Two Potential Market Outcomes
Complementary Infrastructure Requirement
Benefits of Self-Driving Cars
Costs of Self-Driving Cars
Winners
Losers
System Evolution
Driverless (autonomous) vehicles is one of the hottest topics being discussed in the news lately. Some writers have been touting the enormous benefits adoption of driverless cars will bring, emphasizing the utopian scenario associated with the new technology. Others have noted the large industries dislocations their adoption will create, emphasizing the dystopian scenario. This analysis is my attempt to better understand what the market for driverless cars will entail.
Two Potential Market Outcomes
There have been two general market scenarios bandied about in discussions of autonomous vehicles:
- Personal Self-Driving Cars (PSDC): In this scenario people generally own their own vehicles, but instead of people doing the driving, the vehicles drive themselves. This market outcome would yield a vehicle environment that looks relatively similar to the one that exists today, except that cars would have no drivers.
- Shared Self-Driving Cars (SSDC): In this scenario people don’t own their own vehicles. Instead, third-party providers of transportation services own fleets of driverless vehicles, which people hail when they need to go somewhere. In other words, the SSDC scenario conflates autonomous vehicle with peer-to-peer (or sharing) technologies. This market outcome would yield a vehicle environment that is radically different from the one that exists today.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
The degree to which US jobs in the future will be replaced by automation is a question that has been debated for decades. More recently, perhaps during the past decade, it seems that more people believe that a larger portion of jobs will be lost to automation, particularly unskilled jobs. And in the last few years, especially with the advent of Google’s driverless car, it seems that everyone is talking about the potential for a (nearly) jobless future for all but the most skilled, due the imminent automation of a larger and larger portion of jobs.
This series of blogposts is my attempt to better understand the extent to which future jobs in the US will be complemented by or substituted for technology and automation.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
Any attempt to predict what the future might look like should start with an examination of both the present — where are we now? — and the past — how did we get here?
The current global economic environment is unique. A priori, this would tend to suggest that the This Time Is Different view is more likely to prevail. On the other hand, when one examines the specifics of any point is time, one must necessarily conclude that every point is time is different from every other point in time, to varying degrees. In that case, then, this time is no different from any other time. So a priori reasoning doesn’t help.
The circumstances that characterize the current global economic environment include: (i) the increasingly global nature of the world’s economy, (ii) the global recession in which the world has been mired since 2008, (iii) the increasing connectivity of the world’s economies (this goes with increasing degree of globalization), and (iv) the transitional state in which the world’s economies have been engaged over the past few decades. I discuss each of these characteristics below.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
In this blogpost I consider how the nature of jobs in the US has changed over time with the introduction of and adaptation to new technologies. As new technologies have been introduced into the marketplace, they have had three different kinds of impacts on the job market. New technologies have:
• Eliminated many jobs by substituting technology for human labor,
• Created new jobs by creating demand for new types of human labor, and
• Changed the nature of many jobs by complementing human labor.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
Clearly, automation can perform many tasks more efficiently than people. What’s more, technologists continue to improve upon existing technologies that employ artificial intelligence, robots, and other forms of automation. As such, many jobs that are currently performed by humans will eventually be automated. Yet, there are still many tasks that automation will be unable to perform any time soon. This section attempts to clearly define tasks and jobs that automation can and cannot perform.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
The purpose of this series of analyses has been to better understand the extent to which future jobs in the US will be complemented by or substituted for technology and automation.
Summary of Analysis So Far
In Part 1, I established that there are two perspectives on the impact of new technologies on future labor markets:
• The “This Time Is No Different” view suggests that based on what has always happened in the past, we can expect new technologies in the future to eventually create more jobs than they eliminate.
• The “This Time Is Different” view suggests that further improvements in technology will further exacerbate the current trend toward a bifurcation of society, eventually resulting in an unskilled lower class and a skilled upper class.
