Lemon_tm
The pace of technological change has taken a giant leap forward and created opportunities for investors. ChatGPT has been compared to the invention of the internet, robots are being introduced to workforce. And agricultural technology is being more widely implemented to try to solve for a growing global population. Jeff Spiegel, US Head of BlackRock, Megatrend International and Sector ETFs, will help us decode some of these new technologies and better understand the impact on both our lives and our portfolios.
Transcript
Oscar: Welcome to The Bid, where we break down what’s happening in the markets and explore the forces changing the economy and finance. I’m your host, Oscar Pulido. As the stock market continues to see volatility with inflation and interest rate hikes, the pace of technological change has taken a giant leap forward and created opportunities for investors.
ChatGPT has been compared to the invention of the internet, robots are being introduced to workforce. And agricultural technology is being more widely implemented to try to solve for a growing global population. Today we’ll be discussing the top three technologies we see shaping the future.
I’m pleased to welcome Jeff Spiegel, US Head of BlackRock, Megatrend International and Sector ETFs, to help us decode some of these new technologies and better understand the impact on both our lives and our portfolios.
Jeff, welcome.
Jeff: Thanks for having me back, Oscar!
Oscar: Jeff, there was plenty happening during the holiday season – we had rate hikes, we had inflation, we had people maybe taking some time off. But we also had, notably, the release of ChatGPT, which sounds a little bit science fiction-y to me when I say it out loud, but maybe you can walk us through this technology and tell us how is it changing the world? How quickly do you see it progressing?
Jeff: Sure. I took some time off – I hope you did. Unfortunately, I spent most of my time off playing with ChatGPT, so that begs the question, what makes ChatGPT special? Well, it’s an advanced AI chat bot that can understand and create human-like conversation. It was released by a firm called OpenAI, the tools free to use, it quickly captured the attention of social media. It gathered over a million users just in its first five days, which is a record for any sort of application.1
So, the G P T stands for generative pre-trained transformer. It uses a transformer algorithm which is based on a neural network or a computer program that mimics the way the actual human brain works. It was trained using a massive amount of data from the internet, books, Wikipedia articles, and other pieces of writing that amount to about 300 billion words.2
As I said, I’m guilty of having spent way too much time playing with it, seeing how it thinks, seeing whether I could trick it. But the tech is, at the end of the day, a lot more than a novelty.
Jeff: So, two ways in particular that we think ChatGPT could really revolutionize the world.
The first is in search, so GPT could really upend the search model as we know it, and really as we’ve known it since the 1990s. Now, that search model has fostered a massive business, $680bn in digital advertising that occurs on platforms like Google, and that can be disrupted if you think about it, if you don’t actually need to look at a list of hits, but if you could just ask the generative AI a question and be told the answer.3 That would be a massive disruption to a lot of existing businesses.
So, the second area is in the services industry. GPT can perform a wide variety of tasks that are currently done by humans, or at least augment them.
So, with the US still facing a labor shortage, about 4 million workers4 and aging populations around the globe, companies able to create AI tools that enhance productivity and efficiency of their employees could really benefit.
A lot of jobs required good research, vocabulary, and grammar skills. ChatGPT can actually level up a huge portion of the population that may not be as good at those things today.
It’s also worth noting that GPT is only one kind of generative AI – it’s trained on words. Other generative AIs can be trained on other data – visual data. In the case of DALL-E, which is an AI artist also from OpenAI, as well as AIs that use neural networks to process, for example, spatial or numerical data.
Actually, the ultimate breakthrough, what we’re really looking for in AI, and what a lot of people in the field are most excited about, is what’s called Multi-modal AI. Now, that’s AI that can be trained in all of these ways at the same time. To really mimic a human and to operate in multiple domains in the real world, you have to be able to master more than just text-based information – it can think in text, it can think in numbers, it can think in space.
Oscar: So, you mentioned a number of things, and going back, the disruptive nature of this technology to some existing business models like search engine, you talked about how it differs from similar AI technologies, but you also touched on how you were using it and you didn’t realize you were interacting with it, so it tricked you. Therefore, I guess the question is there any safety concerns, with respect to the usage of this technology, or what should we be thinking about if it can trick a human?
Jeff: So luckily, I knew what I was dealing with, but if I hadn’t been told in advance, it probably would’ve tricked me, and so to your point, that’s the danger. People who were testing ChatGPT were quick to realize that they could do a number of things that maybe were a little less than ethical.
Coders realized that they could use the tool to win quick coding contests. Teachers have caught many a student using the bot to plagiarize essays, and some people who uploaded their photos to a DALL-E like app received back renderings that felt sexualized, that made them look skinnier or buffer, younger, even nude.
