Complete Intelligence

Categories
Week Ahead

Will AI Take Your Job? Exploring the Realities of Automation

Explore your CI Futures options: https://completeintel.com/promo

In the latest Week Ahead episode, three experts – Todd Gentzel, Chris Balding, and Sam Rines – discuss the impact of AI on the job market and the enterprise.

The conversation delves into the macro environment and the rise of AI, with Sam Rines framing the discussion by noting the fast adoption of AI tools like ChatGPT and Midjourney, which are taking out low and mid-level writing, creative, and analyst tasks. This is a threat at a scale not seen before as this generation of AI is targeting professional, corporate, and office jobs.

Todd Gentzel, who has consulted and led strategy for some of the world’s largest companies, discusses the current state of AI in the enterprise. He notes that many AI projects are just pet projects to tick a box and the “AI” portion of these projects is extremely limited. However, he believes that AI has the potential to change the enterprise significantly and identifies the factors holding the enterprise back from adopting useful AI.

Chris Balding, the founder of an AI-NLP firm, discusses whether AI will steal jobs. He notes that starting his firm has changed his view of the application of AI and its potential to take on whole job functions. The conversation covers the impact of AI on labor and capital, the potential for AI to be deployed to take on individual functions, and whether AI can only be used to augment job functions or take on whole job functions.

The discussion raises important questions about the impact of AI on the job market and the enterprise, and how it will change the way we work. While the experts have different perspectives on the potential of AI, they all agree that it will have a significant impact on the economy, the job market, and society as a whole.

Key themes:
1. Is the macro environment to blame for the rise of AI?
2. How will AI change the enterprise?
3. Will AI steal your job?

This is the 60th episode of The Week Ahead, where experts talk about the week that just happened and what will most likely happen in the coming week.

Follow The Week Ahead panel on Twitter
Tony: https://twitter.com/TonyNashNerd
Sam: https://twitter.com/SamuelRines
Todd: https://twitter.com/ToddGentzel
Chris: https://twitter.com/BaldingsWorld

Transcript

Tony

Hi everyone, and welcome to The Week Ahead. I’m Tony Nash. Today we’re joined by Todd Gentzel. Todd is an industry and technology strategist spanning healthcare, mining, oil and gas, transportation, and consumer goods. Todd, it’s your first time on the show. Thanks so much for joining us.

Tony

We’ve also got Chris Balding. Chris Balding you guys all know well from Twitter. He’s the founder of a stealth mode AI firm, and he’s also the founder of New Kite Data and a recovering academic.

Tony

We’ve also got Sam Rines of Corbu, who’s on here regularly. So guys, I really appreciate your joining us for the program today. This means a lot.

Tony

I’ve wanted to look at the hype around AI for quite some time. For non-experts, it’s really hard to tell what’s hype and what’s real. We see stuff about ChatGPT or whatever every day, and we can’t tell what’s real output, what’s simulated output, or whatever. So we try to assemble you guys, some experts, to tell us what’s happening. And there’s some real critical answers that we want to address. Why is AI on the rise right now? There are some reasons why AI is coming to the forefront right now. So what are those?

Tony

Will it take your job? A lot of people are, and some people are joking about that. Some people are taking it seriously, some not. But really, will it?

Tony

How will AI change corporate life? What impact will AI have on markets and regulations and so on? These are all things that we don’t know all the answers to right now, but we’re kind of figuring this out as we go along.

Tony

So, just over a year ago, I published a fairly rudimentary illustration showing the pace of impact that I thought at the time AI would take in the workplace and on jobs. So if you notice at the bottom, most of the kinds of infield jobs are retained. A lot of stuff has to physically happen. And my view, at least over the next, say, a few years, is 5% to 10% of jobs need to be automated. And I think that’ll largely grow toward the end of this decade.

Tony

So we have some key themes. First, is the macro environment to blame for the rise of AI? I think that’s a real concern, and we’ll talk about that with Sam. Second is how will AI change the enterprise. We’ll talk about that with Todd. He’s a real expert there, and I can’t wait to have that discussion. And finally, will AI steal your job? That’s kind of a silly question, but I think it’s one that everybody really wants the answer to, and we’ll talk about that with Chris.

Tony

So first, Sam, I want to frame up the discussion with a little bit of an understanding of the macro environment. We’ve had AI enthusiasm before. You have these really robust AI eras, and then you have kind of AI winters. We had a really robust era in 2018 when S&P bought a company called Kensho, which very few people talk about now.

This was just five, or six years ago. They bought Kensho for $550 million and really, nothing happened with it. They were folded into S&P. At the time I talked with people who had visibility to Kensho. They didn’t know what to do with it. It really wasn’t obvious value. But S&P kind of got the opportunity to tick the box on AI. So, in part, S&P wasn’t adopted by S&P’s customers. At least this is my running thesis. It wasn’t adopted by S&P’s customers because wages had been pretty stagnant for 30 years.

