And that's the kind of information a co-pilot could help with in the area of recruiting. I don't need to really belabor this, but you can imagine the experience that a candidate would have with an intelligent co-pilot or a product like paradox, which allows them to look for a job, ask questions about the company, ask questions about the pay, ask questions about the culture, look at work hours, ask about salary, et cetera. That's gonna take some training obviously, but that's gonna be a massive opportunity for you guys.
Speaker 1 00:12:49 And now that the vendors like Success Factors and others are building co-pilots that directly access their skills models and other internal tools, those are gonna be able come easier and easier in the area of employee experience. I talked about this earlier. We're gonna be able to use co-pilots to radically simplify the way employees deal with HR systems and access data. And then in some of the more advanced areas like leadership development and coaching, I see absolutely no reason, and I've talked to vendors about this already, that there won't be co-pilots for learning about leadership co-pilots for new managers or new supervisors, co-pilots that'll give you feedback and support for wellbeing or stressful issues at work. And more and more and more. And I really have to compliment Microsoft on this effort because they see this, because this is the business they've been in forever. They see these opportunities in some sense faster than we do.
Speaker 1 00:13:46 And my experience building one of these things has just been, as I said, absolutely spectacular. Now that leads to, you know, sort of another issue about ai, which is what is going on under the covers and who's going to rationalize all these large language models and tools? Well, I think the way Microsoft looks at it, and I think Google will probably come to the same conclusion, the l m or underlying neural net that accomplishes some of these tasks. Many of these tasks is really sort of a platform. You can buy an l m, you can get an open source lm, but in most cases the value add is the content and the experience on top of it. The prompt engineering, which makes the copilot work correctly and the corpus of knowledge and how it's arranged and how it's organized on the backend. So what Microsoft is essentially doing with open AI is they're treat treating open AI as a platform from which they build all these tools and applications.
Speaker 1 00:14:47 And you know, one could ask yourself, what is the risk of Microsoft investing and being so dependent on a small company in San Francisco for the core of its new systems? And obviously this was a big bet on the part of Microsoft, but I give them a lot of credit. I think Sacha tried to get this work done in Microsoft research. There have been a lot of articles about it. The Microsoft research people were not getting there fast enough and they've invested 10 billion and more in open eye AI to pull this off. And that makes the open AI people essentially a platform group for Microsoft. By the way, I would not be surprised to see a co-pilot for LinkedIn, which they haven't announced, but I gotta believe that's coming too. You know, in the case of Google, they have built a stack themselves. They have the new Palm L L M, which runs on everything from the Google Cloud all the way down to your laptop, PC, and your phone.
Speaker 1 00:15:47 So they can do a lot of this as well. I don't think they have been thinking about this as a productivity tool the same way Microsoft has because Google's revenue tends to come from search and advertising. So my guess is the co-pilot, like experiences in Google, will appear in Bard search, in the search experience and in the generative tools in Google workspace. And Google Workspace, as great as it is, is way, way, way behind Microsoft in its adoption. And Microsoft has been selling and supporting IT departments for as long as I've been really working, to tell you the truth, since the early 1980s and late 1970s. So I think the Microsoft lead in the IT implementations of this are very, very significant. I'm not saying Google won't catch up, but they will probably go into different areas. Now. What about the risks of all this and the risks of the generative AI coming up with bad answers, hallucinating, whatever else you might be worrying about.
Speaker 1 00:16:46 You know, I, I really am confused as to why Sam Altman is flying around the world telling all these governments to regulate ai. It reminds me of Mark Zuckerberg's efforts to do the same thing. It's either some form of self-serving protection where he's trying to get open AI grandfathered into the LE legislation so that he can protect his lead or some sort of a PR move. I mean, the odds of government agencies adequately learning how to regulate a technology that is this new or zero in my mind, I think what they will regulate is the outcomes, the need for unbiased hiring, the need for transparency in pay and pay equity and so forth. And those outcomes will become mandates for AI providers. So if the chat bot or the co-pilot from Microsoft or Success Factors is biased, you can sue them based on the E E O C laws in the United States.
Speaker 1 00:17:48 We don't have to regulate how the algorithms work. So I am personally just not really worried about the dangers of this technology because I really think we have a regulatory framework that handles this. You may have people inside your IT department though that are worried about access to information leakage of proprietary data or other information inside your company, PII and other things. And it's turning out that most of the large companies have people now in IT or in security, IT security that are looking at this and they're testing and validating tools against more and more models to test that they are in fact unbiased and that they are fair. In fact, I talked to two vendors last week that are getting into that business of helping large companies test their tools against various models for bias and other forms of discrimination. But honestly I think that's a sideline.
Speaker 1 00:18:46 I don't think that's the center of the story de, despite the fact that the New York Times loves writing about it, I really think these co-pilots are going to transform our lives, transform our systems, and to me this is the new technology platform. Look at the stock price of NVIDIA last week. Look at the stock price of a SM L tsmc. All of the technology providers, Microsoft, Google that participate in ai, the AI revolution is going to replace a lot of the IT infrastructure we've built over the last decade or two. And it's gonna be very, very positive for us, for the technology providers and a lot of the underlying tools that we use behind this, including this massive new world of chips. And so let's just think about this rise of the co-pilots as a great way to manifest AI in a high productivity useful way of getting our work done, of analyzing data, of building easier to use systems and eliminating the cost of complex, complex, complex user interfaces all in the goal of making us productive as we use our computers to bicycle around the world and get our work done in the most productive and enjoyable way.
Speaker 1 00:20:09 Anyway, something for you to think about. I'll put some links into the Microsoft announcements. I think you'll find them spectacular. And stay tuned for more. We will show you the Josh person co-pilot sometime in the next 60 to 90 days after we preview it at our conference.
