AI Technology: The big picture
In previous posts, I’ve described how Artificial Intelligence (AI)-robotics is the driving force behind the global decline in manufacturing jobs and the middle class. This is especially the case in First World economies but is even beginning to affect the Chinese and other developing economies. Globalization (e.g., free trade, the offshoring of taxes and regulations) as well as Immigration also come into play but have comparatively less impact than most people believe.
This has resulted in a job market split between a modest increase in high paying and a rapidly growing pool of low-paying jobs, with an implosion of middle-rank positions.
A range of possible outcomes
AI scientists’ predictions of the long-term impact of robotics and other AI work applications on jobs range along a continuum from (1) AI technology doing virtually all human jobs to (2) new jobs replacing lost traditional blue collar jobs. Scientists in the first group believe there are no inherent limits to AI capabilities that would prevent them from becoming fully autonomous entities with human attributes. Those in the second group believe that AI-fueled economic growth will generate new jobs requiring sophisticated communication and other skills based on human empathy that robots will never be able to replicate.
- So far, there is more evidence supporting the first scenario. Let me explain. While there is as yet no confirmation of any impending breakthrough in AI replicating the extreme power and complexity of the human brain, the kinds of AI that enhance non-autonomous robotics and other digital functions are rapidly improving.
- The impact on jobs extends well beyond manufacturing, a sector in decline since 1979. Consider how by the 2020s, driverless commercial vehicles, automated fast food restaurants, and advances in health care delivery digital tech will upturn those industries, further reducing jobs.
- While it’s clear that AI applications are driving productivity and economic growth, the continuing decline in jobs will result in people having even less income to pay for AI-manufactured products. This challenge also includes white-collar workers, whose jobs have been in decline since the 1980s as a result of AI software applications.
What can be done?
- After the 2008 crash, countries that invested in training, encouraged entrepreneurs, research and social investment did better than those that didn’t. So, this would seem to be a workable set of policies to help reduce the continuing decline in jobs. However, bringing back the full range of old manufacturing jobs is impossible, given AI and other tech advances since the 1960s.
- Perhaps, governments should actively discourage the rise of AI, by, e.g., blocking the immigration of high-skilled tech workers. Unfortunately, if the White House moves to block future AI developments, breakthroughs would then shift overseas, making U.S. jobs even more vulnerable to global competition.
- Some economists believe that there is only one solution for the decline in jobs: a guaranteed living wage paid to the majority of citizens who will be left out of the workforce. Most of us are alarmed by this possibility. Barriers to its implementation include concerns about reduced incentives for entrepreneurship (that generate new jobs) and the overriding fear that without work, people will be denied their sense of dignity and be more vulnerable to vice. –For a guaranteed income to work there would have to be a radical cultural shift, with our identities no longer centered on work.
Is any job tech proof?
I summarized current thinking about the long-term impact of AI (Artificial Intelligence)-based technology on jobs. Bottom line—if a computer can do a job, it’s going to be ‘hired’ over a more costly human worker. The AI challenge to human labor extends beyond manufacturing robots to software that performs white-collar tasks. These developments account for the worldwide decline of the middle class. In the long term, few positions will escape this dynamic.
Jobs that will disappear
The University of Oxford projects that almost half of all jobs in the Western world could be automated within the next few decades. That includes accountants, paralegal workers, technical writers and many other white-collar occupations. At the same time, image-processing software will replace lab technicians; and driverless vehicles will eradicate truck and taxi driver jobs. Eventually, airline pilots, traffic cops–even soldiers could be displaced.
While some jobs have already been eliminated, others have been conflated into robotics maintenance positions. Any non-creative work that can be broken down into routine components is at risk. AI is accelerating its capabilities to accomplish this across most economic sectors. Moreover, the cost of automation software is decreasing, making it increasingly affordable for small businesses. The greatest single obstacle to applying new AI pattern-recognition breakthroughs is the relatively slow human pace of determining the best applications.
Are there any tech-proof jobs?
Third World manufacturing and low-skilled white-collar workers will be protected, at least for a while, because automation is still more expensive in countries like India than human labor. By comparison, in developed economies like those of the U.S. and Japan, the high cost of labor has driven manufacturing automation.
We’re still decades away from autonomous AI robots taking away all our jobs. Those that require human physical presence and sophisticated technical skills may never disappear (e.g., physical therapists, dentists, athletic trainers, counselors, clergy, and high-level programmers). Jobs that require complex interpersonal and motor skills will also be protected, as with hairstylists, dentists, and surgeons. That said, computers are now being used to improve surgical outcomes, e.g., for eye, prostate, and other operations.
Of the 25 fastest dying industries, ten are in manufacturing. Retraining workers in dying industries for viable jobs is a win-win. Within the imploding coal industry, for example, many former coal miners are being trained in basic computer programming skills. This transition has been surprisingly successful because coal miners already possess considerable technical proficiency in operating equipment and making exacting calculations—skills that overlap those required in programming.
Next let’s explore the kinds of computer-human collaboration that have improved work satisfaction.
How AI is improving the efficiency and quality of work
In the long term, everything I described so far about AI-assisted robotics and software ‘eating’ most human jobs seems inevitable. In the short term, however, this emerging dynamic is helping many white collar and other professionals escape boring, repetitive work tasks while significantly improving productivity. This includes, e.g., versions of email software that learn to read, organize and even respond to some emails.
A period of grace for knowledge-workers
Although this tech/work/cultural revolution is accelerating, there will be a grace period for knowledge work. Why? ‘Knowledge work’ remains competitive because every individual application requires a machine-based algorithm, each of which can cost a million dollars or more. On the other hand, computing power will continue to get cheaper while machine-learning algorithms become more generic, inexpensive and easy to produce. This has already made it possible for IT giants like Microsoft, Amazon, and Alibaba to launch their own cloud machine learning platforms, significantly reducing their R&D budgets. Eventually, those benefits will be passed on to small businesses via affordable subscriptions to algorithm development companies.
Another enormous benefit of the AI-based software is how it streamlines the decision-making process, making it easier for organizations and entrepreneurs to take the calculated risks that all must make. It’s also accelerating the growth of ‘next big thing’ startups.
Three AI-Tech modalities that simplify work
- Automated Machine Learning
All algorithms learn from human input-based big data. This progressively improves accuracy over time. For example, when FB introduces a new emoji, algorithms need to see examples of its usage before they can understand it. There is a labor-intensive side to this process: it takes millions of man-hours for Google to collect and label big data for their algorithms to work. The same goes for any company that wants to break into a new international market with a new language.
- Humans In The Loop
Algorithms that achieve 80% accuracy aren’t good enough. Fortunately, algorithms can self-identify areas where their confidence is low to alert us that our input is required to improve the reliability of AI-assisted processes and decisions. This dynamic can be seen in self-driving vehicle software advances.
- AI Active Learning
This third option combines the first two—i.e., human input helps improve algorithm accuracy both by making decisions the algorithm can’t and at other times, feeding back directly into the AI classification system. This requires a higher level of AI software sophistication that provides it with learning capabilities that are more similar to human learning.
What OWDT can do for you
Our staff at OWDT provide comprehensive services in website development, branding, and digital marketing. Our professionals work rigorously to integrate the latest in website technology and design with leading-edge digital marketing tools to help your company thrive. They know how to finesse a beautiful, brand-defining online presence with coordinated marketing strategies to expand your organization’s visibility and revenue.