The key to overcoming the self-inflicted problems and threats of technology lies in our own intellect. (Getty Images)
The world as we know it, together with the history of all knowledge spheres and bounds, has been governed by long-held beliefs, evidence-based information and proven/tested theories, through rigorous research and innovation. Before the dot-com bubble, expertise and knowledge were found (or formulated) through hours, days, weeks, months and years of reading, listening and searching archives or any available materials and artefacts.
Over the past 25 years, we’ve realised (through epistemology, ontology and axiology) that we reside in a knowledge-rich era where anything, anywhere, can readily be obtained online. Of course, we continue to question and/or verify what is discovered online. However, we now have the opportunity to search for wide sources of information that allows us to bring those very materials closer to us and even quicker.
Enter the era of Artificial Intelligence (AI)
This is not going to be another written piece on ChatGPT, as ChatGPT is only an example of the enormous potential that AI brings to society. Instead, this article aims to distinguish the opportunities, constraints, utilisations and perspectives of AI. And how AI will never replace human expertise (in hard-skilled-driven sectors).
This will be presented through my six Es:
Enablement, Extension, Enhancement, Empowerment, Ethics and Education.
Context (Let me start with setting the scene…)
Chesley “Sully” Sullenberger, a skilled and experienced pilot, landed his aeroplane in the Hudson River after a bird strike disabled one of the engines. This day brought many to their knees with shock and bafflement. Not because any passenger died (thankfully, there were no casualties), but because the pilot had the intuition to land the plane (out of all places) in the river. This real-life occurrence was captured in the epic film ‘Sully’, with Tom Hanks in the lead role.
As tensions settled, a commission of inquiry was held by the National Transportation Safety Board (NTSB) to determine (via computer simulations) whether Sully would have been able to turn the plane and land successfully on one of the landing strips at a nearby airport. All simulations showed that it would have been possible to land the plane on one of the landing strips, without hitting any buildings or objects (despite flying on a lower-than-normal terrain).
However, when accounting for the human factor, without any time to react to the situation, all simulations showed that the plane would have crashed into any building or object. As narrated by Sully, the key here was the human factor. Computer simulations lack the real-life pressures, anxiety and “realness” of the precise situation. Those who ran the simulations knew what steps to take after the bird strike.
In the real-life scenario, Sully and his co-pilot did not know what to do immediately — they had to interpret the risks and take a calculated decision on the next steps. Those extra few seconds can make the difference between life or death. When pressured situations demand it, coupled with skill and expertise, human expertise mostly comes to the fore.
Expertise Theory
The Expertise Theory demonstrates how skill and talent develops and emerges over time, practice and repetition through experiential learning, strategy, knowledge and application. It documents how individuals become experts in their domain or field — through a process of cognitive task analysis — and portrays the necessary skills and expertise more robustly, effectively and efficiently than novices.
Metacognition
Linking to this is Metacognition, which is the ability to think about one’s own thinking; and is built on several cognitive markers such as self-awareness, self-regulation, self-reflection, imagery, goal-setting and evaluation of one’s goals, learning and skills. Unless AI (and robots) will be programmed through an advanced set of cognitive skills, human expertise will remain triumphant in effective problem-solving and strategies to address many of the world’s problems, including the sustainable development goals (SDGs).
The fuel for optimal AI is enormous amounts of quality data. AI enables us to make sense of massive amounts of data. Through this high level of sense-making, it provides various ways of enabling tasks and activities for a diverse range of professions and expertise.
For example, companies have started to innovate smart glasses for the deaf that would enable them to read (through augmentation) what hearing people are saying, via speech-to-text. This enables the deaf to understand those who do not know sign-language a bit better and reduces their need to lip-read instead. This kind of enablement would be similar for the translation of languages in real-time or providing complex and accurate biomedical engineering designs for amputees requiring prosthetics.
Another way in which AI can maximise enablement for humans is through collaboration. An AI-based dashboard could be used to consolidate skills and interests from individuals and/or organisations. To ensure a better fit or collaborative pursuit, the dashboard could provide suggestions for possible collaborations, with virtually no boundaries, in the context of the gig economy and internationalisation.
Considering the examples listed above, the use of AI would still need to be led, developed, designed and governed by human expertise with inter-disciplinary expertise (ranging from both hard and soft skills), to enable various capabilities of human expertise.
In the context of healthcare, it is predicted that there will be a shortage of thousands of health professionals by 2030. AI has been identified as a potential solution to remove arbitrary or rogue tasks and save health professionals time and energy, so that health professionals can consult with more patients.
In this context, AI can extend a helping hand to indirect tasks (such as healthcare administration, patient records, a system for early warning signs, bookings of surgical theatres, or consolidated management of all patients in wards/ICUs on a health professional’s dedicated database or mobile app), so that health professionals can dedicate more of their energy and time to direct medical-related tasks that require only the expertise of trained and skilled people. We’ve already witnessed the potential of robotics assisting with surgery, but even then, the skilled human would lead the treatment, surgery or programme. As a result, AI cannot replace health professionals.
