Episode Transcript
Speaker 1 00:00:06 This week I'm a little bit behind on the podcast because we had a really fantastic offsite in Southern California for our company. I want to talk to you about that for a minute. I also became a grandfather this weekend, just today, and that is the most moving, exciting thing that's happened to me in many, many years. I highly recommend it. I will tell you a little bit more about it as time progresses, but that was a pretty exciting couple of days I've just been through. And then I want to talk about the mania and amazing amount of press that's going on in AI and chatbots and HR technology in general relative to the economy. So the reason I wanna mention our offsite is we are a fast growing, medium sized company and we have a very, very passionate team of people. And we work remotely from all over the world.
Speaker 1 00:00:55 We have people in Asia, we have people in Europe, we have people in Southern Europe, northern Europe, and all over the us. And we spend a lot of time online with each other. But we did, we don't get together that often. So at least once a year or twice a year, we have a big meeting and every time we have this meeting, I come back thinking we should do this more often. And the the point I'm trying to make is, even though hybrid work is ill-defined for most companies and CEOs generally want to get people back in the office, employees want more flexibility. They want the four day work week, they wanna work the days and hours and location they're interested in. They don't necessarily want to commute if they don't have to. There needs to be a purpose to come back together and we're all frustrated by where that is.
Speaker 1 00:01:35 But what we've learned in our company, and I think most of you have learned, and I've seen this in many of the leadership meetings you've had, is there is absolutely no replacement for being together with people. The relationships, the trust, the bonding, the fun, the sense of community and shared vision that you can create, the sense of psychological safety. People can speak up, they can talk, they can let you know what's on their mind. You can get to know them, you can talk to 'em about their families and see what's going on in the rest of their lives. That is so, so valuable at least twice a year. I think you should get together face to face for a couple of days and just spend the money and spend the time and do it. And the outcome will be orders of magnitude greater than you can ever imagine simply because of the discussions and the interactions you'll have that you would never have over the phone or over zoom.
Speaker 1 00:02:31 And those are essential things for strategy, for clarity of decision making, for new directions, for understanding what the problems are in the organization. We have to get together. And so I was reminded of that and it was really, really a valuable experience. And anybody who's interested, I can tell you more about the way we do it. Then there's the issue of ai. Now we're probably gonna see floods of articles in the New York Times articles in every magazine about AI entrepreneurs, venture capitalists, taking all their money out of crypto and putting it into AI companies renaming their websites with AI at the end. This is not a new technology. AI's been around since the 1950s, and if you read the history of it, it's people have been working on vision, language, natural language processing, auditory misunderstanding, understanding the way the brain works and trying to mimic this in computers for a long, long time.
Speaker 1 00:03:27 And it has just picked up speed because these large language models like G P T three or Bard are highly networked billion plus node neural networks that can do a pretty good job of mimicking human interactions. And if you read the articles about them and talk to computer scientists, they do have a life of their own. They tend to learn through the interactions we have with them. So if you chat a pragmatic question to a chat bot that has a large language model, it will give you a pragmatic answer. If you chat angrily to it, it will give you a different kind of answer. If you challenge it, it'll give you a different kind of answer. So it appears to have a personality. Now of course, it doesn't really have a personality like a human being. We have emotions and we have long history of relationships that we bring into conversations.
Speaker 1 00:04:21 We have auditory processing that these chat bots don't have, but they're pretty sentient in many ways. And so there's dozens and dozens of articles out right now on why they're dangerous and why they're um, a mistake and why they're being oversold. And yeah, I mean, I can argue all those things are probably reasonable challenges, but every technology that's ever been invented had problems for some period of time. And the upside of this is so massive that we're gonna have to just plow through the challenges and get to the solution that we want. Because what they can do is they can amass and understand literally billions of nodes of information in a communication form that is unlike anything we've ever done before. And in our world of HR and HR tech, this is going to be very transformational. I've mentioned in the last podcast training, knowledge management, employee experience, support, certification, leadership development, coaching, selection, assessment, scheduling, onboarding, transition management, many, many, many of the programmatic things that we do in HR or we have to design something and we have to dream up a program in this area and then we have to deploy.
Speaker 1 00:05:38 It can be done through ai. And I am very sure that over the next year, most of you are gonna do this and you're gonna get tools and vendors are gonna provide solutions that do this and it's gonna be fantastic, but it's not gonna be cheap. Because one of the things that I've been learning about the AI engines is they're not indexes like search engines at Google. They're intelligent nodes with billions of nodes that use linear processing to compute answers. And they take two to three orders of magnitude more compute processing than a typical search. If you type into Google who is the president of the United States or some similar search to that, it's doing some sort of a beat tree index, which is very, very efficient to find the most likely answer. These don't do that. They actually traverse this node of this massive network and using statistics, they compute the most likely answer based on the words and how the words come together and what words belong before and after and with others.