Tony

So even in 2018, you could kind of throw people at analysis problems and the type of things that Kensho was built to solve. But now we’re seeing ChatGPT, MidJourney, and those types of large language models and image models being adopted pretty quickly.

Tony

ChatGPT, as you guys know, had millions of users in the first hours, in the first couple of days. So we can say that processing power and coding and that sort of thing are responsible for advancement in AI, which is true. But adoption seems to be different than the actual capability. So when we see ChatGPT and MidJourney adopted so quickly, they’re really taking out low and mid-level writing, creative and analyst tasks. That’s what they’re taking out right now, are those tasks. These are things that earlier had 10-15 years ago, had been sent to, say, India and other offshoring places, but now it’s being experimented with doing this stuff virtually in developed countries. So I realize I’m talking a lot today. I don’t normally do this at the top of the show, but I think we need to introduce some of these ideas for people to watch.

Tony

I’m sorry I’m talking so much today, but one key point here is that AI has always been discussed more than robotics. So where it would take over the job of physical laborers, like people in warehouses, blue-collar workers, as Americans would call them. But this generation of AI is different. This generation is targeting professional jobs, corporate jobs, and office jobs, which are new. It’s kind of unprecedented, where this level of fear for white collar jobs is discussed to be replaced by technology. So, Sam, after that long intro, can you talk us through some of your thoughts on this? This is my hypothesis. Is there anything there? Can you talk us through some of the kind of capital versus labor and wage issues that we’re seeing right now? And is that having an impact on the adoption of AI?

Sam

Yeah. So don’t throw too much at me at once. Okay, so let’s take a big view of the history and kind of parse this out, because I do think it’s worth kind of going back to previous periods to look at what exactly spawns the adoption of various technologies. Because AI is a technology and it’s incredibly useful for those people that want to become, or can become much more productive over time. So I think that’s kind of the level set there. But if you look back at 70s and the level of inflation there, it spawned a significant amount of capital investment in things like computers, right. It was expensive to hire an individual, inflation was running out of control, and you wanted to maintain your margins if you were a corporation. So what did you do? You made people more productive by employing technology, specifically the computer at the time. Right. It sounds kind of ridiculous to say that the computer was a productivity enhancer because we all know that now productivity is not necessarily enhanced by a computer in front of you. But then it was incredibly enhanced for productivity. So when you have significant inflation pressures against a business, it spawns the want and the need to go ahead and invest in incremental technologies.

Sam

So kind of fast forward to COVID, and if you were a leisure and hospitality company or a company that faced individuals, you had an incredible incentive to invest in an underlying technology to allow your business to either exist in a couple of years or to survive and maybe even thrive. If you were very good at it. You had to go out and you had to make sure that your website could offer delivery or pickup options for food. You had to really invest in technologies that previously didn’t necessarily have to do. Were they emerging? Were they interesting? Yes. But all of a sudden they became existential to your business and the ability to survive going forward. So you saw an incredible amount of investment in platforms that allowed for delivery and pickup of food, et cetera. Kind of coming out of COVID. Now what you have is an incredible shortage of workers and a significant amount of wage pressures, and you have inflation pressures. So if you’re a business looking to maintain margins, grow going forward, AI is an incredibly interesting potential tool for you to be able to make some of your best workers and best thought leaders and intellectual leaders much more productive and allow you to grow going forward without having to worry about whether or not you’re going to be able to find that incremental employee.

Sam

And I think that really is an understated catalyst for why ChatGPT-4 is so incredible, right? I love it. It makes me a lot more productive at my job. I’m still playing with it and I don’t actually publish anything.

Tony

Can I just give you a tangible example of what you’re talking about? I know that you understand this Sam, but for our viewers. So my staff last week put together a persona in a large language model and called it Nash, and it looked at all of our previous shows of The Week Ahead and then it came up with a persona for Nash. So last week’s newsletter, Complete Intelligence Newsletter, and going forward, they’re largely written by this persona in Chat GPT. So we don’t have to spend the time anymore to actually write our newsletter. Of course we clean it up a little bit, but it has my voice, it has my word choice, sentence structure and so on. And so largely our newsletter is automated and of course there are little tweaks here and there, but for the most part those are the types of things where maybe I had to hire a newsletter person before, even if they were offshore. But now it’s done in three minutes.

Tony

CI Futures is our subscription platform for global markets and economics. We forecast hundreds of assets across currencies, commodities, equity indices, and economics. We have new forecasts for currencies, commodities and equity indices every Monday morning. We do new economics forecasts for 50 countries once a month. Within CI Futures, we show you our error rates. So every forecast every month we give you the one and three month error rates for our previous forecast. We also show you the top correlations and allow you to download charts and data. You can find out more or get a demo on completeintel.com. Thank you.

