Engineered

Review of "Turning Point: Policymaking in the Era of Artificial Intelligence"

I recently read this book, by Darrell West and John Allen, respectively a Vice President and the President of the Brookings Institution. It was a little dry and numbers-heavy but there were some interesting chapters. I particularly liked their comments on the trends re: car insurance (basically they argue that with the advent of driverless cars, car insurance should go down [p.72], although I think that insurance companies will make up for it in the home insurance premiums), and also I was fascinated by the discussion about the zoning rules and congestion p.96. It is a really great point that vehicles like Amazon Prime delivery trucks put more wear and tear on the roads and they also bring with them a whole range of issues from double-parking to storage issues to parcel theft to apartment buildings that have no loading dock. I enjoyed thinking about the environmental impact of truck delivery as well as the joint warehouse and distribution optimization problems to meet customer needs. Also I had never heard about the freight master plan in Seattle but it is indeed a thing! You can read more about it here. Rating: three stars out of five, which on Goodreads scale means I liked it (enough).

It is time to revive this blog! Let's start with a book I read last week, Tim Cook: The Genius Who Took Apple to the Next Level by Leander Kahney. This is to prepare for the engineering management course I will be teaching next semester, which is a required course of all engineering undergraduates. This semester I had assigned Failure is not an Option by Gene Kranz (former NASA flight director) and I loved it, being an engineering MIT nerd, but some of the students didn't like it as much because they had a hard time seeing how it would be applicable to them (and also they didn't understand why they had to take this course since they don't expect to be managers for a while, and of course since my specialty is optimization and I've never been a manager in industry I was not the best person to make the case they should care). So for the coming iteration of engineering management I am reading a couple of other books to find something more suitable. I wanted something available in paperback for cost reasons, not too long because it is only one element of one course (so I was wary of the Steve Jobs bio by Walter Isaacson, for instance) and hopefully something with a bit of diversity because some of the students who are not straight while males were not enthusiastic about the lack of diversity in the Kranz book (what can I say, in the 1960s you just didn't have many engineering managers who were not straight white males and even fewer who achieved enough to be the topic of a book).

I loved the Tim Cook book for many reasons, which I will try to dump in this post. First, I think Tim Cook is a much better leader than Steve Jobs. Jobs was abrasive and, in my opinion, an egomaniac a little too confident in his genius. Maybe that worked twenty years ago, but those days people won't follow too many leaders with such characteristics. Also, Jobs didn't care about sustainability, and Cook has created a lot of programs to achieve sustainability and climate change. Again, twenty years ago maybe CEOs could get away with not caring about sustainability, but with all the talk about climate change, it is not something visionary leaders can do any longer. Cook has also pioneered new products such as the Apple Watch. I also, of course, like that Cook was educated as an industrial engineer and started his career in operations management. There is also an interesting chapter in the book about the controversy surrounding the FBI's request to Apple to develop a backdoor to the iPhone, which raises valuable questions about data privacy - a hot topic these days. Further, Cook is openly gay following his coming out in an op-ed he wrote for Bloomberg Businessweek in 2014, which helps show that CEO positions aren't just for white males, and he has made a strong push toward diversity, for instance on Apple's board. (Of course, challenges remain: while the percentage of underrepresented minorities working at Apple has increased, they tend to be mostly employed in the Apple stores rather than in higher-paying positions as managers.)

Another thing I like is that Harvard Business Publishing Education has many case studies about the events described in the book. This past semester I used the case studies that the previous instructor had been using for years, but the students pointed out that they were too dated (and they were). So I am thinking about also assigning case studies for instance on sustainability, the Apple Watch, etc. Then I would probably also assign some non-Apple cases on related topics so that the entire semester is not about Apple.

Here are a few of the relevant cases on the Harvard Business Publishing Education website:

1. Apple Inc: Managing a Global Supply Chain
2. Apple Inc: Global Supply Chain Management
3. Apple and its Suppliers: Corporate Social Responsibility
4. Apple Pay
5. Apple Watch: Managing Innovation Resistance
6. Financial Policy at Apple (more likely to end up in the engineering finance course)
7. Building a Backdoor to the iPhone: an Ethical Dilemma
8. Apple in China
9. Apple's Future: Apple Watch, Apple TV and/or Apple Car?