In Part 2, I established that the circumstances that characterize the current global economic environment are unique. To briefly summarize, the current environment is global in nature, currently in the midst of a severe recession as well as a long-term state of transition as society adapts to new technologies, and actions undertaken by players (businesses, policymakers, etc.) will have an impact on other players globally. Since the current economic environment is different from those that existed in the past, any generalizations that attempt to project past experiences into the future must be carefully considered.
In Part 3, I compared jobs counts by detailed occupation categories in 1940 to those in 2013, and I established that new technologies have:
• Eliminated many jobs by substituting technology for human labor,
• Created new jobs by creating demand for new types of human labor, and
• Changed the nature of many jobs by complementing human labor.
In Part 4, I established that computers can perform any cognitive or manual tasks that can be clearly defined and specified by a set of rules. On the other hand, computers cannot perform tasks for which data inputs and/or outputs are not well-defined or cannot be fully pre-specified. Such tasks involve (i) solving unstructured problems, (ii) working with new Information, or (iii) performing non-routine manual tasks. Alternatively described, computers do not possess the ability to (i) be creative or form new ideas, (ii) recognize large-frame patterns, (iii) engage in or with complex communications, or (iv) perform tasks requiring sensorimotor skills.
The Future of Jobs
I can use Part 3 job comparisons, together with Part 4 restrictions on what automation can and cannot do, to provide some indication as to how new technologies might affect the future of jobs. However, Part 2 suggests that since the current environment is unique, I must be careful about using past experiences to predict the future. To the extent that the predictions about the future gleaned from Parts 3 and 4 suggest This Time Is No Different, do the forces furthering this conclusion transcend the forces causing the current environment to be unique with respect to past environments?
As computers become faster, cheaper, and more efficient, they're increasingly being used in place of labor to generate products and services. This creates a dilemma for society: If a large portion of the population becomes unemployed and is unable to earn an income due to increasing use of automation, then who’s going to buy all the goods and services generated by producers? In other words, having efficient producers doesn't do any good if there are no consumers who can afford to buy their output.
A universal basic income is a system in which all citizens of a country are paid an unconditional annual income. Can a universal basic income that is funded from taxes on workers and on producer profits be used to solve the problem of joblessness due to automation?
This analysis seeks to answer the following question: Given a society with a large portion of jobs replaced by automation and an associated large portion of its citizens with no employment prospects, would a universal basic income system ever be sustainable from an incentives standpoint? That is, would all members of society ever find it in their mutual self-interest to support a UBI?
Terminology/Technical Information
Players in the Electric Vehicle Game
Current Stages of Adoption of Electric Vehicles
Advantages and Disadvantages of Electric Vehicles
Energy Inputs and Emissions Costs of Electric Vehicles
Should the Construction of Electric Charging Stations be Subsidized by the Public?
A recent article in the WSJ, “U.S. Utilities Push the Electric Car” by Cassandra Sweet, notes that electric companies nationwide are seeking to charge electricity consumers extra fees to fund construction of electric vehicle charging stations by the electric companies. The rationale is that having more charging stations available will speed adoption of electric vehicles by consumers, thereby leading to fewer pollutant emissions, and thus higher air quality for everyone.
Should all electricity consumers be required to pay the construction costs of electric vehicle charging stations?
The answer to this question requires understanding the underlying distribution of the private and social costs and benefits associated with manufacture and use of conventional versus electric vehicles.
Category Gains and Losses
Subcategory Gains and Losses
New Job Categories
Conclusions
By 2000 much outsourcing and automation had already taken place. So a comparison of how private (i.e., non-government) jobs have changed between 2000 and 2015 should provide a better sense of where things are headed in the future than would a comparison of jobs in 2015 with those from an earlier time.
The BLS provides job counts for the US by year and by private (i.e., non-government) Occupational Code, which sorts jobs into “major,” “minor” and “broad” categories. After perusing the data and considering the classifications provided, I decided to create my own job categories and subcategories that I thought were more meaningful to the way I’d like to consider the information. From here on, when I refer to categories and subcategories, I’m referring to the categories I created, not those used by the BLS.