Sam

No, again, that’s productivity enhancing for your team, right? And it allows you to say, okay, now that we really kind of come up with a way to automate this newsletter, what else can we do? So it allows you to be not only productivity enhancing, but potentially revenue enhancing, potentially bottom line enhancing, producing new products, new services, et cetera, et cetera. So in my mind, that is the one of the tailwinds to AI adoption at this point is that you really have not only called a curiosity with it, but also a need to replace the incremental employee because you can’t find them. If the incremental employee doesn’t exist, you’re not destroying jobs, you’re creating/enhancing ones that exist. The idea I’m kind of running ahead of us. I know, sorry. But to me that’s really the catalyst behind the current adoption, right? And if you look at one of the most labor intensive businesses out there and we kind of touched on this while we were chatting before reporting if you look at agriculture, I mean, John Deere has been working on AI tools for farmers for a decade and has bought up a significant amount of IP around that to not only allow farmers to become much more productive, but potentially make it so the farmer doesn’t have to be in the tractor during planting, during when they’re spraying the plants early on and during harvesting, the farmer can go do other stuff.

Sam

So I think as we begin to really understand that there aren’t enough farm workers out there. That there aren’t enough people to hire into various businesses, I mean, just look at the participation rate. The participation rate is not exactly coming back the way anybody thought it would after COVID, and it’s unlikely that it’s going to recover anytime soon with the number of retirees. Retirees have a significant demand for services. If you’re going to provide those services, you’re going to need to not only adopt new technologies and new tools, you’re going to have to come up with new ways of doing things generally. So I think AI always was going to be something interesting, but it’s something interesting at the right time with the right catalyst moving forward. And this is not something that’s going to be… There’s a little bit of fattiness to it in different ways, but I don’t think it’s going to be one of those passing fads that everybody’s like, “remember when AI was a thing?” I think it’s much more of something that we’re going to interact with on a daily basis across a whole lot of services and a whole lot of businesses that we did not anticipate prior.

Tony

So two things there. Technology generally is deflationary, right? I mean, aside from like $1,400 iPhone or whatever, generally, technology is deflationary for kind of status quo activities. Is that fair to say?

Sam

Sure.

Tony

That’s good. And then you said something like, we’re going to X with AI. But people are already experimenting with that stuff. So we do have people who are already doing that. And it’s really a question of it going at things going broad market. Like, I don’t want to be the AI hypester here. I’m really just kind of asking these types of questions just to understand your view on this stuff.

Sam

Sure. I think it’s pretty straightforward. Right. You have to have some way of replacing a nonexistent labor market, and AI does that in a fairly efficient manner.

Tony

So it’s demographics, wages, participants, demographics, wages.

Sam

Demographics change slowly than all at once. It’s not as though you can simply incentivize the demographics to change. Right?

Tony

Exactly.

Sam

That ship sailed a long time ago. Generally, to your point, demographics are a powerful force where when you have a significant amount of people that are older and out of the labor force demanding a significant amount of services, you have to figure out a way to deliver those services into them. With fewer people in the labor force, which is a massive long term catalyst to tools like AI, like ChatGPT, that type of thing, and it’s not going to stop there.

Tony

Yes. Okay. Good points. Okay, so let’s move from the kind of context and thanks for that, Sam.

Tony

Let’s move into how will AI change the enterprise? Todd, you’ve consulted and led strategy for really some of the world’s largest companies. In enterprise circles, we hear about AI projects from big consulting firms or a firm like Palantir, which really is a consulting firm. These are largely pet projects to tick a box. But at least in my mind, the kind of AI portion of these projects is extremely limited at this point. So given the economic context that Sam discussed and the corporate dynamics that you’re aware of, is AI in the enterprise a real thing right now?

Todd

Yeah, I think that you probably have to break it into a couple of groups. I think the earlier statement about agriculture and John Deere is true in oil and gas is true in healthcare. I mean, there are lots of companies that have been at this for a while, and they’ve got relatively mature environments, and in those environments, they’re really playing a different game. It’s not a check the box. It really is kind of fundamental to business models. I think there’s sort of a sort of much larger group of organizations that are just beginning to be aware of the opportunity in the kind of intermediate and long term. I’m super positive. I think this is unquestionably, the direction this has been headed for a long time. I think in the short term, we’re going to see what we always see during these periods of technical transition. It’s going to be messy. I think it’s important to always remember that there are real power dynamics around any adoption of new technologies. And in a lot of cases, the people who are in leadership and the people who are making these decisions are the authors of the current state.