Interesting cases would also be:

1. China Moves Up the Value Chain: FoxConn's Dilemma
2. Toyota: The Accelerator Crisis (from 2010)
3. Cisco: Collaborating on New Product Introduction (from 2009 but relevant to new product introduction in engineering management)
4. Lean Process Improvement at Cleveland Clinic (also a little older)
5. Wyeth Pharmaceuticals: Spurring Scientific Creativity with Metrics
6. GlaxoSmithKline: Reorganizing Drug Discovery (from 2005)
7. IDEO: Human-Centered Service Design
8. Vertex Pharmaceuticals: R&D Portfolio Management
9. Ant Financial
10. Implementing LEAN operations at Caesar's Casinos
11. Disruption in Detroit: Ford, Silicon Valley and Beyond
12. Zara: An Integrated Store and Online Model

I only plan to use 12 cases since there are only 13 weeks of the semester after the first introductory week and who knows if we will have a mega-snowstorm again after Texas's continued failure to winterize the electricity grid, so it is best to have one week's worth of margin. This semester I was covering the cases on Thursdays and on Tuesdays we would talk about the assigned book (Failure is not an option this semester) and about another engineering management textbook I liked (I may also change that textbook but that will be the topic of another post.)

I probably should give some thought as to how to keep the course engaging although the sections of the discussion-heavy course are getting stuffed with as many students as there is a classroom of that size, but since leadership seems opposed to opening another section to maintain the quality of the student experience that is a hallmark of education at a private university, I'll take it as a signal that I shouldn't worry about the inability to do small-group discussions in the classroom the course was dumped in next semester or the surge in enrollment over the past year. It is a pity because what the students seemed to like most about the course was to talk with each other about their opinions of the cases and their experiences as summer interns, but there is nothing I can do about it, except taking the students out to hold discussions on the campus lawn, weather permitting. At least I will have tried to make it interesting for me to teach, and hopefully some students will find the material interesting to learn. 

Online Learning

For today’s post, I want to talk about online learning. Zoom has obviously made the continuity of instruction possible in higher education in spite of the pandemic, and for that I am very grateful. To me, synchronous education is better than asynchronous. But I also want to point out that (through no fault of Zoom’s), turning online synchronous lectures into Internet versions of on-campus ones is an implicit choice that we don’t necessarily need to make when considering online education by itself, rather than as a companion to in-person instruction.

By that I mean that a semester in a brick-and-mortar campus has always had a certain number of weeks, let’s say fourteen weeks, followed by finals. And the lectures have always had a certain duration that worked well for in-person instruction, such as fifty or eighty minutes. But if we consider 100% online instruction, it is not clear to me completion of the course and retention of the material are best achieved when the online course is also fourteen weeks long, followed by finals.

For instance, many online students are full-time professionals. They may be able to set aside four weeks or maybe even eight weeks for a course, but three months may give them pause, because their boss may wonder if they can really give the team their best. I think for online models it makes more sense to have a course split into modules and having one week or two of break in between the modules. Also, the modules could be offered repeatedly throughout the year so if a student wants to take a break from learning to focus on their full-time job, they could be off for a few weeks and then resume easily. You may also have to teach only once a week to make the schedule work for students who also have a full-time job and can’t make the commitment multiple times a week.

Further, I’d love to see more one-morning or one-afternoon online courses: something of very narrow duration and therefore very small cost that lowers the bar for interested would-be students who are not sure of what it would mean to take an eight-week course with a certain professor or at a certain school. Offering day-long (or half-day-long) courses would lower the barrier to entry into an online graduate program by giving students a better sense of what to expect.

In summary, for online education to deliver its potential we should design online courses that are best suited for the medium, and the best of them may be radically different in duration (of a lecture), length (of the module), course organization and testing, as well as size of enrollment, than their in-person counterparts.

I mention size of enrollment because part of the fun of an in-person course is to run into one’s classmates in the classroom before the start of the class and catch up. In the Zoom it is a bit harder to do since students may be reluctant to talk all over each other in trying to greet a specific classmate, it is awkward to have everyone else listen it to the conversation and it may feel invasive to send a stranger private chat messages to continue the discussion.

So from experience I would say ten students with the instructor is ideal to maximize interaction without having to overly duplicate instruction across sections. This means that some courses that had twenty or twenty-five students enrolled in person would be better served by opening several smaller online sections, if the course instructors seek that their expertise transfer well to online education. Online education is certainly more complicated than just putting an off-line course on Zoom. But whatever form online education takes next, Zoom is sure to play a critical role in delivering it and remain part of the future of online learning.

