We are all apprehensive about artificial intelligence (AI) and the loss of jobs. Artificial intelligence and the future loss of jobs has been making us curious. How far AI job loss prediction is accurate makes us bewildered. Hence, Elon Musk created a news headline claiming that AI will render all jobs obsolete. On November 02, 2023, Elon made this observation while exchanging views with the British Prime Minster. He termed artificial intelligence (AI) the most disruptive force in history. At one point in time, AI will make all jobs needless. Against the backdrop of such a claim, the Future of Jobs 2023 report of the World Economic Forum (WEF) has downgraded the role of AI in jobs. Instead of stating AI’s role in killing jobs, the Future of Jobs 2023 of the WEF report looked upon the role of AI as a job creator and augmenter. This report states that AI will augment humans and help them perform jobs better than before instead of killing jobs. Hence, there will be additional AI skill needs for existing jobs—creating new jobs.
There is no denying that Elon Musk’s AI job loss prediction contradicts the recent view of the Future of Jobs report of the World Economic Forum. However, such a contradiction creates a Dilemma in the minds of students and policymakers. If students believe in Elon Musk’s view of AI in jobs, they are hopeless in preparing for their future careers. Similarly, policymakers and development practitioners are clueless in driving economic growth by creating jobs. On the other hand, if we believe in WEF’s recent view that AI will not kill jobs, many AI Innovation promises, like autonomous vehicles, find no direction. Believing in only augmenting AI’s role, do we ignore AI’s fundamental proposition?
Key takeaways of AI rendering all jobs obsolete
- AI for automating knowledge summarization and reproduction–large language modeling-based tools like CharGPT4 will likely take over jobs of automating summarization and reproduction of existing knowledge.
- AI for repetitive sensing and cognitive role–the development of sensors and sensing and perceiving algorithms will likely allow AI to take over humans’ sensing and cognitive role in delivering repetitive service jobs.
- AI faces a barrier to automating Innate abilities–despite the ease of automating Codified knowledge and skill, AI faces high barriers to automating the role of humans’ innate abilities in jobs.
- AI leaves dull jobs for humans–due to the requirement of innate abilities like dexterity, warmth, and common sense, jobs considered to be mundane, like household chores and elderly care, will likely be left for humans.
- AI leaves knowledge and idea creation jobs to humans–despite knowledge summarization capability, as curiosity and creativity appear to be out of reach of AI technology, knowledge and idea creation jobs will likely remain only for humans.
- Humans for empathy, Passion for Perfection, and imagination–despite having human-like looks, voice, and language skills, due to a lack of empathy, passion for perfection and imagination, AI tools will be predominantly assisting humans in performing most of the jobs.
- Merits of claims about AI will render all jobs obsolete–due to the inherent weakness of AI technology core and unfolding knowledge about humans’ role in work, claims like AI will render all jobs obsolete appear to suffer from lack of substance and hype-creating attempts.
Purpose of AI in Jobs
The primary purpose of AI has been to imitate human-like intelligence so that we can replace the cognitive role of humans in the control loop of machines with machine intelligence. As a result, we will overcome the limitations of humans’ sensory, physical, and cognitive abilities. Consequentially, we can make machines with AI abilities for more effective, efficient, and safer performance. For example, a robot butcher with X-ray sensing can see through meat, resulting in a more accurate cut than a human butcher can ever deliver. Similarly, robot cars have the potential to overcome 700ms cognitive latency of human drivers to make roads safer. Hence, AI’s ultimate goal is to make machines’ operations free from humans. Therefore, loss of jobs due to AI appears to be inhabitable. AI and job loss have a natural correlation. Thus, such objective and possibility contradict the notion of WEF’s Future of Jobs Report 2023 about the consequence of AI on Jobs.
However, does it mean AI will succeed in rendering all jobs obsolete? What about the jobs in developing AI itself? Are we under the impression that AI will itself be improving AI, creating an alien race? Besides, what about the jobs requiring empathy, love, affection, and warmth? Are we expecting that AI will kill those jobs as well?
AI job loss predictions—a review
The most dramatic prediction about AI job loss has been Elon Musk’s claim. According to such a claim, all the jobs will be taken over by AI. There will be no jobs for humans to perform better than AI. Well, that may sound utopian. Here are a few AI job loss predictions:
- 45 million Americans may loss jobs due to AI by 2030
- PwC survey finds that 30% of respondents from 44 countries are apprehensive about losing jobs to AI.
- Microsoft’s 2023 Work Trend Index report reveals that 74% of the workforce in India is nervous about AI’s potential to job loss to AI.
- In knowledge-based service roles such as customer care and secretarial service, 40% of working hours could be impacted by ChatGPT-4 type large language models.
However, AI job loss statistics is not frightening. According to CBS News, in the USA, jobs lost due to AI in May 2023 were only 4,000. Indeed, at this rate of job loss, the job loss number due to AI will not add up to 40 million by 2030. Hence, fear of job loss from AI needs investigation. We need to dig down like possibility and limitation of job loss due to AI.