Todd

And so they struggle to sort of conceptualize what the world would look like under a completely different set of norms. And I think unlike some of the previous generations of technical advancement, I would argue we’re coming out of the age of digital enablement. We’ve talked about transformation. I think there’s been very little transformation. I think it’s mostly just enabling some core things we were already doing and gaining some minor improvements in productivity. AI is one of a dozen exponential technologies that plays a very, very different role in accelerating innovation and accelerating business model development and changing operating models. That’s where things get really dicey. And I think there are going to be winners and there’s losers. And I know, Tony, you and I have talked over the years about when you do scenario planning, you sort of right off the bat, assume that there’s really no good or bad future. It’s good for some and it’s bad for others, and I think that’s going to be true here. I think what we’re going to see is there are organizations who have spent the last decade really creating the kind of agility, the kind of resilience that’s necessary to make a transition like this and really capitalize on it.

Todd

And there’s going to be some organizations that really struggle. And that’s why I actually think that this may not be the age of the incumbents. I think that the people who are really intending to disrupt have a window of opportunity here while people are kind of working through the internal dynamics of what it means to adopt these new technologies and brand new ways of working. People who are unencumbered by those cultures and those kind of leadership norms are going to be able to move much more quickly and likely be able to sell into that world. And I think that’s going to give rise to a whole new group of consultants. I think there’s always the system integrator model and we’re going to sell the big thing and we’re going to work it out over five years and rest of that. I think that the people who will play most prominently in this next phase really are hyper specialists and they’re going to come in and they’re going to solve significant real problems.

Tony

When you say that, I think you said the current operational architecture is a reflection of the current leadership or something like that. And it sounds like they won’t change willingly. Just to be a little bit brutal here, is there going to have to be a wave of retirements or something like that for AI to really hit larger firms or what would push larger firms to attract or to adopt really interesting levels of, say, technology and productivity?

Todd

I think that we’re at a kind of a unique place where a lot of the things that made us successful in the past are the things that actually inhibit our progress. And you know, if you’ve got folks who are relatively intransigent, I mean, really the only option is to move on. We used to have a firm I worked for. This sounds really crass. We had a phrase you either change the people or you change the people. And I think we’re at that kind of a moment where if you find yourself in an environment where the leadership and the operating norms really are not particularly conducive to making these key pivots, everything Sam said is right on the money. I mean, these are economic realities. You’re going to have to make these changes to remain competitive and you’re going to have to find a way to a new way of operating that will allow you to do that again and again and again. Because this isn’t an embrace AI. It’s embrace tool after tool after tool that’s solving these problems. It’s a very different discipline, but it’s also spinning up a bunch of interesting challenges. I was just talking to somebody this week that was working on some things around material science and leveraging AI in that space.

Todd

And we are so rapidly spinning up new materials that it’s difficult to find people who are capable by way of their training, of conceptualizing the utilization of those materials. And so these opportunities in some cases take a little while not just to ingest but to train up people to leverage these to their full extent. Which is why I think the short term is going to be really a story of fits and starts. There’s going to be some big wins and there’s going to be some significant resistance. One of the places where I’m kind of most interested right now is what was mentioned earlier about sort of the top of the food chain right. You’re talking about very elite, top level professional jobs. We’re already seeing some really incredible things in the healthcare space around second reads of scans.

Tony

What does that mean, second read? Can you walk us through that process? Yeah.

Todd

So the radiologist takes a look at your X ray or MRI and says, this is what I see. And then it automatically goes out to an AI engine that goes in and makes sure that everything was caught. And what we’re finding is that we’re routinely catching things with the AI. Well, that’s beginning to tell a story, not just about supporting the work of a radiologist, but potentially, over time, the machine actually becoming a superior mechanism to leverage as a first read and a second read, and you can actually create alternate models. And these are things that are not science fiction. These things are already happening. These are institutionalized systems are doing it really to mitigate risk. I now can say I’ve looked at it multiple ways, and we feel fairly confident at what we’re seeing. That’s happening in industries right now, where we’re actually seeing real life, serious use cases that are mitigating risk, lowering costs, improving outcomes that needs to be scaled. And that’s really what I’m getting at. I think that you see these really interesting spot treatments, right, where we’re looking at something saying, I can solve that. The question is, how do enough of those actually begin to be leveraged?

Todd

It becomes a way of working rather than just a tool in the box that we go to in very specific and very narrow circumstances.

Tony

So what about those people who say, “oh, I’ll never let AI be my doctor, I’ll never have a robot for a doctor, or I’ll never let AI be my CPA” or something like that? Will they have a choice?