Speaker 1 00:06:40 So they cost a lot of money. So a startup AI company that's trying to build a large language model from basically what I've been reading, is probably gonna spend a minimum of four to $5 million just to run the model. Forget about hiring peop and doing design and building a website and doing sales and marketing and all that just to get the model to work. So as interesting as this technology is, the business model side may be the biggest challenge. In the case of Microsoft Bing, the company's got a lot of money and they're willing to expend their Azure processing on open AI's chatbot to get bing up and running. But when you consider the fact that a query on Bing might be hundreds of times more expensive than a query on Google and they're selling ads, it may not make a lot of money. So these AI systems have to have big ROIs.
Speaker 1 00:07:35 It's not clear to me that search and advertising is necessarily the best location for this yet. I think some of the other applications are much more high ROI and more financially viable. Um, I'll probably be proven wrong, but for example, I was talking to an engineer over the weekend when we were out of town about some of the things that can be done in basic database processing. So let's suppose you go into your financial system and you say, how are the sales of winter jackets this month versus last year at this month? The system goes in, it does the comparison. It determines that they're maybe 30% lower than they were a year ago. And it starts asking questions. Was it the seasonal weather change? Was it the availability of materials? Did we have fewer stores? Did we have more stores? Did we spend more money on marketing?
Speaker 1 00:08:24 Did we spend less money on marketing? Did we have different colors? You know, there could be thousands of factors that weigh into that change that human beings may not eventually find. They could or they may not. I don't see any reason why AI couldn't do that. That's the kind of thing one of these large language models could do. And think about that type of a question in virtually every part of our work in the human resources area where we're trying to figure out why we're having turnover, why we're having retention problems, why we have low performance, why we have management issues. I think these machines are going to be very, very successful at doing this. And they're gonna take talent intelligence and they're gonna make it super intelligent. Now, we're still early days here. The systems that are being promoted now in the market are not open yet.
Speaker 1 00:09:13 And these large language models are not completely available to IT departments. Although Facebook just announced the availability of one. I don't know if it's any good, but it's probably got a lot of r and d into it. So if you have a lot of money and you have a lot of horsepower in your IT organization, you can start building one of these things. You can use it for knowledge management, you can use it for all these training applications. I think we're gonna see hundreds of new applications of this. And for the big vendors, Oracle, Workday, sap, adp, all of the payroll providers, all the HCM providers, including the large learning companies, certainly big companies like Cornerstone, they're gonna have to get good at this because this stuff obsoletes a lot of the traditional linear processing that we do in our traversing of technology to implement many, many of the solutions that we do.
Speaker 1 00:10:02 So I am not buying all of these GAD flies who think that this stuff is going to hurt us. I mean, every piece of technology we've ever invented had the potential to hurt us. And maybe it did a little bit, but the upside of it was massively positive. I do think, by the way, one of the examples that's often given is the self-driving car from Tesla. And Elon's main argument for the accidents and deaths that have been created through the AI and the self-driving car is that in general, the self-driving car is something like a hundred times safer than human-driven cars, according to him. However, that's not a validated test. Nobody's looked at the data, he just says that. And most of the studies I've seen say that in that particular case, the self-driving car is usually used on open roads. It's not used on highly complex driving situations.
Speaker 1 00:11:00 And so there are factors that make it look like it's safer than it is. So the ethical question you have to ask yourself in that case is, are you willing to be a Guinea pig to test the self dive in car while Tesla makes it better and better with your life in somebody else's hands? I think the answer is most of you would say no. So there's all sorts of ethical questions on how far we push these things to get them to become smarter, and what risks are we willing to take? These ethical issues are one of our responsibilities in HR because many of the things that we deploy AI for will directly impact job seekers, candidates, employees pay promotion, and very important issues. So we have to rely on the vendors who sell us this stuff for testing and validation in anti-bias algorithms on these large language and large data systems.
Speaker 1 00:11:51 But we've been doing that for for a while. And there are now laws entering certainly New York State and other places that are requiring vendors to test things that are used for human capital decisions. We could get sued. So I'm not minimizing the potential risk of some of these things, but my gut feel on the technology I've seen, given the guardrails that Microsoft is now put in it, by the way, most of you know Microsoft Limited the number of queries you can do in Bing, so you can't get it to go off kilter like it has in the past. You're gonna do some amazing things with this stuff, and we're starting to talk to more and more HR tech vendors about this, and we'll talk about a lot more over the next couple of months for sure. So that's a pretty exciting thing. Now, the second thing I want to talk about relative to HR is the economy.
