© Bank of America Corporation. All rights reserved.
With global spending on AI rapidly increasing, so will investment opportunities
Not too long ago, artificial intelligence (AI) may have seemed decidedly futuristic, but in today’s world, that is far from being the case.
If you drive a car with adaptive cruise control, tell your smartphone to search for a nearby restaurant or watch a recommended movie on a streaming service, you’re already using AI technology.
Companies whose products you use — from financial services companies to retailers — lean heavily on AI to learn more about their customers, improve manufacturing efficiency and even help choose new employees.1
It is hardly an exaggeration to say we’re no longer at the dawn of the Age of AI — we are living within it.
Defining AI by type
At its most basic, AI is made up of algorithms, or blocks of computer instructions, that can learn from the data they process. Using that data, they can create new — or modify existing — algorithms, essentially giving the appearance of being intelligent, at least to a degree. Broadly speaking, there are three types of AI: narrow, general and super.2 Much of the AI we see today is considered to be of the narrow variety.
“Narrow AI allows a machine to perform tasks that are usually associated with human cognition — but only in a single domain in which it has been specifically trained,” says Ehiwario Efeyini, senior analyst, Market Strategy, Bank of America Private Bank. “For instance, Deep Blue, the AI system that defeated chess master Garry Kasparov in 1997, couldn’t play checkers or tic-tac-toe because it lacked the ability to automatically use knowledge from one area and transfer it to another area, as a human might do.”
General AI, which has yet to advance beyond the conceptual stage, “would, in theory, allow machines to understand and reason, not unlike humans,” Efeyini says. “But this does not apply to the AI systems in use or under development today.”
And then there is a hypothetical third category, Super AI. If that ever becomes a reality, many decades from now, we’ll have reached what some call “the singularity.” That is the point where machines become smarter than people in almost every area and AI autonomously engages in a kind of runaway development cycle that’s beyond human control.3 That may seem like a daunting scenario, yet Super AI may also hold the promise of helping to solve many of the world’s most complex issues, such as disease and climate.
The concept of artificial intelligence has been around since at least the 1950s (in the writings of British mathematician Alan Turing and others), and AI has shown up in Hollywood movies for almost as long (the sci-fi classic “2001: A Space Odyssey,” featuring a seemingly sentient — and murderously paranoid — AI-based computer named HAL,4 recently turned 50). The technology itself has been making occasional headlines since the 1990s (see paragraph on Deep Blue). But only in recent years has the technology insinuated itself, seemingly quite quickly, into our everyday lives.
Here’s how that happened:
A product of bigger, stronger, faster technology
There are a number of factors in AI’s dazzling expansion, says Joseph P. Quinlan, head of Market Strategy at Bank of America Private Bank. “One key driver is recent technological improvements,” he says. “More advanced semiconductors, new programming techniques such as machine learning, faster and more pervasive networks and the expansion of cloudbased data centers have all led to faster computing, faster communications and a greater ability to move huge amounts of data in a shorter time.”
This techno-supercharging has also enabled the spread of linked devices, or what’s often called the Internet of Things (IoT), by the billions. As Quinlan notes, “With the IoT, people are using more connected devices, from watches to washing machines, and those items themselves can communicate with each other via the cloud.”
Newer technology has also enabled the proliferation of voice-activated virtual assistants, like the kind on your smartphone, which seem to “chat” with you in real time. In reality, you’re probably not talking with the device itself, but rather with AI on a distant server, where most of the voice recognition and speech generation occurs (and where, perhaps, your voice is recorded).
At the same time, Quinlan says, “we are living in an era of what’s often called Big Data.” And how big is big? “Some research suggests more data has been created in the past few years than in the entire history of the human race,” he says. “And that mass of data could become 40 times larger in just a few years.”5
Indeed, every time you send a text message, load a photo to the cloud, post on social media or use a search engine, you add to the data mix. And it’s safe to say that as millions more people become connected (and set up social media accounts), data production will continue to skyrocket. 6
BANK OF AMERICA AND THE RESPONSIBLE DEPLOYMENT OF AI
The potential held by artificial intelligence to transform business and society is vast. But as the medical, travel and financial services industries consider how best to use it, they should ensure that they do so wisely, says Cathy Bessant, Bank of America’s chief operations and technology officer.
“We’re responsible for the outcomes that the technology creates,” says Bessant, adding that financial institutions — given their impact — have “a monumental responsibility” to get AI right.
To meet that responsibility, Bessant advocates inclusion of ethical and moral considerations in the technology’s up-front development, rigorous testing before implementation and strict governance throughout its use. For Bessant, it’s critical to understand that AI is only as unbiased as the people who create it.