McKinsey on The future of work in America

 McKinsey has been in the headlines for the wrong reasons recently (and more recently) so I'm not sure how much credibility I put in their research reports anymore but there is no doubt that, given the renewed interest in artificial intelligence and its implication for future job prospects in the U.S., many people are wondering what data science and automation will mean for them, both in terms of their own jobs and the economic prosperity of the area where they live - the latter point being a novel focus of the present report, understudied in the rest of the literature. So here is an overview of McKinsey's recent report The future of work in America. 

Of its analysis of over 3,000 US counties and 315 cities, it concludes that job growth will be concentrated in 25 megacities and high-growth hubs (Dallas-Fort Worth being one of them), while 54 cities and 2000 rural counties, home to 1/4 of the US population, have "older and shrinking workforces, higher unemployment and lower educational attainment." The report's authors write: "Automation technologies may widen these disparities at a time when workforce mobility is at historic lows." It will become even more important to hold a Bachelor's degree or higher, which has implications in terms of the job prospects of underrepresented minorities too, since they often have more limited access to education. 

Here are some sobering statistics: "Jobs held by nearly 15 million workers age 18-34 may be automated, so young people will need new career paths to gain an initial foothold in the working world. Roughly 11.5 million workers over age 50 could also be displaced and face the challenge of making late-career moves. The hollowing out of middle-wage work could continue."

According to the authors, the largest occupational categories in the US economy have the highest potential displacement rates, while strong job growth is expected in "healthcare, STEM occupations, creatives and arts management, and business services." The McKinsey consultants recommend mapping alternative career pathways for displaced workers by identifying "growing occupations that utilize compatible skills and require similar education credentials."

My favorite graph in the report is naturally Exhibit E8, which shows Collins County (Dallas/Plano, TX) high in the top right quadrant of high employment growth and high index of educational attainment of the population. McKinsey estimates that educational attainment index for the Dallas area, by the way, much higher as that of Austin. I also found Exhibit E10 very interesting, but most valuable was Exhibit E11, which details the different challenges various types of communities will face, from the urban core to the mixed middle and low-growth and rural areas. I won't post them here so that you get to read the whole report on the McKinsey website, but I'll tell you that I wholeheartedly agree with the consultants' assessment that a challenge of megacities and high-growth hubs will be to increase affordable housing near employment centers. As for the need to "involve employers in creating high school and community college programs to develop key skills needed in growing fields," that is already been done in Dallas thanks to a groundbreaking partnership between SMU, Toyota and the Dallas ISD. (Also read this great article about it in Dallas Morning News, Bridging the economic divide in Dallas is a moral imperative.) The report's authors also raise the issue of retraining workers and providing lifelong learning, which may have to take the form of "short, flexible courses that follow the boot camp model". While successful training programs abound throughout the country, it remains to be seen how they can be scaled. 

Beyond the impact on workers, it is worth also discussing the impact on companies, as "organizations will face cross-cutting challenges that require multidisciplinary solutions." "While some will simply pursue labor cost reductions, those that deploy automation and AI technologies in pursuit of innovation and growth can accelerate profitability growth and even add jobs." The consultants discuss employers' opportunities and challenges based on company footprint and workforce characteristics. 

At the end of the report, I concluded that, although McKinsey has received a lot of negative press, that report does show the type of superb work it is capable to produce in consulting engagements. The top decision makers may have committed errors in judgement in their dealings in Saudi Arabia and South Africa, but McKinsey's foot soldiers - its army of analysts and associates - remain the best in the business. A must-read.

 

Women in Science

Not too long ago, I came across this series of wonderful posters commemorating women in science, available on the website insidetheperimeter.ca, which presents "big ideas from the Perimeter Institute [for Theoretical Physics]". What amazed me is that, in spite of my obvious interest in women in science and engineering documented in my own path, I had never heard of many of the women profiled. Wouldn't talking about the stories and the people behind the discoveries help make science education more interesting for more middle school and high school kids? 

Thankfully, the informative webpage on insidetheperimeter.ca has some details on who these women are and why they deserve to be remembered. The webpage also has the colorful individual posters, which makes it a lot more enjoyable to read than this post, but I will still say a few words on each.

