Introductory Chapter: Artificial Intelligence Will Not Replace Humans – Humans Using Artificial Intelligence Will Replace Those Who Do Not

1. Introduction

We have witnessed an unprecedented technological revolution during the past 3–4 decades [1, 2, 3]. A confluence of highly synergistic factors, including the widespread adoption of the Internet, ongoing advances in microchip technology, the evolution and refinement of “specialty microchips” able to hyper-focus on a specific computational task, blockchain and “internet of things” (IoT), and the growing body of associated theoretical knowledge, all combined to create a perfect environment for the entrance of artificial intelligence (AI) and machine learning (ML) into the mainstream [4, 5, 6, 7, 8]. At this time, the rate of progress is becoming so rapid that our “vantage point” on this remarkable evolutionary process is in itself becoming a “moving target,” where “the way things are today” will not necessarily reflect tomorrow’s reality.

2. The evolving interaction of humans and intelligent machines

As artificial intelligence becomes more than just a “catch phrase” and moves into mainstream applications, human-AI interactions are bound to exponentially increase. In fact, these interactions have the potential to become too intricate and complex for humans to identify when the so-called point of “singularity” is reached, denoting a state of irreversible and ever-increasing “machine dominance” over human intelligence [3, 9, 10]. Concurrently, it is hoped that a substantial amount of new knowledge regarding these previously unexplored interactions can be generated within a reasonably short time frame, effectively facilitating a continuous learning/feedback-loop cycle that will hopefully lead to a much better understanding of the human civilization’s initial “non-human intelligence” encounters [11, 12, 13, 14].

At the same time, as both the use and utility of AI-driven applications continue to increase, important observations and trends emerge. One of the most important trends is the realization that AI can be only as good (e.g., accurate, effective, and useful) as its algorithms and data inputs [15]. Another important observation is that “human-guided AI” maybe more suitable in everyday applications than “autonomous AI,” at least at the current point in time [16]. In addition, it is becoming increasingly apparent that the contemporary set of technological limitations (both related to hardware and software) makes AI-based applications most optimal in the broadly defined area of “productivity enhancement,” where humans equipped with AI assistance can be significantly more productive, with both greater quality and quantity of output [17, 18], and hopefully with an enhanced operator work experience. Within the latter context, autonomous Black Box AI (BBAI), a system whose inputs and operations are not “visible” to the user or other parties, inherently requires human oversight, mainly because of its propensity to “confabulate” output based on nontransparent data processing pathways [19], further strengthening the case for human-guided/supervised AI implementations.

3. Artificial intelligence in healthcare

AI is rapidly transforming healthcare, fundamentally altering how we diagnose, treat, and manage diseases. The convergence of the above-mentioned emerging technologies has created an unprecedented opportunity for AI to enhance healthcare delivery and achieve what is often referred to as the “quadruple aim”: improving patient experience, enhancing population health, reducing costs, and positively transforming the work life of healthcare providers [20, 21].

Currently, AI is being deployed to automate repetitive and time-consuming tasks, particularly in precision diagnostics. For instance, AI algorithms analyze medical images for conditions such as diabetic retinopathy and lung cancer with a level of accuracy that often rivals or exceeds that of human specialists [22], without the associated “human fatigue” and fluctuating performance. These systems not only improve diagnostic accuracy but also significantly reduce the time required for image analysis, thus expediting the overall diagnostic process [23, 24]. Furthermore, AI-powered virtual assistants and large language model (LLM) based “chat bots” maybe used in certain settings to assist patients in managing their symptoms and navigating healthcare systems [25, 26]. These tools, integrated with wearable devices, could provide real-time health monitoring and personalized feedback, enhancing patient engagement and enabling early intervention for various health conditions.

4. The promise of tomorrow

In the future, AI systems will likely evolve to incorporate ambient intelligence, creating a more seamless and intuitive interaction between technology, healthcare professionals, and patients. This will involve the large-scale adoption of precision imaging technologies and the development of AI systems capable of integrating diverse datasets, including electronic health records (EHRs), genomic (and in general other “omics” [27, 28]) information, and real-time patient monitoring data [29, 30, 31]. Such systems will enable more precise and individualized treatment plans, moving us closer to the realization of the promise of precision medicine.

In the field of drug discovery, AI is poised to revolutionize the development of new therapeutics. By analyzing vast amounts of biomedical and known molecular data, AI can identify potential drug targets and predict the efficacy and safety of new compounds more quickly and accurately than traditional methods [32, 33]. This capability is expected to significantly reduce the time and cost associated with bringing new drugs to market, ultimately leading to more effective treatments for a wide range of diseases. It will also supercharge the emerging field of personalized/precision medicine [33, 34].

In the long term, AI has the potential to transform healthcare into a fully integrated, predictive (and thus preventive), and highly personalized system. On can hypothesize that healthcare systems of the future maybe interconnected through a unified digital infrastructure, enabling the continuous monitoring and analysis of patient data [35, 36]. Blockchain may play a prominent role in such a global paradigm [7, 37]. AI-driven predictive analytics will ideally be able to help anticipate health issues before they become critical, allowing for early intervention and more effective management of chronic conditions. Additionally, the development of AI-powered tools such as “digital twins”—virtual replicas of patients—will enable healthcare professionals to simulate and evaluate the impact of different treatment options before exposing patients to any undue risks, thus ensuring the most effective, highly optimized, and truly personalized care [38].

Finally, AI will play a crucial role in addressing some of the most significant challenges facing health systems today, including the shortage of healthcare professionals and the increasing complexity of medical knowledge. By automating routine tasks and augmenting the capabilities of healthcare practitioners, AI will help to alleviate the burden on healthcare facilities and providers, and ensure that high-quality care is accessible to all. In addition, adaptively evolving care guidelines will provide frontline healthcare staff with the most effective and up-to-date information without delays and/or potential “expert consensus” biases [39, 40].

5. Synthesis and conclusion

As artificial intelligence becomes more prevalent and universally accepted, and as it increasingly penetrates into more domains of our lives, it will be critically important to identify, categorize, develop, and optimize various “utility-implementation” pairs applicable to specific use-case scenarios. In the near future, the most likely benefit and use-case for AI-based applications will be the broadly understood “increase in productivity” enjoyed by humans utilizing AI in their private and professional lives. Beyond this immediate period, it is likely that our understanding of AI and its risks, benefits, and alternatives will guide more autonomous applications of this transformational technology across selected functional areas and sectors of our economy. This may include more liberal applications of AI/ML in healthcare, where specific benefits may manifest as greater clinical throughput, increased clinical accuracy and efficacy, whilst significantly increasing provider job satisfaction by reducing administrative burdens and various nonclinical, minimally productive, and/or counterproductive activities (e.g., excessive time spent completing medical records or extended wait times to receive diagnostic results). Any such advancements and implementations will need to be very carefully supervised and closely monitored, with strong provisions and hard stops built into the overall implementation system to ensure continued focus on the quality, safety, and ultimately value of care provided.