Jeff: So, the saying “garbage in and garbage out” really applies on this latter point. To the extent that the data available to train AIs has pervasive bias or represent societal features that may not be ideal, an AI generally can’t make that determination. It can’t filter out that sort of bad data – or inappropriate data – effectively.
Solving for this is going to be critically important, and it’s really going to be a hotly contested debate, because if you think about it, there’s a lot of subjectivity involved in where bias lies and what’s culturally positive versus what’s culturally negative amidst all of this massive amount of data, or remember 300 billion data points, that the AI can absorb. How do you sift through that?
Jeff: But we do think there are ways to use these safely and effectively and ethically today. And probably three things to keep in mind.
The first is understand the limits and consequences of what an app like this can do. AIs can fool us into thinking that they understand more than they do, and that can cause problems. ChatGPT gives really confident answers, but it doesn’t think independently. And if it takes as gospel, for example, a Wikipedia page, it may very confidently make claims that are debatable or simply outright wrong.
Second, know what you’re giving up. Every time we ask ChatGPT a question, the conversation is logged and it’s stored.
Sometimes it’s analyzed by a human to make improvement, sometimes not. But either way, it’s important to remember that we’re giving up some data to online tools like ChatGPT, and that means it’s probably wise not to share sensitive information or photos.
Third, use these tools to improve your work not to do your work. So today I’d say AI can be helpful if we’re looking for a light assist. You could ask a chat bot to rewrite a paragraph in the active voice. A non-native English speaker could ask ChatGPT to remove grammatical errors from an email before sending it. A student could ask the bot for suggestions on how to make an essay more persuasive, but you need a human in the loop to make sure the outputs are correct, not to mention that they’re in context, that they’re appropriate, and that they’re sensitive.
This is all fascinating as I’m listening to it, and something that comes to mind is why don’t we put this to the test right now. And in fact, for our audience, we’re going to introduce another special guest here that Jeff is going to ask some questions to. And that special guest is ChatGPT.
ChatGPT: Hello, Jeff and Oscar.
Jeff: Okay, let’s do this. ChatGPT, what is 153 times 257.
ChatGPT: The answer is 30,061.
Jeff: So pretty basic question, basic work for the kind of calculator people have had in their pockets for years and years. But for any of you lightning math wizards, actually you know the answer is not 39,061. It’s 39,321. Pretty amazing that an app that can trick me into thinking it’s a person can’t do basic math.
So, let’s ask another. Hey ChatGPT, no offense, but why are you bad at math?
ChatGPT: As an AI language model, I’m not bad at math but I don’t have the capability to perform calculations as humans do. My training data includes vast amounts of text about mathematical concepts and equations, so I can answer questions about math, provide explanations and generate mathematical text, but I cannot perform arithmetic operations or solve mathematical problems.
Jeff: So, ironically, what we’re learning here is ChatGPT actually “knows” that it’s bad at math, but it didn’t tell me that – it gave me a wrong answer with confidence.
I’ll elaborate a little bit more on ChatGPT’s answer to why this isn’t working, why it can’t do what seems like a basic function for a calculator.
The idea is that the multiplication process involves carrying over digits and performing complex operations on large numbers. That’s beyond the capabilities of this underlying technology because it’s in the deterministic domain, meaning there’s a definitive right answer. Multiplication is a simple rule, but text-based generative AI cannot comprehend deterministic rules.
Neural networks cannot learn rules, simply because they are not designed to do so. They’re multiplying and adding up inputs, weights, and biases, but they’re not designed to learn simple, or we would think simple “If-Else” logic.
Oscar: So, as we’ve been talking about ChatGPT and artificial intelligence more broadly, you’ve referred to it as a way to perhaps help humans with tasks that they’re performing. Let’s take that a step further, now let’s talk about robots that are using artificial intelligence. We’re certainly seeing more robots in the workplace and production lines; how might robots impact the labor market? Is there a risk that they are making jobs that humans perform obsolete, or is it actually they’re helping?
Jeff: So, I’m a history major, I’m a history buff. The first industrial robot was actually rolled out by General Motors in 1959.5 The robot, called The Unimate, was a hydraulic arm that performed repetitive tasks such as welding. Seven years later, the Unimate, a friendly robot, entered pop culture when it appeared on The Tonight Show with Johnny Carson to play golf.
But that raises the question, what’s happened from now to then? Well, what’s happened is that the average cost of an industrial robot has fallen by 50{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} over the past 30 years, increasing the affordability of adoption.6
So, robots are largely economically viable today in a way that they hadn’t been historically, but country to popular concerns, they generally haven’t been used to replace human factory workers. Instead, they’ve been used to free up those workers to focus on higher-value tasks or supercharge the workers’ output by essentially engaging together.
