AI: The Good, the Disruptive, and the Scary
As a starting point for this issue, Cutter Fellow Steve Andriole presents a brief, multifaceted overview of AI — “the good, the disruptive, and the scary” — and sets the backdrop for further exploration. He outlines some recent advances in and key limitations of AI and explores how AI could disrupt several domains, such as insurance, banking, law, real estate, and education. He then discusses the impact that the deployment of intelligent systems will have on jobs and the professional opportunities that will arise.
Those who develop and sell “intelligent” applications understand the extraordinary implications — and opportunities — of artificial intelligence (AI) solutions. Amazon, IBM, SAP, IBM, Oracle, Google, Cisco, Microsoft, Alibaba, Comcast, HP, Intel, Facebook, Apple, eBay, Thermo Fisher Scientific, Samsung, Dell, Foxconn, Huawei, and Baidu — just about all the players — are racing across the value chain to sell what they have to sell: smart applications that can save money and make money. CEOs, COOs, CIOs, CISOs, CDOs, CFOs, and CTOs are all waiting impatiently to deploy applications that will save them time, effort, and money — especially money they now spend on expensive humans. They see AI as a cost manager and a profit center. Is this a revolution waiting to happen? Absolutely yes.
AI will continue to impact industries and professions vulnerable to deductive replacements and routine automation. But — for the first time — AI will also displace a whole lot of knowledge workers — well-educated professionals — especially in the banking, legal, education, real estate, and insurance industries, among a broad range of service industries. Its impact on the transportation and manufacturing industries will accelerate as well. In fact, there’s no limit to the applied potential of AI, which will conceivably become as ubiquitous as databases, networks, and personal computing devices of all kinds.
The good news is that artificial intelligence is already many things to many people and represents a variety of alternative methods, tools, and techniques either already in the field or under aggressive development. An important distinction among intelligent solutions is strength. “Weak” AI solutions are ones essentially preprogrammed to “solve” problems in well-bounded, usually deductive domains like routine task management (e.g., record-keeping, FAQs and answers, and customer service). “Strong” AI solutions are those capable of extrapolating and “reasoning” beyond well-bounded domains, such as those that can “diagnose” problems in real time.
Weak AI system “intelligence” is defined as the consistent, persistent performance of repetitive, repeatable tasks. Filing a tax return only requires weak AI, since tax rules are well-defined (well-bounded), deductive, and codified. Strong AI systems can “think” by generalizing intelligence across problem domains. While we’re some years away from intelligent systems with “consciousness,” strong AI systems will eventually mimic human capabilities, though there’s heated debate about how “human-like” they can ever become.
The good news is that weak AI is exploding. But don’t confuse “weak” with ineffective. Some of the domains to which weak AI is applied include financial trading, market analysis, insurance underwriting, fraud protection, plagiarism checking, email management (including spam filtering), and multiple forms of tax preparation. These applications are inexpensive to develop, easy to deploy, and therefore extraordinarily cost-effective. In fact, their use has already become best practice.
Strong AI is on its way in several forms. Intelligent automation, especially robotic process automation, is adaptive to changing tasks. Machine learning — especially deep learning — super-charges intelligence with powerful knowledge representation techniques like neural network modeling. Enabling these applications are tools like machine vision, robotics, and natural language processing.
Billions of dollars are pouring into the development of both weak and strong AI. According to CB Insights (and other industry investment trackers), AI remains one of the largest investment areas for private equity venture capitalists, corporate venture capitalists, and companies across multiple industries that the technology industry has ever seen. High levels of financial interest are predictors of technological progress.
“Disruption” is a matter of perspective. It has been placed in quotes here because artificial intelligence is indeed disruptive, which many industries, companies, and executives find exciting as well as threatening (which may explain why most disruption comes from startups and not industry incumbents). There are some industries that AI will disrupt more (and faster) than others, such as insurance, real estate, banking, law, and education. Each of these industries will be disrupted not just by weak and strong AI, but also by regulatory changes that will clear the way for major disruption. Weak AI is already hard at work in these industries, but stronger AI is banging on the door. Let’s look at some of the major processes in these industries and how AI will disrupt not only these processes and but whole business models.
The insurance industry has already been attacked by digital agents, but the digital army is now poised for a takeover. Most millennials do not use home, auto, and life insurance agents:
Instead of working with local agents to find the right coverage, 67% of millennials are purchasing directly from insurance companies, leaving agents out of the picture.
Why in the world would anyone with a computer or smartphone make an appointment with a human being and physically travel to an office? If present trends continue, insurance agents will disappear completely about the same time we bury the last baby boomers. Agents and the entire process will be replaced by intelligent chatbots and virtual assistants, which will optimize insurance products according to the individual circumstances of each client. The same assistants will handle claims. The processes that are especially “vulnerable” include underwriting, claims processing, transaction management, fraud detection, risk management, and, perhaps most importantly, insurance planning. The shift from weak to strong AI will be disruptive, especially when intelligent virtual assistants replace human insurance agents, which will occur steadily over the next five to seven years.