There has also been a tendency to predict less loss of jobs requiring innate abilities like physical and manipulative. On the other hand, job loss prediction for knowledge workers has been rising. Significant language model-centric AI applications like ChatGPT-4 have shown vital signs of service production using or compiling existing knowledge.
AI job creation prediction
Although Elon Musk made a headline claiming that AI will render all jobs obsolete, there have been predictions that AI will create jobs. For example, the WEF study predicts– in 2023–about a 40% increase in the employment of AI and machine learning specialists by 2027. The study also indicates a significant jump in jobs like information security analysts, data analysts, and scientists. These areas, combined, may witness the creation of 2.6 million jobs during the 2023-2027 period. Such predictions proceed further stating that during the 2020-2025 period:
- AI will eliminate 85 million jobs &
- create 97 million new ones.
Defining AI and Machine Learning
During the early era of computers, the urge to create machines capable of simulating human intelligence led to the term “Artificial Intelligence” by John McCarthy in 1956. Of course, they succeeded in simulating human intelligence–partially, though. They developed a computational model to simulate the trajectories of ballistic missiles or explosions of nuclear bombs. But did Mr. McCarthy use the term AI to refer to artificial general intelligence (AGI)–software capable of any intellectual task humans can do? Perhaps, no. The definition of AI has been evolving, resulting in vague expectations.
The next issue is about machine learning (ML). Is it AI? ML has been tempted to be called AI since it imitates human learning behavior through training samples. Unfortunately, it creates gross overestimation and suffers from blown-up expectations. As a result, early progress of ML leads to inflated expectations. The current state of ML refers to the approximation of a set of limited variations of practical situations. Deep learning or Neural network-based ML is about approximating the reality represented by training samples. Hence, it’s a kind of averaging type algorithm defined by a weight matrix. Therefore, ML’s performance is limited by the approximation of training samples—let alone imitating humans’ imagination and ability to deal with the unknown. The current ML cannot develop a scalable model from training sample data. Hence, ML fails to recognize all kinds of horses by learning about horses through one or a few images of a horse. Even recent splashes such as OpenAI’s ChatGPT and other generative AI tools apply data science methods for compiling data only—let alone imitating human intelligence in creating knowledge.
Extremism in AI job loss—creating AI hype cycle
Demonstration of autonomous vehicles in flowing lanes, detecting and obstructing obstacles, and following a route to reach a destination created the impression that robot cars were around the corner. Hence, the fear of losing millions of jobs for professional drivers became headline news. Similarly, ChatGPT-4, like significant language model-based applications, has been fueling the speculation of rendering all knowledge-intensive jobs obsolete. Does it mean that millions of knowledge professionals like medical doctors or teachers will lose jobs due to artificial intelligence?
In many cases, approximate knowledge compilation and synthetization by generative AI applications serve practical purposes. For example, such applications could be helpful in predicting which customers are most likely to cancel or which credit card transactions are fraudulent. As human judgment often faces difficulty reaching more than 80 percent accuracy, 90 percent prediction accuracy of AI tools could be a better alternative. But is such performance acceptable in prescribing medications to patients? Upon losing more than $2 billion in automating cancer dose prescriptions, IBM got a lesson.
Due to initial demonstration and lack of clarity about what it takes for AI to meet expectations in different jobs, we suffer from AI hype cycles.
Will AI render all jobs obsolete?
There is no denying that AI will make jobs obsolete—but not all of them. AI isn’t a silver bullet. For example, an internal combustion engine or steam engine relieved humans from supplying physical energy to rotate a wheel or pull a cart. However, it did not completely eliminate human roles in providing energy. There will be a gradual transition of giving increasing roles to technologies like AI.
Of course, ML plays a helpful role in compiling knowledge. However, its ability to generate knowledge is beyond the underlying science of machine learning. Besides, complete representation of reality through a finite number of data sets appears to be impossible in many practical cases. Furthermore, a significant amount of knowledge we apply is beyond codification. More importantly, we keep creating knowledge and feeding it to creativity to generate ideas for dealing with unfolding limitless variations. Hence, the proposition of AI rendering all jobs obsolete suffers from a lack of science base.
Besides knowledge, human beings are rich in empathy, affection, love, and passion for perfection. These abilities are precious in many jobs like teaching, healthcare, customer service, and innovation. The current state of scientific disciplines, from neuroscience to computer science, do not have adequate insights to codify them—let alone imitate them as AI applications.
Last but not least, it is about human’s innate sensory and physical capabilities and their roles in Getting jobs done. For example, intuitively, we use the rich natural abilities of our hands in various jobs, starting from laying a bedsheet or holding a cup of tea. After 300 years of effort, our robot hand is highly primitive to our hands. Many such challenges need to be overcome to take over even mundane jobs. Hence, it’s time to draw lessons from dead robots or ASIMO robots about AI job loss.