• Claudia Alexander (1959-2015), PhD U.Michigan, geophysicist, worked for NASA's Jet Propulsion Laboratory including as last project manager of Galileo mission to Jupiter 
• Vera Rubin (1928-2016), PhD Cornell, astronomer who did pioneering work on galaxy rotation rates
• Vivienne Malone-Mayes (1932-1995), PhD UT, active in civil rights demonstrations, fifth African-American math PhD in the U.S., first African-American on faculty at Baylor, 
• Ada Lovelace (1815-1852), mathematician, worked on the early "computer", 
• Chien-Shiung Wu (1912-1997), PhD UC Berkeley, "experimental physicist who made significant contributions in the field of nuclear physics. Wu worked on the Manhattan Project, where she helped develop the process for separating uranium into uranium-235 and uranium-238 isotopes by gaseous diffusion" (according to her Wikipedia page)
• Emmy Noether (1882-1935), PhD U.Erlangen, described by Albert Einstein and others as "the most important woman in mathematics" (according to her Wikipedia page) developing the theory of rings, fields and algebras. Was removed from her teaching position by the Nazis, took refuge in the USA where she started teaching first at Bryn Mawr and then at Princeton.
• Annie Jump Cannon (1863-1941), "American astronomer whose cataloging work was instrumental in the development of contemporary stellar classification" (again according to her Wikipedia page).
• and of course two-time Nobel-Prize-winner Marie Sklodowska Curie (1867-1934), who needs no introduction.  

You can read about the Emmy Noether Initiatives at the Perimeter Institute here. 

 

"The doctor of nearly lost causes"

In this days of conoravirus panic and pandemic, I figured it might be nice to have the example of Dr. William Gahl, MIT'72, to cheer us up. I read his profile in the latest issue of Technology Review and that was the most uplifting thing I had read all week. Gahl heads the Undiagnosed Diseases Program at the National Institute of Health, where he "takes on the baffling cases that drug companies won't touch." The article follows Donna Appell as she attempts to find answers for her daughter Ashley, afflicted by the Hermansky-Pudlak syndrome. She worked as a nurse and cold-called NIH after hearing one of the doctors say that her daughter was a "NIH case" and, by sheer luck bordering on a miracle, got patched through Dr. Gahl, who has become one of the world's foremost experts on rare metabolic diseases. 

Dr. Gahl "more or less singlehandedly launched" the Undiagnosed Diseases Program at NIH in 2008 with only $280,000 in seed funding, and has discovered to date 23 new genetic disorders or genetic phenotypes. After listening to Donna (Ashley's mother) on the phone, Dr. Gahl invited Ashley to the NIH, where she became the index case (the first person with that disorder studied by the center's researchers) for her rare genetic disease. The article explains: "That investigation ultimately triggered a sea change in medical understanding of the disease. Over the course of several decades, Gahl and his colleagues identified three of the nine genes now known to cause Hermansky-Pudlak and determined the condition’s natural history, or progression. This research allowed them to help pinpoint the variations that lead to the scarring of lung tissue known as pulmonary fibrosis, which makes it difficult to breathe and is the most common cause of death for those with certain types of Hermansky-Pudlak." What an accomplishment on D. Gahl's part, to have that sort of impact on patients' lives. 

What amazed me about Dr. Gahl is the genuine care he brings to his patients. For instance, Donna Appell started the Hermansky-Pullak Syndrome Board, of which Dr. Gahl is a board member. In 27 years, he has never missed one of the annual conferences. (Ashley is still alive, in case you are wondering.) He is also level-headed about his success, for instance saying: "“There are a lot of people smarter than I am in the world, but working helps me understand things,” he explains. “I can’t lie on the beach. I work Saturday and Sundays. That’s when you get things done."" Of his rise through the ranks at the National Institute of Child Health and National Human Genome Research Institute, he says, jokingly: "My bosses kept quitting." 

A life changer for Dr. Gahl was to become friends with someone "who served as a liaison with the FDA and was a key architect of the Orphan Drug Act of 1983, which gives the pharmaceutical industry incentives to invest in research on drugs for populations too small to promise blockbuster profits." This is how he learnt that getting a diagnosis was the most important thing for patients with rare diseases (so that they could look for patient advocacy group, research protocols elsewhere in the world etc). The article then describes the work he does with patients at the Undiagnosed Diseases Program (UDP) and his great sense of humanity. I also loved learning about UDP's first breakthrough (read the article). 

This was a fascinating read, and the way the article ends - with Dr. Gahl excusing himself to attend the birthday party of one of his colleagues, who brought cake - was just the perfect illustration of his humanism, which shines as brightly as his talent. 