Todd

Yeah, I don’t know that they will. I will tell you that there’s some pretty sophisticated tools that are already on the market that are very close to being able to achieve the same level of efficacy and diagnosis as the very best physicians that we have. When you think about that as a language model, I mean, if you think about, like, a Physician Desk Reference and you’re asking questions and you’re getting the medical history and you’re making decisions and there’s things that the machine is capable of doing that’s, just far more capable in the human mind in evaluating the different levels of risk and the likelihood that this is what I’m seeing versus this other thing. Because we’ve seen such a remarkable advancement just on that front in the last four or five years, and you’ve seen its adoption. You look at the NHS or you look at Medicare and you say, there’s absolutely no way, at least at that first level of diagnosis, that we’re not moving very aggressively in that direction for a lot of reasons. Number one, it’s much cheaper, but number two, it’s super available. It’s easy access. We’re actually catching these things long before they become genuinely problematic and cost the public a whole lot more by way of health care dollars.

Todd

So I get it. I understand it. I think there’s sort of an impulse initially to say “I’m very uncomfortable with that.” But increasingly there is a whole lot of diagnostic stuff that’s happening behind the scenes that people aren’t seeing that’s already in place. That’s pretty significant part of their care.

Tony

Right. Okay, so this is where I’m going to give a little shameless plug for complete intelligence, just to give people a little tangible idea of what can be done.

Tony

So we do budget forecasting for companies, and we have one company, a client, $12 billion in revenue. They have 400 people who take three months to do their annual budget process. We did that in 48 hours, taking one of their people less than a week of their time to transfer knowledge to us. We had better results in 48 hours than what 400 people did over three months. And this is a very tangible way of identifying the opportunity that’s available with AI tools and other technology tools. It’s not just replacement. It’s not RPA, robotic process automation. It’s not that it’s better. Right? And that’s where people should be a little bit aware, where we’re talking about doctors, we’re talking about people with MBAs, we’re talking about highly educated professionals where we can have a machine do that work better and faster. And that brings us to Chris Balding to give us great news, Chris. Thanks, Todd. I really appreciate that. And you guys jump in on this anytime.

Tony

Chris, the real question here is, will AI take my job? Right? My job? I’m hoping it does. But for most people, will AI take their job? I think you’re about to launch an AI NLP, a natural language processing firm. First question, I guess, is how has starting that firm changed your mind about the application of AI today versus even just a few years ago?

Chris

I think there’s this discussion about will it take people’s jobs? And if you look back on really any technological breakthrough from the cotton gin to fracking, what you really had is the per unit price would drop of a T shirt or how much it costs to get that oil and gas out of the ground. But what happened was it consumed people that had the technical training, higher levels of technical training. If you think about AI, people will say, well, hey, we don’t need as many coders. Well, you know, what’s going to happen is that opens up a whole new field of cybersecurity risks. And all those coder jobs are going to migrate into cybersecurity because all you’re doing is opening up cybersecurity risks, as a simple example. If you talk to any IT guy inside big companies or whatever, there’s typically a list of about 40 projects management wants them to work on, and there’s 20 that are constantly at the top of that field and they never get to those more advanced, maybe investment, longer term types of product. Well, if you’re able to blow through those 20 faster, as a simple example, you can move on to those more creative, risky type of projects.

Chris

So when I hear people talk about, well, it’s going to take my job, I think it’s absolutely going to change how people work. I think it’s going to change the types of jobs that we do. For instance, one type of coding might move more into cybersecurity. Is it going to eliminate these jobs so that the total level of employment disappears? Absolutely not. It’s just going to change how we work and the specific jobs we do.

Tony

So is it at least at this phase, is it more augmentation than it is automation?

Chris

So it really kind of depends on what you’re specifically saying. One of the things, and I think OpenAI has, has even said things to this effect, you know, we talked about macro and other stuff, but really, what has, what is undergirding this is that really, for the past, let’s say five to ten years, you’ve basically seen this exponential increase in AI type stuff. And that is really driven by, just to be blunt, the hardware of what you can do with GPUs. And part of the reason that we talk about this is, going forward, the amount of GPU capacity that you’re going to need is I mean, you’re going to start sucking down. I mean, the the amount of energy that they were sucking down from GPUs to do bitcoin will pale in comparison if it really takes off the way people say it will. I’ve used it for a lot of coding and similar types of things. And what you really see is, especially on more complex types of projects, you kind of use it to kind of seed what you’re doing, maybe take specific steps. It absolutely, I don’t think, is near the point where it can basically manage entire significant projects.

Chris

And so it’s absolutely a time saving tool. We talk about this with coders. It’s absolutely a time saving tool. Is it taking over their job? No, absolutely not. It’s going to help them do things faster, move on to more complex types of processes that they’re trying to automate.

Tony

Okay, but if it helps people do things faster, then that means they’re spending less time doing the job they have now. So somebody’s losing, right? Somebody’s losing a job, right?