As a result, robots represent one of the most powerful tools in fighting a lot of today’s economic challenges. Open jobs in US actually exceed unemployed workers by about 5 million people.7 Annual wage growth is 6{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca},8 supply chain pressures are a standard deviation above historical averages,9 a record number of jobs may be near or re-shored,10 and the US and global populations are aging really rapidly.
Let’s look at some of the most powerful robotics use cases and why economic conditions and innovation together are why the market estimates a 23{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} compound annualized growth rate for robotics through 2030.11
And we’ll look at this through the lens of logistics and the lens of healthcare. Let’s start with logistics.
80{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} of warehouse operators agree that they’re going to rely more on automation in the future.12 Look forward in 2023 to seeing Autonomous Mobile Robots – AMRs – proliferate. AMRs, unlike their more common cousins, Autonomous Guided Vehicles, or AGVs, don’t require predefined paths or human oversight. AMRs are particularly helpful in healthcare settings, hospital settings. Cleaning robots are leveraging UV light to sanitize surfaces while navigating crowded spaces with large physical equipment and multiple moving pieces. Robots are already reducing warehouse error rates by up to 67{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca}.13 And robots help address warehouse injuries.
So today, over 200,000 non-fatal accidents happen every year in US warehouses.14 That causes companies a cost of over $84m a week.15 And most importantly, it causes pain and suffering for employees. So, we can solve a lot of those problems with modern robotics, and again, focusing on the ability to help the individual human and make our lives better, we come to healthcare, surgical robots are game changers.
A surgical robot will have one arm with a camera, giving the surgeon a view of an affected area at up to 10 times magnification versus the human eye.16 A second robotic arm will hold surgical instruments, allowing a degree of precision and steadiness that simply isn’t possible with human hands. The smaller resulting incisions improve outcomes and shorten recovery times, reducing pain, reducing blood loss, and reducing scarring.
Today, globally, only about 3{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} of surgeries are performed robotically.17 And you can imagine how much tremendous runway that allows for growth. But again, as we were talking about with ChatGPT, no one’s pulling the surgeon out of the equation. The surgeon needs to be involved. The surgeon needs to be monitoring, the surgeon needs to be operating the robot. This is really about how AI, robotics, and human beings can work together.
As technology is becoming more powerful, efficient, and cheaper, robots are actually more ubiquitous. North America robot sales have reached record highs over three straight quarters.18 We’ve also seen a range of robots enter our day-to-day lives. They vacuum floors, they mow lawns, they craft the perfect cup of coffee, and in some places, they even provide companionship.
Oscar: You’re certainly painting a picture where robots are helping the labor markets, not replacing humans, but I think you used the word “supercharge” what the humans can do in their day-to-day. Maybe we can talk a little bit about agricultural technology and where you see robots and AI playing a role there. Because I think as of last count, there’s about 8 billion people on planet Earth, or we’re approaching that number by 2050. There is projected to be 10 billion people on planet Earth that we need to feed.19 So how do robots help the agricultural processes that we have in place now?
Jeff: So, Thomas Malthus, a philosopher, famously proclaimed hundreds of years ago that we would run out of food because the population of the world would just continue increasing. And there was only so much food that could be created.
What we’ve actually learned in the hundreds of years since is through better technology, better cultivation, genetically modified crops, we can actually increase the total food yield. Even as population rises, AI and robotics are going to be the next game changer in that progression. As we look to the future, disruptive technologies such as precision agriculture and agricultural robots can affect meaningful change towards more sustainable production and consumption.
AgTech makes it possible to produce more food on less land with less water in more locations. The resulting efficiency gains and lower food production costs are really game-changing, and we can dig into a few specific innovations that you’re going to see bring together a lot of what we were already talking about in AI and in robotics.
We’ll start with smart crop monitoring. This is really about AI. Here we’re using real-time sensors and imaging to deliver water and nutrients within a farm system to where they’re most needed, helping produce more food with less, and that enables increased yield through more efficient use of natural resources and inputs, and improved land and environmental management.
So, agricultural producers are already turning to this technology to drive growth. The global market for precision agriculture reached 7 billion in 2021.20 By 2030, it could almost triple to over 21 billion.21 That’s a 13{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} compound annual growth rate.22 Second, area controlled environmental agriculture, often referred to by the initialism CEA, takes place in enclosed growing structures like greenhouses and warehouses, which provide protection from outdoor elements and maintain optimal growing conditions. The term vertical farming is really evocative here – literally, stacking layers of agricultural production atop one another. CEA makes it possible to produce food year-round in urban environments and much closer to consumers.
So, imagine walking into your local grocery store on a snowy winter morning but picking up freshly harvested lettuce, juicy tomatoes, and sweet blackberries, all of which were harvested in your home city just a few hours prior. Vertical farming makes this possible because it allows farmers to grow crops indoors at scale in stacked layers all year round.