There are, of course, other professions and jobs that one would need to study to determine whether AI could be at risk of replacing people. In summary, it is only those jobs that include repetitive tasks, redundancy, perplexity or lack of adaptability, that would be at a higher risk of being replaced by AI. For professions and jobs that require versatility, adaptability, critical thinking and higher-order applied skills, human expertise is here to stay.
In a nutshell, AI can extend human expertise with predictive modelling, automated tasks, decision-making becoming simpler and analysing large amounts of data.
AI can also enhance what humans do. For example, in the cooking industry, thousands of recipes for a particular dish could be summarised by AI, and recipes could be designed to enhance the quality of a dish that tastes and looks better. This would enhance the capabilities of a chef or cook, irrespective of how many years of experience they have in the kitchen.
A robot chef could certainly follow and execute a recipe, but a human chef would be key to oversee the process that is being administered, which is hugely different to an automated manufacturing or production process. A similar approach could also lend itself in the field of agriculture to optimise farming methods and reduce the risks and exposure of pesticides. Enhancements of AI allow us to further improve the quality or value of something or intensify the quality of a system or process. With such enhancements, the human touch would still hold more weight than AI.
AI would be in a key position to support the ability of others by empowering them through better insights, skills and training. It could do this by providing and facilitating continuous and tailored learning that matches the individual abilities and circumstances of the person. It can also allow individuals to learn at their own pace and achieve better, scalable outcomes, in their domain or field of expertise. Furthermore, AI can empower financial analysts to make better decisions through measurable and delineated sources of patterns, trends and predictive modelling in the financial market. Very similar to taking the horse to the water analogy (without drinking it for them), when AI empowers, humans can do better.
There has been considerable information and reflections on the ethical foundations of AI and what it means in many industries. My prediction is that AI is only going to get smarter and more sophisticated. For example, we might begin to see a more adaptable version of robotics being applied in numerous sectors.
Some television films and series such as iRobot and Better Than Us explore the fiction-based capabilities that we could experience with robots. For example, the three laws of robotics (also based on fiction) is grounded on the premise of 1) no harm to humans and 2) obedience to humans. Robots would be trained to inflict no harm on humans, obey what humans say and protect its own existence provided that it does not conflict laws 1 and 2.
There will come a time where robots will also have the ability to think and do, just like humans (with minimal emotional connection). In this applied context, it raises several questions in terms of who is at fault should a robot take its own liberty, to cause harm to humans or the environment. For example, a child crosses the road, and instead of the automated car stopping or hitting the emergency brake to avoid colliding with the child, instead, it proceeds driving and bumps the child (without any mistake or programming error, but with deliberate intent).
The key question is: who is at fault here, the robot, the human that has programmed the robot or the company that sells/distributes the robot? Verily (and hopefully), humans would have programmed robots to serve or be of benefit to society. But a robot taking a decision on what to do in that very moment, through independent processes, is the human programmer or company still at fault?
Such an example would need to be governed by robust laws, robotic constitutions, regulations, legislations and intellectual property (including copyright) for the interplay between human and machine/robot. Ultimately, there may come a time where robots could replace certain human activities. However, it is probable that they would still lack the emotional intelligence and adversity quotient that would govern their decisions and activities. Importantly, empathy and human touch remains key to understanding human emotions and behaviour (as well as behavioural change).
From an educational perspective, there has been an overwhelming amount of content regarding advantages and disadvantages of AI (as well as ChatGPT) and the implications it may have on reducing one’s ability to think critically as well as the risks for increased plagiarism and decreased academic integrity. However, just like any other tool, technology or process that evolves in society, there are solutions to mitigate such risks or concerns. For example: diversifying assessment pedagogies to counteract the need for students to plagiarise or optimising academic journal submissions to prevent researchers jeopardising their integrity. On the contrary, instead of putting contingencies in place for such, there has been a plea for educators to also look at this through an optimistic lens, where technology (a recent example being GPT-4) can actually enhance and facilitate the learning process better.
Summary
AI has proven to be a supercharger for humanity, where more time and energy can be freed up for humans, so that they can use it to uniquely solve various problems that require advanced strategy and reduce the likelihood of doing repetitive or redundant daily tasks. AI remains an X-factor for augmenting human expertise, and through my six E’s, it’s X’plained why AI cannot replace human expertise.
This paper was not written by ChatGPT.
Habib Noorbhai is a Professor (Health & Sports Science) and Director of the Biomedical Engineering & Healthcare Technology (BEAHT) Research Centre, at the University of Johannesburg.
The views expressed are those of the author and do not necessarily reflect the official policy or position of the Mail & Guardian.