Tony

Because if it’s helping people do stuff faster, then companies have to spend less time on headcount. Right? I’m trying to get out of the, hey, this is replacing jobs. But we kind of end up there with this type of technology.

Chris

Yeah. So think about it two ways. Let’s assume you have an IT department. All of a sudden, that IT department is doing less work, making sure that there’s not a paper jam at the printer and that the computer can talk to the printer. Okay. There’s less time spent doing that. But I guarantee you there’s hackers in Russia that are now using ChatGPT to say, “how do we break into this?” Part of the issue is that guy who started out in It is probably going to move over to cybersecurity. Okay? Or they might say, “hey, we can let go of a couple of people, but now we want these other guys to focus on these bigger investment type projects that maybe we had kept on the back burner because they just didn’t fit within our budgetary priorities.”

Tony

Okay, so those are relatively fungible skills. But if you’re like the Radiologist that Todd’s talking about, can those skills be repurposed to something else?

Todd

Well, honestly, I think it’s case by case, but I mean, Radiology is a great example and just health care generally. I think we’ve all probably heard that we have a nursing shortage and that you can’t find an endocrinologist and we’re constantly dealing with this really serious labor issue. A lot of that is because across the board in healthcare you have people really failing to operate at the top of their license because they’re spending an incredible amount of time doing the paperwork, meeting the CMS requirements. And so you have doctors who are doing 30% doctoring because the rest of their time is basically meeting all of the obligations to all the different stakeholders. Right.

Todd

I think what we’re likely to see is these people who are sitting in that sort of, again, that sort of top tier of kind of professional expertise, really spend more of their time doing value creating work. I think if you think about what’s really going on, we have effectively an opportunity cost that’s baked into everything that we’re just not doing because we’re doing all of these things that really don’t require somebody operating at that level.

Tony

Right.

Todd

What we’re trying to do. I think and I think this is really the way we should be framing the future of AI is that if you really get focused on value creation and you start talking about that opportunity cost gap, I need every one of these employees operating at the very top of their capabilities, regardless of whether they’re a physician or a coder. And I need most of their time being pushed against real value creating activities rather than all the stuff that really should be relatively easy to put off to this other way of operating. And I think you can be threatened by it or you can recognize that the greatest inhibitor to innovation over the course of the last decade has not been our ability to produce technology. It’s our ability to free up capable people to really focus on the innovative things that need to get done in order to make things go to the next level. This is that linchpin moment. And every leader ought to be asking the question like, “how do I maximize the value of every single human asset that I have and really get them operating at top their license.”

Todd

And if that’s not the focus, then this probably is going to be a challenging period and it will become about cost and it’ll become about reducing by way of eliminating positions. That’s not, I think, the way to go. I think that’s actually probably the wrong way to think about it. I don’t doubt that there will be people who will be in that trap because they just are going to have a hard time to make the move, but the smart companies are going to be able to understand that very quickly and move aggressively to make that happen.

Sam

Yeah. And I think that’s a critical point that should not be overlooked is you can be scared of it or you can embrace it and use it as a tool to enhance your one, your life, because none of us like doing the lower end of the spectrum stuff that we always have to do. If you use it to eliminate that and get to do the stuff that is much more highly value add, that is incredibly accretive not just to the business but also to your lifestyle in general. Right. I think embracing it and actually having a positive attitude about it and saying, how can I use this to make myself more productive and generally more happy? Because hopefully we’re doing things that we love to do. How do I use this to do that? I think it’s all about the mentality of approaching it rather than saying, “oh my word, is this going to take my job?” I think it’s a fundamental thing that if you think it’s going to take your job, it probably is simply because you’re not going to embrace it and learn and try to adapt to the new technology, you’re going to fear it and shut it.

Sam

And I think that’s going to be the fundamental difference between those that succeed with the new technologies that are coming and those that fail and fail in a meaningful way.

Tony

Yeah, but I think fear is a natural response to something like this. Right. I mean, we’re all kind of not all of us, but a lot of us are afraid of new stuff. We’ve had our same job for 10-20 years. We have a routine, we go in, we do our work, we leave it five and call it a day. That’s most people, the vast majority of people, and I don’t necessarily think maybe I’m a skeptic here and maybe I’m a bad person for thinking this, but as Todd you talk about people want to look at the greatest value add they can have within their job and that will help them from being kind of automated. I don’t know that most people think that way. Maybe they do. But I think most people are just kind of going in for hours to do a routine job and those are the things that are the most dangerous, I think the positions that are the most dangerous.

Tony

Before we kind of wrap this up, I don’t want people to think that I just kind of loaded this with people who I knew would have the same view as me.

Tony

So, guys, let’s take the other side of the table for a little bit. And I’m not accusing you of having the same view as me, but let’s take the other side of the table a little bit. Let’s assume that large language models and Chat GPT and all these things are overhyped right now, okay? What could stop the implementation of these technologies so that they aren’t adopted across companies and across the economy? What could stop this stuff? Chris, you’re muted.