Barry X Ball: 3D scans and robots in art

One thing I like about Dallas is the ability to listen easily to great artists coming through the city to talk about their art and their creative process. While I usually stick to the Arts & Letters Live series of the Dallas Museum of Art (Zadie Smith! Ann Patchett! Paul Auster!), in late January I attended the talk by New York artist Barry X Ball as part of the Nasher 360 Speaker Series at the Nasher Sculpture Center, on the day his retrospective opened at the Nasher. 

What is unique about Ball is the way he uses 21st century tools such as robots and 3D printing to re-imagine masterpieces, such as Sleeping Hermaphrodite on view at the Louvre Museum in Paris. The talk provided fascinating insights into Ball's process - how he scans sculptures using 3D imaging and makes alterations to the 3D scans and then produces a small-scale version of the sculpture he has in mind followed by a reimagined sculpture thanks to an army of assistants and robots carving the stone. Ball also donates the original 3D scans to the museums that own the work of art he has scanned, to help with conservation. The process itself is painstakingly slow because the robotic arm has to go very slowly. As an example of alteration, Ball showed his reimagined version of Pieta by Michelangelo, who died a few days after completing it at the age of 89. Because it was Michelangelo's final work of art, Ball replaced the face of Jesus Christ with the face of Michelangelo. He also removed a piece of wood near Mary behind Christ to have cleaner lines. The Wall Street Journal explains the statue took Ball seven years to produce at a cost of $1m. This brochure explains his process too.   

The Wall Street Journey has a great 6-minute video of Ball's process. Ball has a great new studio (at the tune of almost $20m) that will allow him to create even bigger statues and takes pride in being able to make changes to the sculptures that were not possible at the time the original artists made them. The video also discusses the financial side of art-making and Ball's financial pressures. 

In his talk at the Nasher, Ball also discusses his sculptures of two heads representing Envy and Purity and he had an interesting take about one of the pairs, which was supposed to represent the violence women in some parts of the world are still subject to, especially acid attacks in India. Amusingly enough, he used the word duality a lot to describe those pieces, and it made me think of duality in linear programming every time. Given my work on robust optimization, duality is probably "the" word that I cannot hear in everyday settings without thinking about my professional life.  

Shockingly, the exhibition at the Nasher marks the first survey by a major U.S. museum of Barry X Ball's work in three decades. His work is so good it certainly deserves more attention - some people might find it kitsch or derivative, but I love how he uses modern engineering tools to add his spin on famous statues and make us rethink how we view them - and hopefully this retrospective will be followed by many others. 

GrAIt Expectations: AI in Business

 

I recently came across this special report on AI in Business, entitled "GrAIt Expectations", in an old issue of The Economist. Overall I found the special report thoughtful and well documented, with lots of interesting examples. Some are better known than others, for instance it is no surprise that big companies such as Johnson & Johnson and Accenture are now using Artificial Intelligence (AI) to sort through job applications and pick the best candidates, but AI can also be used in more cutting-edge and perhaps disturbing ways: the Chinese insurance company Ping An thinks it can spot dishonesty by having prospective borrowers answer questions by video.    

I am not sure if I agree with the definition of AI: "AI and machine learning (terms that are often used interchangeably) involve computers crunching vast quantities of data to find patterns and make predictions without being explicitly programmed to do so." I tend to think that artificial intelligence is the ability of computers to make decisions, rather than predictions, without being explicitly programmed to do so. Or we can adopt the definition in the Oxford Dictionary, which defines Artificial Intelligence as "the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages." According to a report by MIT's Sloan Management Review and the Boston Consulting Group, "around 85% of companies think AI will offer a competitive advantage, but only one in 20 is "extensively" employing it today."

This got me interested in reading the MIT/BCG report itself. In fact there are three reports, since this is a yearly effort that started in 2017. The authors give the example of Airbus, which uses AI to match about 70% of the production disruptions to solutions used previously, shortening the amount of time it takes to deal with disruptions by more than a third. Because the report is already about 2 1/2 years old, we are nearing the time where we can check the validity of predictions such as "within the technology, media and telecommunications industry, 72% of respondents expect large effect from AI in five years." They split their survey's respondents into 4 groups: pioneers (19%), investigators (32%) [referring to organizations that are deploying AI at the pilot stage only], experimenters (13%) [organizations that are piloting or adopting AI without deep understanding, which frankly sounds like a bad thing] and passives (36%) [organizations with no adoption or much understanding of AI]. The 2017 report particularly emphasized the need for good data in order to create good AI models. Apparently a lot of companies didn't have the historical data required. Some applications such as drug development need negative as well as positive data (what didn't work as well as what did work), while only the positive data tends to get published. The report's authors make an interesting point about the make-or-buy problem applied to AI capabilities. They say that the leaders in using AI for business develop their own in-house AI capabilities, while the followers tend to work with independent AI vendors. But it is going to be important for organizations to have their own AI in-house capabilities. 