One important caveat to the analysis: The analysis includes only those jobs captured by the BLS. There is reason to believe that there are a nontrivial number of private jobs that are not being captured in the BLS data. For example:
- According to one source, as of Oct 2015, Uber had “327,000 active drivers on the road in the U.S.” Considering that the BLS reports 180,960 Taxi Drivers and Chauffeurs in 2015, it is clear that the BLS is not capturing Uber drivers. And there are many other gig jobs currently in the economy, most of which are probably also not being captured. See, for example, “Top 100 On Demand Jobs Like Uber And Sites Like Airbnb.”
- There are a lot of small producers (see, for example, Jeffrey Sparshott, “Big Growth in Tiny Businesses” in the economy, many of whom are likely not captured by the BLS.
The Technology Triangle
Years ago I attended a meeting on intellectual property (IP). One of the speakers, a sharp IP attorney named Pat Ellison, gave a talk, which greatly resonated with me. He said that a successful technology requires a balance between technology, business, and law, as represented by the triangle in Figure 1. (I recently contacted Pat about the origin of this idea and he said he was fairly sure that the idea was developed collaboratively with others, but he couldn’t remember who the other contributors were.) Very succinctly, descriptions for the requirements are:
- Technology: The technology must work well.
- Business: The technology must be cost effective, that is, is must able to be manufactured and sold for a profit.
- Law: The legal and regulatory underpinnings of the technology, including intellectual property foundations and liability issues, must be sound.
A successful technology will exhibit balance in each of the three areas in the sense that if any of the three is too weak – the technology doesn’t function well, the technology cannot be sold for a profit, and/or the intellectual property is invalid or ineffective or other regulatory issues have not been settled – then the technology will not become commercially successful.
Figure 1
Persuasion technologies include methods and techniques derived from behavioral psychology and behavioral economics used to shape the choices people make. The favorable environment for using such methods, enabled by people’s increasing use of computers and smartphones, has led to the proliferation of their use by software developers.
Like any technology, persuasion technologies can be used for good or evil. However, the increasing dependence of people on digital technologies, together with the increasing prevalence of software developers’ use of persuasion technologies has created emergent behavior in society that’s downright ugly: the emergence of extremism, outrage, and divisiveness among members of society.
This analysis is closely tied to a previous analysis I performed, “Information Distortions on the Internet.”
This analysis will examine
- The nature of persuasive technologies
- The game between software developers and users that has created an environment of good, bad, and ugly
- How the environment might be changed to create more favorable social outcomes
AI Basics
Definition
Why Now?
The Controversy
The Letter
Current AI Ecosystem
Categorization of AI Technologies
Organization of Companies in the AI Ecosystem
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
OpenAI, the organization recently cofounded by Elon Musk, has been receiving a lot of press lately. The company was introduced as follows:
OpenAI is a non-profit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return.
Since our research is free from financial obligations, we can better focus on a positive human impact. We believe AI should be an extension of individual human wills and, in the spirit of liberty, as broadly and evenly distributed as possible.
Two issues in particular have been generating most of the attention surrounding the founding of the new organization:
- OpenAI will focus its research on discoveries that will have positive benefits for society; and
- OpenAI will be open source, that is, its discoveries will be freely available to all.
Recent advancements in AI have enabled researchers to provide valuable new products and services in the marketplace, and the promise of continuing advancements suggest that even more valuable discoveries are on the horizon. As such, what motivations lay behind the decision of Elon Musk and his cofounders to make their new organization open source, rather than establishing it as a for-profit company? They have said that their intent is to provide discoveries that benefit humanity. But are the founders really as altruistic as they, themselves, and the media have made them out to be?
This analysis is an attempt to better understand the dynamics underlying the AI ecosystem so as to better understand what motivated the founders of OpenAI to designate the organization as open source and whether or not there may be other agendas out there besides pure altruism.