“As a product of humans, artificial intelligence is subject to the same foibles of human nature,” she says. For example, an algorithm using historically, statistically or geographically biased data might reflect those biases in its decision-making and negatively impact clients. According to Bessant, avoiding such unintended consequences is a key goal of responsible AI development.
As part of its efforts to do that, Bank of America has partnered with Harvard University’s Kennedy School of Government to establish the Council on the Responsible Use of Artificial Intelligence.7 The council brings together leaders from private business, education and government to consider how best to capture the benefits of AI while taking a deliberate approach to minimizing its risks.
“It’s incredibly important that AI deployment doesn’t outpace our ability to understand and act on the implications of what we’re doing,” Bessant says. “Ultimately, responsible AI is not about what we can do, it’s about what we should do.”
Shining a light on dark data
Without question, a mountain of stored data has accumulated, but it turns out that only about 10% of it is usable in its current form; the other 90% is considered “dark,”8 or unstructured — a bit like a city map without street names. Nonetheless, many corporations have continued to collect this unstructured data with the hope of using it down the line. With AI, that is now possible.
Certain machine learning and analytical tools featuring AI are helping companies extract usable information from this deep pool of data, meaning that once-buried customer purchasing histories, patient records, traffic patterns and old documents are being made newly accessible. One key issue here is customer or patient privacy, which all too often has proved difficult to protect in recent years. As more of this unstructured data becomes useable, maintaining anonymity and using information responsibly will become ever more crucial.
Beyond factories and medical facilities
Robots have been used in automotive factories and hospital operating rooms for years, but AI is pushing long-standing boundaries. In auto plants, for example, recent developments have equipped machines with the ability to perform tasks like welding and painting, while simultaneously “looking” for product defects.9 In the operating room, AI is bringing more autonomy and precision to surgical machines, in part by using improved visual inputs.10
Meanwhile in Japan, Quinlan says, AI-driven robots are helping to provide care for a growing elderly population in a country that has a declining number of care providers. (For more, see “A Robotics Revolution in Japan.”)
A ROBOTICS REVOLUTION IN JAPAN
By Joseph P. Quinlan, head of CIO Market Strategy, Bank of America Private Bank
Japan is leading the way in confronting an aging society and the knock-on effects of a labor shortage and soaring costs for elder care. In doing so, it’s also part of the larger pursuit of a robotics revolution that could transform Japanese society and be replicated around the world. The revolution goes well beyond the factory floor. Here are a few examples of how:
The revised Japan Revitalization Strategy envisions “a new industrial revolution driven by robots,” and the country is considered among the best at producing such technology, along with Germany, Switzerland and China.
Robots that can learn
Using neural networks and machine learning, AI is now capable of creating its own rules of operation. “One way it does that is by finding patterns in large datasets, and then using these insights to develop its own rules of operation and to make decisions,” says Efeyini. In short, some AI can learn and then change an approach, as needed. And on the horizon are AI systems that will have the ability to correct themselves. “Researchers have been focusing on developing algorithms that can help AI more easily learn from its own mistakes,”11 Efeyini says.
Since digital images are essentially just another form of data, AI can often find patterns within them. “The software powering autonomous cars works this way,” Efeyini says. “You want the car to recognize other objects on the road such as cyclists, pedestrians, road signs and other vehicles. So instead of writing computer code to specify the features of another car, which would take considerable time and expense, you can now feed a computer many images of cars, and it’ll learn visually.” A similar kind of AI is at play when your smartphone organizes the photos of your parents into a single album (whether you asked it to or not).
Similarly, in recent advances in the health care field, he says, “an AI system was able to detect early-stage skin cancer from patient images at a rate of accuracy comparable to human dermatologists.” 12 Other improvements in AI-supported medical detection have been seen in lung cancer screenings, as well. 13
Another type of image identification operates in facial-recognition (FR) technology. For an example of this process, look no further than your smartphone. AI can unlock your phone by projecting a swarm of infrared dots onto your face and reading the reflected light, comparing what it “sees” with the image of you stored in its memory.
As use of facial recognition grows, it is raising considerable concerns about privacy. To help understand the broader (some might say Orwellian) uses of FR, consider China. As Quinlan notes, “AI-powered FR is being tested by the Chinese government, with a primary goal of using it for law enforcement.” In this case, multiple cameras scan crowds, isolating each individual face; AI algorithms then search a central database for those faces.