The MIT-BCG report lists several managerial challenges: (1) developing an intuitive understanding of AI, (2) organizing for AI (especially having the organizational flexibility that enables collaboration between humans and machines on project teams, taking into account whether AI resources should be managed centrally or locally or in a hybrid model) and (3) re-thinking the competitive landscape. Among the authors' recommendations (ensure customer trust, perform an AI health check, brace for uncertainty, adopt scenario-based planning, add a workforce focus), I found the scenario-planning one the most intriguing. "Companies need to think even more expansively about their businesses, build cohesive future scenarios, and test the resilience of their directional plans against such scenarios."

In a 2018 update, the MIT-BCG authors identify four major patterns in their survey and interview data:

• Pioneers are deepening their commitments to AI ("fully 88% of Pioneers invested more in AI than in the previous year")
• Pioneers are eager to scale AI throughout their enterprise ("85% [of Pioneers] agree that they have an urgent need for an AI strategy and 90% say that have a strategy in place already.")
• Pioneers prioritize revenue-generating applications over cost-saving ones because "a much higher penalty is incurred by missing an opportunity in the external market" rather than internally. Examples include "complex pricing applications, churn analysis in the customer portfolio, new delivery forms that significantly speed up [the company's] ability to meet customer demand" as well as "wholly different business model[s]."
• AI is creating both hope and fear among workers due to the possibility of workforce reduction.

Many respondents also expect change to their business model. Also, it is important not only to get more data science resources but also train more businesspeople in the potential of AI to think about potential use cases.

The 2019 update started on a less upbeat note: "Many AI initiatives fail. Seven out of 10 companies surveyed report minimal or no impact from AI so far." The authors argues that "companies that capture value from their AI activities exhibit a distinct set of organizational behaviors. They integrate their AI strategies with their overall business strategy, take on large, often risky AI efforts that prioritize revenue growth over cost reduction, align the production of AI with the consumption of AI, through thoughtful alignment of business owners, process owners and AI expertise to ensure that they adopt AI solutions effectively and pervasively, unify their AI initiatives with their larger business transformation efforts, invest in AI talent, data and process change in addition to (and often more so than) AI technology. They recognize that AI is not all about technology." As an example, Aetna already uses AI to "design provider networks, prevent fraud and discover overpayments - traditional applications of analytics within the insurance industry" but "is now pursuing strategic initiatives to create more customer value with AI. In one Medicare-related offering, product designers used an AI-based method to customize benefit design." Interestingly enough, the percentage of Pioneers + Investigators remains at 50% of respondents, with 20% of Pioneers. Risk from AI is particularly strong in highly regulated industries like banking, which are being disrupted by new entrants such as Apple Card and Amazon Cash. "Out of three major possible applications of AI - cost reduction, revenue generation and new product development - most of the Pioneers are applying AI in at least two ways." The report's authors also emphasize the need to focus on the consumption of AI, not just its production. This means "developing a fertile environment in which producers can develop, champion, and implement AI solutions in strong collaboration with the business" and "developing sufficient expertise among business users so they can properly use the probabilistic nature of many AI solutions." Basically, it is about cultivating nontechnical leaders and business users.

AI is "not just a corporate race but an international one," especially between the United States and China. The MIT-BCG authors even started to conduct a separate survey with Chinese executives in 2018. "Chinese AI Pioneers are investing more aggressively and report a greater focus on business model transformation."

The MIT-BCG authors argue, based on their written questionnaire and interviews, that the key benefit of AI is not cost cutting but revenue generating, in particular in areas such as new product design. This actually seems to contradict The Economist, which claims that "in private, many bosses are more interested in the potential cost and labour savings than in the broader opportunities AI might bring." A consultant insists: "If you just cut costs and don't increase value for customers, you're going to be out of the game." Both The Economist and the MIT-BCG teams agree that AI pioneers may develop an unsurmountable competitive advantage, leading to corporate concentration and monopoly power.