Speaker 1 00:12:36 And the way I see it, we're in a very confusing time that forces you to think about productivity. And here's why. The job market is incredibly competitive. We have a shortage of labor, shortage of skills, shortage of leadership in virtually every country in the world. Even if the economy slows down significantly, that's not gonna go away. It might become a little bit better, but we just don't have enough people. The economic output per human being has gone up a lot, but we don't have enough human beings. And any economist will tell you that no economy grows when the population shrinks. And that include companies too. Now, sometimes companies can become hyper efficient with a very small number of people, but that's fairly unusual and it's difficult to do, and the economy is slowing. We don't have a recession, at least in the us, but profits are slowing.
Speaker 1 00:13:29 Consumer spending is slowing, bank balances are going down, credit debt is going up. I mean, all of the evidence is that we're gonna keep on slowing for a while. We're not out of this at all. If you can sort of ignore the stock market, it's very emotional. And so you're going to be asked to work on projects to make people more productive, to reduce costs, to improve efficiency, whatever language your CFO uses. And you're gonna have to figure out what your role in this. I firmly believe, and we will show you how to do this, that HR is one of the most important domains in business for making this happen. We know how to make teams productive. We know how to set goals. We know how to train leaders. We know how to do organization design. We know how to reduce bureaucracy. Hopefully, at least you should know how to do that.
Speaker 1 00:14:17 And AI, by the way, will be a big part of this. And as we learned in our org design study, by the way, which is coming out in a super class in the JBA in another couple months or less. So stay tuned for that. You should be taking time to work on productivity projects. You should fork off some of the resources in the HR department. I talked about systemic HR last time. We can talk to you more about how to do that. And you should be looking at sales productivity, supply chain productivity, manufacturing, productivity, whatever the the part of your company that you think is the most scalable. What can you do to do more with less human effort with fewer people, fewer costs, fewer, fewer expense. And you'll be surprised how much of that is right and centered in the domain of hr. I know this because for myself as a business person, maybe first and analyst second, or maybe the other way around, I'm not sure I think about this all the time.
Speaker 1 00:15:12 Every time we hire somebody, I think to myself, is this person going to increase our productivity or revenue per headcount or decrease it? And if they decrease it, when is it gonna turn around and grow again? And what are we gonna do to make this person more productive? Now that we have more people? And as you've read in some of my articles, the last couple of months, most companies were going the opposite direction. They were hiring more people and becoming less productive step by step as they were layering off more on more bureaucracy. And this is where the AI is gonna fit in these tools, which are expensive. They cost a lot of computing power. They're not inexpensive systems to build a large language model, but the vendors are gonna offer them at a low price. To get us hooked are one of the ways you can improve productivity.
Speaker 1 00:15:55 You can improve productivity by improving accountability, by making teams smaller, by training leadership to be more focused, by giving people more time and flexibility to improve their skills by moving people around so they're not stuck in jobs that they're not good at, or jobs that you may not need people to be doing anymore. And obviously by skilling, skilling, skilling people all the time and giving them time to learn from each other, to learn from their mistakes and share things that they've learned in the organization, technology will be a part of this. Many, many technology projects do not improve productivity. In fact, a lot of H HCM replacements do not improve productivity unless they're part of a transformation. If you read our HR technology Definitive Guides, you'll see that HR technology projects have to be coupled with organizational transformation and job transformation work also, which is really exciting and interesting and fun.
Speaker 1 00:16:49 So that is going to be our mantra for the next year. Anyway, that's kind of my comments for this week. It's been a very big week with our offsite and my very, very beautiful, lovely young grandson who, I won't mention his name yet cuz my son is still trying to decide. But we've had a very, very interesting week. Stay tuned on the AI stuff. I'm gonna be talking about this at the HR Technology Conference coming up as well as some other events. And we're doing a lot of work right now on the new HR operating model. I talked about that. More work on internal mobility and the use of talent marketplaces, a whole bunch of projects on the gig workforce, what I call the pixelated workforce, which is now 36% of Americans work in a contingent basis. That is a massive, massive number. And and also updating the J B A our Academy and lots of new things coming out in the global HR capability model. Have a great week and let's hope this weather gets better soon. Bye everybody.