The real estate industry is under attack from companies like Open Listings and the larger “for sale by owner” (FSBO) community. But the traditional players have some powerful friends that lobby endlessly to keep their hold on how real estate is bought and sold. There are so many hands in the typical transaction that it’s impossible to easily identify all of the financial vested interests in real estate transactions — which makes the industry difficult, but not impossible, to disrupt. Consider executive coach and author Bruce Kasanoff’s point of view:
95% of a broker’s role could be handled better by well-designed technology systems. Bidding, for example, could be handled by an automated system that includes legally-binding documents that would be instantly accessible to each party’s attorney … the fact is that one thing keeps the broker’s role alive today: the regulations that govern the real estate industry.
The implications for AI here are everywhere. The entire list/bid/buy/sell/regulate process can become victim to automation. Much of the automation only requires weak AI; strong AI will manage advanced bidding and negotiation. Customized listing processes will be developed, implemented, and managed by intelligent systems. Property search will be automated based on preferences and life circumstances that virtual assistants will calculate. The property search process will become proactive the moment a buyer expresses interest in moving or renting, when buyer and market databases will merge into personalized and customized options.
According to Moven founder Brett King (as reported by Eric Rosenbaum of CNBC):
The biggest banks in the world in 2025 will be technology companies, and banks that grew through branch acquisitions in the ‘80s and ‘90s, that grew by physical bank presence, will have a real problem.
Money is also disappearing. Way back in 2012, International Business Times correspondent Jacey Fortin reported that:
In Sweden, monetary transactions made with physical cash are down to three percent of the national economy. In most Swedish cities, public buses don’t accept cash; tickets are prepaid or purchased with a cell phone text message.
The US lags, of course, but it’s only a matter of time and money — especially because of the control that cashless transactions provide governments and the financial gains banks accrue from closing physical branches and going cashless. What aspects of banking cannot be automated? Which processes would not benefit from weak and strong AI?
AI-with-blockchain will be the preferred banking architecture, but basic banking processes are not the only ones vulnerable to intelligent systems. Wealth and portfolio management will be empowered by advanced analytics capable of customized predictive and prescriptive planning, the holy grail of financial management.
Why are there so many lawyers? There’s no need for so many physical, organic, living, breathing legal professionals in the digital era. Expertise defined around “rules” can be automated and distributed at the touch of a key, a verbal command, or a reasonably intelligent (even weak) assistant. Automated reasoning (strong AI) will replace many lawyers (and, for that matter, doctors, accountants, professors, and engineers), primarily because “the law” and other professions are well-bounded, codified domains — precisely what intelligent systems require to excel. Disruption has already arrived. According to data privacy lawyer Sterling Miller, three events are imminent:
1. Some legal jobs will be eliminated (e.g., those [that] involve the sole task of searching documents or other databases for information and coding that information are most at risk).
2. Jobs will be created, including managing and developing AI (legal engineers), writing algorithms for AI, and reviewing AI-assisted work product (because lawyers can never concede the final say or the provision of legal advice to AI).
3. Most lawyers will be freed from the mundane task of data gathering for the value-added task of analyzing results, thinking, and advising their clients.
Consider also legal vendors like LegalZoom and Atrium that understand how the process from weak to strong AI will evolve. Weak AI applications — what we still call “expert systems” — power many of the self-service functions at LegalZoom and Atrium. Soon, legal advice dispensed by human lawyers will be replaced by intelligent assistants with de facto JDs informed, again, by personalized and customized databases.
And why are there so many teachers? There are so many teachers, trainers, and professors because there’s a tremendous need for their services. But there are problems. Many of the domains that they teach, such as science, technology, engineering, and mathematics (STEM), are dynamic. The currency of the teachers, trainers, and professors who teach domains that frequently change is a challenge. Pedagogy also tends to be static among veteran teachers, trainers, and professors.
Weak AI can automate the basics of education, such as enrollment, curriculum optimization, and grading. It can also customize and personalize learning experiences, correlating with personal and behavioral data that will improve the learning process. Intelligent analytics can measure effectiveness, which results in teacher/trainer/professor assessments, as well as assessments about the effectiveness of curriculum content. Said differently, total quality management can be automated by intelligent assistants. Intelligent tutors can assist where necessary.
First, let’s acknowledge the lack of intelligence around artificial intelligence. Members of the US Congress, for example, know little or nothing about the technology — which is worrisome on so many levels, especially when we consider the technology’s inevitable impact on the US and global economies. Most CEOs — and even most CIOs and CTOs — also know relatively little about AI — though, when surveyed, they list AI as one of the most important technologies of the 21st century. The judicial system has its head in the sand. The general population only appears to understand AI in the way Hollywood dramatizes it, like the way it was exhibited in 1992 in Minority Report, in 1999 in The Matrix, and, more recently, in the movie Her, HBO’s Westworld, and several episodes of Netflix’s Black Mirror.