The best known application of AI is for operations, specifically early maintenance. "Companies can combine data on past performance with those generated by smart sensors on machinery (part of the much-hyped "internet of things") to predict when a jet engine or a wind turbine is likely to fail," says The Economist. Other improvements will occur in demand forecasting and inventory management, leading to decreases in overstocking cost (estimated at $470bn worldwide) and understocking cost ($630bn). Packages are routed more efficiently - and INFORMS long-time member Jack Levis is even cited in the article, saying that "for every mile that [UPS] drivers in America are able to reduce their daily route, the firm saves around $50m a year." There are also enormous opportunities in customer service, where text mining (for emails) and voice analysis (for phone calls) can triage the most urgent service requests from the others. 

In human resources, a novel application will be to identify potential candidates for new positions from the pool of past applicants, and re-direct candidates who are not a good fit for the position they applied to toward positions that better match their skill set. But companies need to monitor their algorithms regularly to make sure they are free of bias, in particular for groups that are under-represented in the training set. The workplace of the future, though, may be much "creepier" than what we are used now due to employee monitoring. The example of startup Humanyze was particularly telling. Employees wear an "ID badge the size of a credit card and the depth of a book of matches. It contains a microphone that picks up whether they are talking to one another; Bluetooth and infrared sensors to monitor where they are; and an accelerometer to record when they move." This is supposed to provide a "full picture of how they spend their time at work." 

The potential corporate market for AI software, hardware and services is estimated at around $58bn by next year. Cloud providers such as Amazon and Microsoft have a clear advantage, while consultancies have been strengthening their capabilities. For instance McKinsey bought QuantumBlack in 2015. It remains to be seen whether AI wizards go to work for conventional consultants. The Economist argues that even IBM "finds it hard to get hold of the best talent. None of the top doctoral candidates in AI goes to work for IBM."

Personalized recommendations and faster deliveries sound like good things, and creating jobs that only exist because of AI or developing new drugs thanks to the power of machine learning is very exciting. But according to the McKinsey Global Institute, 14% of the global workforce could have their jobs automated away. This will require workers' retraining, which companies are reluctant to pay for. The second issue is of course data privacy. A third concern is a winner-take-all situation where the companies that win the AI race put their competitors out of business, increasing concentration in the tech sector and beyond.

What is 5G?

There is a lot of talk about 5G these days, from the discussion of broadband in "The Future of Texas" which I blogged about in a previous post to the latest DARPA BAA, so I figured I'd write a quick post about it. The main question, rather than "What is 5G?", may well be "Why do we care about it?"

According to a New York Times article from December 2018, "The transition to new fifth-generation networks, known as 5G, will affect how you use smartphones and many other devices." Those other devices include security cameras, autonomous driving vehicles, industrial robots and drones. 5G is "a set of technical ground rules that define the workings of a cellular network, including the radio frequencies used and how various components like computer chips and antennas handle radio signals and exchange data." At the time, Qualcomm predicted "initial median speeds of about 1.4 gigabits" or "roughly 20 times faster than the current 4G experience," with a goal of speeds around 10 gigabits per second, an obvious game-changer for streaming video, although frankly I prefer when technology breakthroughs affect society at a more meaningful level. 

A recent report by McKinsey starts with: "The fifth generation of wireless technology promises lightning-fast speed, incredibly low latency, and the capacity to carry massive numbers of connections simultaneously. Not surprisingly, the imminent arrival of 5G is creating a buzz in both the industry and the wider world." (Low latency means that data messages are processed through the network with only small or minimal delay. It is important for applications that need to work in real time, such as virtual reality, augmented reality and remote surgery.) There are actually two flavors of 5G: "low- and mid-band 5G focused on spectrum below 6 gigahertz and high-band 5G on spectrum above 6 gigahertz, particularly in the millimeter wave bands". The spectrum above 6 gigahertz is needed for ultra-high broadband speeds. But "Delivering the promised performance improvements of 5G through high-band spectrum, on the other hand, would require a fundamentally different architecture with much denser networks—something like 15 to 20 sites per square kilometer in highly populated urban environments, as opposed to two to five sites today. " 5G has the potential to unlock the potential of the Internet of Things. It could also benefit people who live in areas without wired broadband. Yet the McKinsey report is distinctly lukewarm about the financial prospects of 5G. It may be the equivalent of a tech arms race, where wireless operators know that if they didn't offer it, their customers would switch to their competitors. The report advocates seven "no-regret moves": 1. Develop 5G offers 2. Plan at the micromarket level 3. Rethink network deployment and operations 4. Consider sharing the cost 5. Secure access to key resources 6. Work closely with regulatory authorities 7. Develop a comprehensive spectrum strategy.