Chris

I think one of the things is Todd has alluded to this is you’re going to need so basically the basic technology that ChatGPT used is really probably just ten years old. They just added a lot more data and a lot more GPUs. I mean, the fundamental technology is not new in the least. What you’re really going to need, what is going to stop this is now you have to get domain experts coupled with those tech geeks to say, what can we do together? So whether it’s an endocrinologist, whether it’s a financial analyst, whatever it is, and one of the things is outside of the mainstream that you’ve seen a lot, is how can you develop these language models that are providing very precise answers for very specific fields? I’m a tax accountant. I am an endocrinologist, I am whatever. So if you don’t bring those domain experts together with those tech geeks and you’re just stuck with ChatGPT, which is basically trained on the Internet, you’re going to get a lot of bad answers rather than being able to augment what those humans can do.

Todd

Well, I would go further on that and say that those domain experts are critical, especially at this moment in time, right? Like, you start thinking about healthcare, aviation, mining, oil and gas, places where there’s really some very significant risk, and you say, look, those domain experts working side by side, they see that risk coming, they bake that into the conversation. They talk about what to actually put in that learning model to actually create an environment where you accomplish those kind of incremental improvements, but without exposing the organizations to exponential risk. I would tell you right now, the issue is it’s early. And so there’s not a lot of domain expertise that’s actually fluent enough in this to have a dialogue that’s meaningful to kind of push this forward. And the risk that’s inherent to that is the sort of ugly pre adolescence, as we sort of learn our way into using the technologies appropriately, getting out over our skis and getting some things really profoundly wrong, that really creates sort of a downdraft, right? Like, oh, this failed, or this didn’t work or it opened up this massive amount of risk, that’s a human error question. That’s really just a function of moving more.

Chris

Just to kind of add to that, Todd. Give me 1 second, Sam. I’m sorry about that is one of the issues that especially in an issue like the medical field, and I’ve heard this talked about in multiple other fields, is humans are there for a reason and especially if there’s a license, if there’s legal liability, et cetera, et cetera. No human, no matter how good the technology is, even if the technology is demonstrably far superior to human, no human is going to turn that legal liability over to a computer without saying, I’m going to sign off on this, I’m going to check it. And as you said, Todd, that machine learning was basically double checking what the radiologist was doing, just verifying.

Sam

Yeah, to Todd’s point and to Chris’s point, and I think this is really important, if we don’t get the domain experts in there to actually help and make better decisions, better outcomes, better reporting by the by ChatGPT 4, 5, 6, 7, 8, we are going AI in general is going to end up being regulated in a meaningful way. It only takes a couple of really big incidences, car crashes, et cetera, before you end up with the FAA, before you end up with the Transportation agency, et cetera, et cetera, Department of Energy. However you want to look at it, the amount of regulation that will come down on top of this in a landslide like way if you don’t get it right from the beginning and have some sort of self regulating mechanism, whatever it might be, is another, I think, understated suffocating factor, right? There’s nothing that suffocates innovation like regulation. And if you don’t get it right and you don’t get it right pretty quickly the amount of regulation that’s going to come down on this, particularly when it’s consumer facing, when it’s labor facing, those are some very powerful lobbies that are going to absolutely hammer this if it’s deemed to be unsafe or dangerous. I mean, it’s that simple.

Tony

Interesting. So basically what I get from you guys is we’re likely to have at least a few years where it’s more augmentation, where those experts are feeding back into the models to help them understand what they do before these things can really go off on their own. Is that fair to say? So we can’t just open the box today, replace a bunch of jobs and everyone’s on government payments or whatever for the rest of their lives. It’s going to take a few years for this stuff to really get some practical momentum in the workplace.

Todd

I think that’s right. But I think to that previous comment, the industry has to be very careful to sort of self moderate here. I mean, there are going to be folks who really very diligently go about the process of ensuring that we do it right. And then there will be people who inevitably will play it fast and loose. It’s the folks on that side of the fence that actually create the downward pressure from the legislative and regulatory environment. And so it’s just kind of an interesting moment in time because it’s sort of the learning period that really puts it on a solid footing. But it’s also a period where there’s a great deal of volatility and potential for there to be some kind of significant things that happen that actually harm the long term ability to get it implemented in a way that makes sense for the public.

Tony

Very interesting. Yeah, I think that regulation point is so super important. Okay, guys, anything else to add before we wrap this up? This has been hugely informative for me. Anything else that’s on your mind about this?

Sam

I’ll just say don’t fear it. Use it. If you’re not using it, if you’re not trying to learn about it, then make it make you better or get out of the way.