Produce grown in indoor vertical farms travels only about 43 miles on average to reach grocery store shelves, compared to 2000 miles for conventionally cultivated product.23
So, bringing together our topics today, robots are also on the rise in farming drones overhead that monitor crop health, robots in field automatically watering plants, and autonomous tractors driving themselves all working 24X7 around the clock.
While adoption is still nascent, commercial agricultural robots are helping forward-thinking farmers and agribusiness to produce more food on less land, and with fewer chemical.
Oscar: It sounds like a very futuristic world that you talk about, but that’s the point I think, is that the technological innovation that’s going on is required.
On the topic of technology, maybe we can talk about the CHIPS Act that was passed last year, and maybe you can describe, what did that piece of legislation actually mean?
Jeff: It’s critical for any discussion of technology in 2023 to note that de-globalization and geopolitical competition are really important considerations that are impacting the sector. When you think about the Chips and Science Act and its $53 billion to support American Semiconductor capacity, “[it’ll] lead to the geographical realignment of manufacturing capacity away from Asia,” and that’s according to Tony Kim, our leading technology portfolio manager here in the fundamental active equity business at BlackRock, who oversees 30 billion in tech assets.
Here and in other sensitive technologies like batteries, quantum computing and artificial intelligence, government intervention through carrots – think fiscal spending – but also sticks – tariffs and regulation are likely to continue. That’s going to create opportunities for investors. It’s also going to create challenges.
Oscar: When you hear about all the technological innovation going on, a lot of which you just spelled out, people draw the comparison to the change that went on during the 19th century industrial revolution. There was a lot of fears at the time and misconceptions, which also then carry through to today – people fearful of all this change that you’re talking about going forward. So, how could these fears and misconceptions challenge progress, and what should investors be thinking about?
Jeff: I think President Kennedy (sic) actually said it best, when looking at this sort of progress, “the only thing we have to fear is fear itself.”
Personally, I think the future is bright. I think the rise of robotics and AI will be a milestone. So, stay tuned, stay positive, because if we embrace change in progress, our descendants, I believe, will look at today’s robotics and AI revolution as improving the world at a scale even beyond that of the Industrial Revolution itself.
Oscar: Jeff, I’ve said this before, when you’ve been a guest, you’ve given us a lot of great facts to consider, and I want to thank you for joining us on the bid, and actually we had two guests today. I want to thank ChatGPT as well.
ChatGPT: You are welcome.
Jeff: Thanks for having me, Oscar.
Oscar: Thanks for listening to this episode of The Bid.
Sources:
1 Yahoo News, Steve Mollman, 12/9/22
2 Sciencefocus.com, “ChatGPT: Everything you need to know about OpenAI’s GPT-3 tool”
3 Shopify, Digital Ad Spend 2021-2026
4 5U.S. Chamber of Commerce, Understanding America’s Labor Shortage
5 Automate.org, “The First Industrial Robot”
6 McKinsey, “Automation, Robotics and the Factory of the Future”
7 Bureau of Labor Statistics, August 2022
8 Atlanta Fed Wage Growth Tracker
9 NY Fed, Global Supply Chain Pressure Index
10 WSJ – U.S. Companies on Pace to Bring Home Record Number of Overseas Jobs, 8/23/22
11 Market Research Future, “Global Robotics Market Research Report,” August 2021.
12 FetchRobotics, July 2022
13 Robots 247, 8/14/21
14 Bureau of Labor Statistics, November 2021 “Employer Related Workplace Injuries and Illnesses”
15 Liberty Mutual Insurance Workplace Safety Index
16 Proliance Puyallup Surgeons, “How Robotic Surgery Improves Patient Outcomes”, 3/11/21
17 Medtronic Press Release, 10/11/21
18 TheRobotReport.com, “Robot Sales Hit Record High for 3rd Straight Quarter”. 8/29/22
19 United Nations, “World population to reach 8 billion this year, as growth rate slows,” July 2022.
20 Grand View Research, “Precision Farming Market Size, Share & Trends Analysis Report By Offering (Hardware, Software, Services), By Application (Yield Monitoring, Weather Tracking, Field Mapping, Crop Scouting), By Region, And Segment Forecasts, 2022 – 2030,” 2022.
21 EIT Food, “Is vertical farming really sustainable?,” August 2018.
22 The White House, “FACT SHEET: CHIPS and Science Act Will Lower Costs, Create Jobs, Strengthen Supply Chains, and Counter China,” August 2022.
This post originally appeared on the iShares Market Insights.
Editor’s Note: The summary bullets for this article were chosen by Seeking Alpha editors.