This idea of arms race to get 5G can also be extended at a national level. According to a December 2018 Wired article, which may be the best article I've read on the topic and also provides an interesting history of the various generations of wireless technology (I wonder what 6G will be!), "Industry experts and political leaders across the spectrum warn that it's possible for US companies to fall behind the curve if China or some other country is able to build the foundations for 5G more quickly... The fear is is that if China is first to 5G, its burgeoning tech industry will create the next global mobile platform; 5G could also give China an edge in the [Artificial Intelligence] race." 5G will make possible a new level of connectivity. The Wired article also explains some of the challenges better than the McKinsey report did: "The idea most commonly associated with 5G is millimeter-wave technology, which takes advantage of the very high end of the wireless spectrum, where there’s plenty of unused bandwidth. It's this technology that could enable those 10-Gbps speeds, but it comes with a huge trade-off. Millimeter-wave signals are less reliable over long distances and are easily disrupted by obstacles like trees, people, and even rain. To make it practical for mobile use, carriers will need to deploy huge numbers of small access points in cities, instead of relying on a few big cell towers as they do today." Because of those challenges, "carriers around the world are now turning to what's known as the "mid-band" of the wireless spectrum... [where] signals can travel farther and more reliably, making it possible to blanket areas with coverage using large towers." As a sobering note, "Consumer advocates worry that the rush to 5G is undermining public oversight of the wireless industry." Read the article for examples. It also has a number of helpful links for further readings at the bottom.

If you are interested in the FCC's perspective on the topic, here is an overview of its 5G FAST plan, where FAST stands for Facilitate America's Superiority in 5G Technology. The strategy is described as having three main components:(1) pushing more spectrum into the marketplace; (2) updating infrastructure policy; and (3) modernizing outdated regulations.

Ars Technica, in a December 2019 blog post, also concludes that "5G marketing has been writing checks that actual 5G technology will have a lot of trouble cashing." It also pokes some holes in a lot of the hype of the mobile carriers, especially in terms of coverage maps and hardware. Here is an example: AT&T "now has three flavors on the menu: “5G E,” which [is its fastest 4G service, rebranded, and] isn’t 5G at all; “5G,” the low-band 5G that doesn’t have the speeds the industry’s touted; and “5G+” millimeter-wave service. Branding that calls out millimeter-wave 5G as the high-performance variant... invites people to think less of low-band 5G—perhaps as “5G minus,” “5G Light,” or “Basic Economy 5G.” This isn’t a great situation for app developers, either—and without apps that can show off 5G, consumers may wonder what the fuss was about." Someone mentioned that: “If you want your app to be hugely successful, it still needs to reach a lot of people. By requiring 5G, you would be limiting your consumer base pretty quickly.”

DARPA recently organized a "Spectrum Collaboration Challenge" (SC2) to use AI in order to better allocate the radio frequency spectrum given fast-rising demand. The website explains: "Today’s approach, which is nearly a century old, isolates wireless systems by dividing the spectrum into rigid exclusively licensed bands, which are allocated over large, geographically defined regions. This approach rations access to the spectrum in exchange for the guarantee of interference-free communication. However, it is human-driven and not adaptive to the dynamics of supply and demand. At any given time, many allocated bands are unused by licensees while other bands are overwhelmed, thus squandering the spectrum’s enormous capacity and unnecessarily creating conditions of scarcity." This obviously begged for a novel optimization approach. "Competitors reimagined spectrum access strategies and developed a new wireless paradigm in which radio networks autonomously collaborated and reasoned about how to share the RF spectrum." MIT Technology Review had an article back in October about it. University of Florida students won first place ($2m) and Vanderbilt students won second place ($1m).

China recently announced that it is already starting to develop 6G, which is supposed to bring even bigger capacity and lower latency. Some researchers also argue that it will be more human-centric and bring more security, secrecy and privacy. Others claim that 5G is a transitional technology and that 6G will be characterized, among other things, by the "multiplication of cells leading to huge overall investment and extremely high complexity if one approaches the architecture in a centralised way. Much more effective to go for a massively distributed control, with massive use of artificial intelligence both in the network resources and in the service applications (with terminals becoming network nodes and service platforms)." For that, though, we will probably have to wait another decade.