Tony

Exactly. Watch a few videos, learn how to do some mundane tasks. Use it to your advantage and do things like we do with our newsletter. Just get some really routine tasks automated and then just start learning from there. So guys, thanks so much. This has been really, really valuable. Thank you very much. Have a great weekend.

Todd

Thanks, Tony.

Sam

Thank you, Tony.

Categories
News Articles

How AI-based ”nowcasts“ try to parse economic uncertainty

This post was published originally at https://www.emergingtechbrew.com/stories/2022/06/17/how-ai-based-nowcasts-try-to-parse-economic-uncertainty?mid=13749b266cb1046ac6120382996750aa

This month, the S&P 500 officially hit bear-market territory—meaning a fall of 20+ percent from recent highs—and investors everywhere are looking for some way to predict how long the pain could last.

Machine learning startups specializing in “nowcasting” attempt to do just that, by analyzing up-to-the-minute data on everything from shipping costs to the prices of different cuts of beef. In times of economic volatility, investors and executives have often turned to market forecasts, and ML models can offer a way to absorb more information than ever into these analyses.

One example: Complete Intelligence is a ML startup based outside Houston, Texas, that specializes in nowcasting for clients in finance, healthcare, natural resources, and more. We spoke with its founder and CEO, Tony Nash, to get a read on how its ML works and how the startup had to adjust its algorithms due to market uncertainty.

This interview has been edited for length and clarity.

Can you put the idea of nowcasting in your own words—how it’s different from forecasting and the nature of what you do at Complete Intelligence?

So Complete Intelligence is a globally integrated machine learning platform for market finance and planning automation. In short, we’re a machine learning platform for time series data. And nowcasting is using data up to the immediate time period to get a quick snapshot on what the near-term future holds. You can do a nowcast weekly, daily, hourly, or minutely, and the purpose is really just to understand what’s happening in markets or in a company or whatever your outlook is right now

And what sort of data do you use to fuel these predictions?

We use largely publicly available datasets. And we’re using billions of data items in our platform to understand how the world works…Macroeconomic data is probably the least reliable data that we use, so we use it for maybe a directional look, at best, at what’s happening. Currencies data is probably the most accurate data that we use, because currencies trade in such narrow bands. We use commodities data, from widely traded ones like oil and gold, to more obscure ones like molybdenum and some industrial metals. We’re also looking at individual equities and equity industries, and we track things like shipping times for goods—shipping times…are usually pretty good indicators of price rises.

Who are your clients, and how are the nowcasts used in practice?

Our clients range from investors and portfolio managers, to healthcare firms and manufacturing firms, to mining and natural resources firms. So they want to understand what the environment looks like for their, say, investment or even procurement—for example, how the current inflation environment affects the procurement of some part of their supply chain.

In fact, we’re talking to a healthcare company right now, and they want to nowcast over the weekend for some of their key materials. In an investment environment, of course, people would want to understand how, say, expectations and other variables impact the outlook for the near-term future, like, days or a week. People are also using us for continuous budgeting—so revenue, budgeting, expenses, CFOs, and heads of financial planning are using us…to understand the 12- to 18-month outlook of their business, [so they don’t have to have an annual budgeting cycle].

Tell me about how the AI works—which kinds of models you’re using, whether you’re using deep learning, etc.

There are basically three phases to our AI. During the pre-process phase, we collect data and look for anomalies, understand data gaps and how data behaves, classify data, and those sorts of things.

Then we go into a forecasting phase, where we use what’s called an ensemble approach: multiple algorithmic approaches to understand the future scenarios for whatever we’re forecasting. Some of those algorithms are longer-term and fundamentals-based, some of them are shorter-term and technical-based, and some of them are medium-term. And we’re testing every forecast item on every algorithm individually and in a common combinatorial sense. For example, we may forecast an asset like gold using three or four different forecast approaches this month, and then using two forecast approaches next month, depending on how the environment changes

And then we have a post-process that really looks at what we’ve forecasted: Does it look weird? Are there obvious errors in it—for example, negative numbers or that sort of thing? We then circle back if there are issues…We’re retesting and re-weighting the methodologies and algorithms with every forecast that we do.

We’ve had very unique market conditions over the past two years. Since AI is trained on data from the past, how have these conditions affected the technology?

You know, there’s a lag. I would say that in 2020, we lagged the market changes by about six weeks. It took that amount of time for our platform to catch up with the magnitude of change that had happened in the markets. Now, back then, we were not iterating our forecasts more than twice a month. Since then, we’ve started to reiterate our forecasting much more frequently, so that the learning aspect of machine learning can really take place. But we’ve also added daily interval forecasts, so it’s a much higher frequency of forecasting and in smaller intervals, because we can’t rely on, say, monthly intervals as a good input in an environment this volatile.