Generating Value from AI Systems
Essential Components
Feedback Loops
Stated Benefits of Open Source Systems
Focus on Projects that Benefit Humanity
Mitigate Power of Single Entity
Benefit From and Improve the Technology
Attract Elite Researchers
Why Do I Think OpenAI Was Established As Open Source?
The More Obvious/Discussed Justifications
The Less Obvious/Discussed Justification
In Part 1 of this analysis, I provided some background information on AI as a foundation for the discussion. In this part of the analysis I continue on to discuss why I think Elon Musk designated OpenAI as an open source entity.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
Generating Value from AI Systems
Essential Components
I’ve mentioned several times throughout the analysis that AI technology involves three essential components: IA algorithms (software), AI platforms (hardware), and big data. In this section I describe the nature and use of these components in more detail.
I like the way Neil Lawrence describes the AI system in “OpenAI won't benefit humanity without data- sharing.” He uses the analogy of cooking, where AI algorithms are the recipes, the data are the ingredients, and the platform is the stove or oven.
Anyone who has tried to come up with an original recipe will tell you that it generally needs to be tweaked before you come out with the ideal output. Similarly, researchers design AI algorithms, test and train them by running data through them, then tweak them to improve their performance.
Generally speaking, the better cooks are those with more experience, and they tend to be the ones who come up with the best recipes. Of course, occasionally unknown or unpracticed chefs come up with excellent recipes, but that’s not the norm. Similarly in AI, the better, more experienced researchers are the ones who will probably generate most of the advancements in AI. However, that does not preclude the possibility that some unknown savants will be able to come up with advanced solutions on their own.
Also, in cooking, better ingredients produce better dishes. Similarly, in AI, higher quality data lead to better results – as the saying goes, garbage in, garbage out. At the same time, AI algorithms become more accurate (trained) as they run more data. This means that having access to larger volumes of data will generate more accurate algorithms. So when it comes to data, both volume and quality are important.
Finally, when cooking, the sizes of the ovens constrain the volume of food that can be produced. Similarly, with AI algorithms that need to run through large volumes of data to become properly trained, larger, more efficient hardware systems produce results much more quickly than do smaller systems.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
The following are the essential factors at issue when considering batteries for use in powering electric vehicles:
Amount of Energy that Can Be Stored
The batteries of any given size that are able to store the greatest amount of energy in terms of both weight (specific energy) and volume (energy density) of the battery are the most desirable (efficient) to power electric vehicles. Perhaps the largest current disadvantage in terms of the state of battery development for electric vehicles (EVs) is the fact that currently EVs cannot go very far without having to have the battery recharged, creating so-called range anxiety. Lower battery range would be less of a problem if (i) there were more fueling stations around (currently there are very few refueling stations), and/or (ii) it didn’t takes so long to recharge the battery (20 minutes to several hours, depending upon the technology of the charger). Currently, EV manufacturers are working fiercely to increase both the specific energy and/or energy density of batteries for EVs so as to achieve greater vehicle range.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
From the beginning, the biggest problem facing all-electric vehicles has been their short range, that is, they cannot go very far without having to recharge their batteries. Since lithium-ion (Li-ion) batteries offer the greatest energy capacity and density of all the batteries, and thus the greatest potential for longer range, Tesla chose to use Li-ion batteries to power its first all-electric vehicle, the Tesla Roadster. As Tesla notes:
Tesla battery packs have the highest energy density in the industry
...
Nickel Metal Hydride (NiMH) batteries are commonly used in hybrid cars. However, a 56 kWh NiMH battery pack would weigh over twice as much as the Roadster battery. Instead, Tesla uses Li-ion battery cells which dramatically decrease the weight of the Roadster pack and improve acceleration, handling, and range.
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
Now let’s take a look at the characteristics of the current offerings of electric vehicles across manufacturers, which are presented (above in Figure 2 and) in Figures 4 and 5.