“If the system flags individuals with a criminal record, or spots someone breaking a minor law such as for jaywalking,” Quinlan says, “local police, wearing special augmented-reality glasses, can spot and, perhaps, arrest the offender.”14 While use of FR in the U.S. may not be as acute as it is in China, some police departments here routinely employ it,15 and several major U.S. airlines are already using “curb-to-gate” FR systems in airports.16
It seems almost inevitable that in time, facial recognition technology will see a wider role among government agencies and corporations, and calls for its responsible use will surely grow louder.
(For more on corporate responsibility, see “Bank of America and the Responsible Deployment of AI.”)
Despite what seems to be AI’s already ubiquitous presence in our lives, we are likely to see an even greater proliferation in coming years. Indeed, global spending on AI is expected to reach $77 billion a year in 2022, up from about $24 billion in 2018.17
As Quinlan notes, “We think AI will become even more widespread, in time, as sales of devices that connect to the internet rise, more companies begin making sense of dark data, and millions of people gain access to the internet for the first time.”
Efeyini adds that, “in contrast with the dot-com mania of 20 years ago, today’s new technologies are experiencing widespread commercial application, generating cost savings and delivering a range of new value-added products and services, whether through artificial intelligence software, cloud computing, robotics, digital media, high-speed telecom networks or connected objects.”
While the use of AI will continue to raise genuine social and ethical concerns — about privacy, political control, malevolent robots and the like — it is already improving lives and creating efficiencies in ways that even more imaginative minds may never have envisioned. And the same may be true for investment opportunities.
For more on AI and sectors that may benefit from the application of AI, contact your advisor.
1 “A.I. as Talent Scout: Unorthodox Hires, and Maybe Lower Pay,” The New York Times, December 2018.
2 “What is Narrow, General and Super Artificial Intelligence,” Tech Talks, 2017.
3 “What is the Singularity?” The Economist, May 14, 2019.
4 HAL: Short for “Heuristically programmed Algorithmic computer,” according to author Arthur C. Clarke, in his The Lost Worlds of 2001, 1972.
5 “10 Key Marketing Trends for 2017,” IBM, 2017.
6 “Internet Usage Statistics,” Internet World Stats, June 2018.
7 “Bank of America and Harvard Kennedy School Announce the Council on the Responsible Use of AI,” Forbes, April 23, 2018.
8 “Dark Analytics: Illuminating opportunities hidden within unstructured data,” Deloitte, 2017.
9 “Five Ways Artificial Intelligence is Impacting the Automotive Industry,” Ignite Outsourcing, 2018.
10 “Artificial Intelligence, Robots and the Operating Room,” ieee.org, 2018.
11 “New Algorithm Lets AI Learn from Mistakes, Become a Little More Human, futurism.com, 2018.
12 “Skin Cancer Classification with Deep Learning,” Stanford University, 2017.
13 “End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography,” Nature, 2019.
14 “Inside China’s Dystopian Dreams: A.I., Shame and Lots of Cameras,” The New York Times, 2018.
15 “How facial recognition became a routine policing tool in America,” nbcnews.com, 2019.
16 “Delta to use facial recognition in Atlanta’s international terminal,” Atlanta Journal-Constitution, 2018.
17 “Worldwide Semiannual Cognitive Artificial Intelligence Systems Spending Guide,” International Data Corp., 2018.
Opinions are those of the author and subject to change. The investments or strategies presented do not take into account the investment objectives or financial needs of particular investors. It is important that you consider this information in the context of your personal risk tolerance and investment goals. Due to the time-sensitive nature of the content and because investment opinions may have changed since the time any comments were made by research analysts, the latest Merrill investment opinion and investment risk rating for any particular security discussed should be reviewed, including important disclosures, before making an investment decision.
International investing presents certain risks not associated with investing solely in the U.S. These include, for instance, risks related to fluctuations in the value of the U.S. dollar relative to the value of other currencies, custody arrangements made for a fund’s foreign holdings, political risks, differences in accounting procedures and the lesser degree of public information required to be provided by non-U.S. companies. Investing in emerging markets may involve greater risks than investing in more developed countries. In addition, concentration of investments in a single region may result in greater volatility.
Currency risk is the risk that exchange rate fluctuations will reduce the value of returns. This arises when investments denominated in foreign currencies are purchased.
The Chief Investment Office, which provides investment strategies, due diligence, portfolio construction guidance and wealth management solutions for Global Wealth & Investment Management (“GWIM”) clients, is part of the Investment Solutions Group (“ISG”) of GWIM, a division of Bank of America Corporation.