Try this: go to a party and randomly ask the guests what they think about AI. I’ve done it several times and the word cloud illustrates robots, Alexa, Watson, and Westworld, but nothing about machine learning, knowledge representation, or neural networks. The gap here is scary.
Weak and strong AI will displace labor, but what happens when widespread job displacement occurs? Hardly any of the pundits describe specific displacement management strategies or plans. This is the scary part of the story (not AI hostages or AI-instigated Armageddon). How many industries and companies will know how — or even want — to manage displacement, especially if displacement management offsets savings from labor reduction?
Corporate HR departments will explode with complaints and lawsuits, and collapse under the weight of the exit packages they’ll be forced to give, at least temporarily. Young and aging factory workers — along with lawyers, accountants, and educators — will forget their purpose. Politicians will stare into the technology headlights — again — frozen by their own confusion and vested self-interests. Executives and shareholders will squeal with profitable delight. Universities will adjust their curricula or rapidly lose customers. Pain will pervade the corridors (but not the boardrooms) of the hard and soft industrial worlds, though this time the corridors will be wider and prettier than they’ve been in past displacement revolutions because knowledge workers typically ply their trades in prettier places.
Transition — and displacement management — will be the challenge. This is the part of the AI story that deserves much more attention, though very few analysts talk about what happens after displacement occurs. What happens to the paralegals, lawyers, accountants, medical diagnosticians, manufacturers, supply chain managers, and customer service representatives when they’re displaced? Where will they go? What will they do?
Consider the doom-and-gloom proposition from strategy and management consultant Xavier Mesnard at the World Economic Forum:
The risk we are facing in the near future is mass unemployment for some categories of workers, combined with lack of skills in other categories — and the political and social implications of such imbalances.
While Mesnard’s predictions may be scary, they deserve serious attention. (What if he’s right?)
The assumption is — as it always is — that we have time to think all this through (or ignore it for as long as we can); perhaps even as much time as it took cars to totally replace the horse-and-buggy, or about a quarter of a century. But displacement this time will be much faster and more brutal than it’s been in the past. There won’t be anywhere near as much time as in past industrial revolutions because this revolution is digital and therefore, by definition, pervasive and exponential. We’re already digitally connected, so displacement has an infrastructure in place, which is an unprecedented revolutionary reality (note, for example, how the adoption of automobiles was constrained by the lack of a highway infrastructure).
One example boldly illustrates the point: self-driving vehicles will be widely deployed in five to seven years — not decades from now. The enabling digital technologies that locate, direct, and power autonomous vehicles are in place, which will accelerate adoption and displacement. What will happen to the operators of these vehicles and those who enable the entire vehicle and transportation supply chain?
The emerging crisis is not about inevitable displacement but about reaction and replacement policy. Since corporations will largely benefit from displacement (in the same way they benefit from cheaper global labor markets), governments will be expected to intervene at some level. But given how clueless government already is, and how slowly it moves even when it’s informed and committed, the prospects for effective displacement management are low. It may be that we’re focusing on the wrong problem — and the wrong problem-solvers. It may be that completely unpredictable events will define displacement management policy. So how do we manage displacement?
We have less than a decade to figure all this out, which is a problem given the current level of intelligence about AI. There’s a lot of education and analysis necessary to minimize transitional and displacement damage and optimize another global transition. The backdrop is complicated by a variety of other competing priorities. The first step is awareness.
We have seen whole industries disrupted by digital technology over the past two decades. But all that disruption combined only represents Disruption 1.0. The five industries discussed here represent just the beginning of Disruption 2.0. Like the iPhone, there are likely to be many incarnations over time. As machines get smaller, smarter, and cheaper, we’ll see more and more industries disrupted by digital professionals and their digital tools. The implications of continuous disruption are extensive and unpredictable. The world as we think we know it will never be (just as it never was).
AI has the potential to quickly supplement and displace huge numbers of hard and soft skills. We need some real intelligence to avoid a rocky process that could have major social, economic, and political consequences. Are there members of government willing to launch these discussions at the national level? Are there editors of major media willing to raise red flags? Would it make sense to seek global agreement? (CNN has time to analyze the power of music and meals. Maybe it could do a series on AI.)
This is clearly a multidisciplinary problem. Colleges and universities should focus on the implications of displacement. Government and foundations should fund research in the area. Corporations should develop displacement management plans. The major social platforms — Facebook, Instagram, Twitter, and so on — should proactively encourage discussions about the displacements about to occur.
AI will disrupt knowledge- and production-based professions and the fields that prepare and maintain these professionals. Note that “trickle-down disruption” will be much more impactful and sustained than “first-order disruption.” You can decide how much about AI is “good,” “disruptive,” and “scary.” There’s enough to go around, that’s for sure.