Figure 4
Figure 5
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
This section examines the structure of costs associated with manufacturing Li-ion batteries for use in electric vehicles.
The battery packs used in electric vehicles consist of numerous individual batteries connected together and packaged into modules, which are then connected together and packaged into battery packs. David L. Anderson, in “An Evaluation of Current and Future Costs for Lithium-ion Batteries for Use in Electrified Vehicle Powertrains” explains this process in a bit more detail:
[F]or automotive applications, individual cells are typically connected together in various configurations and packaged with associated control and safety circuitry to form a battery module. Multiple modules are then combined with additional control circuitry, a thermal management system, and power electronics to create the complete battery pack…
A copy of the full analysis can be downloaded by clicking on the link at the bottom of this blog entry.
In Part 1 we learned that the essential factors at issue when considering batteries for use in powering electric vehicles include (i) the amount of energy that can be stored, (ii) longevity, (iii) cost, and (iv) safety.
In Part 2we learned that (i) theearliest EVs (hybrids) used NiMH batteries, due to their greater safety, longer life, and lower cost; and (ii) two factors led to the industry-wide adoption of Li-ion batteries as the battery family of choice for electric vehicles: (a) their potential for greater vehicle range, and (b) patent access problems to NiMH battery technology.
In Part 3 we learned that (i) current EVs use Li-ion batteries because they offer the greatest potential energy capacity and density; (ii) Li-ion batteries include a family of batteries composed of different materials; (iii) the cost of the battery is the largest cost component of electric vehicles; of the battery costs, the most significant contributors are the costs of the raw materials, which vary greatly in price; and (iv) different material constructions of Li-ion batteries generate differences in battery performance, where the ranking of battery potential from least to greatest is (a) LCO (1st gen) and LMO (2nd gen), (b) LFP (3rd gen) and NMC (4th gen), and (c) NCA and LTO.
In Part 4 we learned that information on current EV offerings provide three indications: (i) many of the current EV offerings are “compliance cars”; (ii) the performance of most EVs is clustered around similar levels of energy capacity and range; and (iii) the battery manufacturing industry is consolidating around a few key suppliers.
In Part 5 we learned that (i) high quality control standards for the manufacture of batteries for EVs result in low manufacturing yields, on the order of about 60%; (ii) materials account for about 75% of total manufacturing costs of batteries for EVs; and (iii) cost reductions in the manufacture of lithium-ion batteries may be achieved through larger scale production volumes and technological breakthroughs.
Putting it all together yields the following insights.
The future of jobs is a serious concern.
The most popular opinion I’ve seen is that the answer is more education. Consider, though, that we are in a period in which historically high levels of the population have some amount of postsecondary education. Yet, less than half the population (about 42%) has at least an associate’s degree, and only about a third of the population has at least a bachelor’s degree. How much higher is it realistic to think these levels can actually go? Not to mention that student debt has reached massively unsustainable levels ($1.3 trillion).
So then what about the other half of the population?
I recently did an analysis of changes in employment by firm size over the past few decades. My analysis showed that
- Increasing percentages of employees have been employed in large firms, at the expense of employees in small firms.
- New firm creation has increasingly come from openings of smaller firms, while consolidation has been rampant among the largest firms.
- The economy has become less dynamic than it was during the 1990s, with relatively greater decreases in job activity at smaller firms.
Taken overall, the data are consistent with economic/market conditions that
- Are less hospitable to firms overall, and
- Favor small firms for new innovations, but large firms for continued market success.
Factors consistent with this environment include
- More regulations, capture by special interests, and/or uncertainty over-all that inhibit business activity;
- Regulations, capture by special interests, and/or uncertainty that favor large firms over small firms (e.g., Obamacare, bank regulations that favor large and/or less risky loans over small/more risky loans, minimum wage laws, etc.);
- Bureaucracy in larger firms that prevents new ideas from developing and/or gaining traction; and
