Overview
The Dawn of General Intelligence?
The rapid advancements in artificial intelligence (AI) over the past decade have moved beyond narrow, task-specific applications. We are now witnessing significant progress in developing systems with more general cognitive abilities, often referred to as Artificial General Intelligence (AGI). While the precise definition and timeline for achieving AGI remain subjects of debate, the potential implications of such systems are profound and demand careful ethical consideration, especially from those who will be building and deploying these technologies.
The idea of technological convergence, where seemingly unrelated technologies intersect to create something new, is a key concept in understanding the trajectory towards AGI. This phenomenon has been observed throughout the history of computing. One example is the development of the smartphone, which merged various distinct technologies, including voice telephony, text messaging, mobile Internet access, photography, and entertainment into a single device. In a similar vein, AGI represents a convergence of advanced hardware processing capabilities, highly refined machine learning, classification, and inference architectures, and the vast amounts of human-generated information available on the Internet.
Technological emergence is the sudden and rapid growth of a new technology from roots in diverse and often disparate lines of development. This new emergent technology is not easily predictable from its sources and often comes from a novel and innovative application of unrelated technologies. One example is the emergence of Bluetooth(opens in new window) data communications for computing-based devices. These technologies had their origins in short-range radio communications developed during World War II as a possible secure battlefield communication tool, but that remained mostly dormant for decades. When the development of truly portable computers (laptops) and the broader availability of cellular telephony created a need for data connectivity without cables, the old work in short-range, high-frequency radio was resurrected and developed into the now ubiquitous Bluetooth data communications used in many devices today.
While there isn't a single universally agreed-upon definition of AGI, Goertzel(opens in new window) asserts that it generally refers to AI with human-like cognitive abilities – the ability to learn, reason, and understand across a wide range of tasks and contexts [1]. The Google DeepMind classification scale(opens in new window) provides a useful framework for understanding the potential capabilities of AGI. Systems reaching the "expert" level can perform tasks at or above the level of a human expert in a specific domain. More significantly, the prospect of AGI achieving "collaborator," "expert," or even "agent" levels in interactions with humans and other AGIs raises a host of novel ethical challenges. A "collaborator" AGI might work alongside human professionals in complex decision-making, while an "expert" AGI could independently solve intricate problems. An "agent" AGI could autonomously pursue goals in complex environments [2].
While the development of such advanced AGI is still largely in the research phase, the ethical considerations are not merely theoretical. The decisions made by researchers and developers today regarding the architecture, training data, goals, and control mechanisms of these early-stage systems will have a significant impact on the trajectory of AGI development and its future societal consequences. As future computing professionals, particularly those in the early stages of their careers, you will likely encounter these issues firsthand, whether you are directly involved in AGI research or working on applications that leverage increasingly sophisticated AI.
This forum will explore critical ethical questions surrounding the development and benefits of AGI, including accelerating scientific discovery and solving global challenges. These benefits may come with potential risks, including job displacement, algorithmic bias at an unprecedented scale, and the potential for concentration of power. It is also important to remember that the field of AGI is rapidly evolving; what is seen as merely conjecture today may be a prototype implementation next week and a commercial product in just a few months.
[1] Goertzel, B. (2007). Artificial general intelligence: Concept, state of the art, and future prospects. Journal of Experimental & Theoretical Artificial Intelligence, 19(3-4), 247-292, Available at https://intapi.sciendo.com/pdf/10.2478/jagi-2014-0001(opens in new window).
[2] Morris, M. R., Sohl-Dickstein, J., Fiedel, N., Warkentin, T., Dafoe, A., Faust, A., ... & Legg, S. (2023). Levels of AGI for Operationalizing Progress on the Path to AGI. Available at https://arxiv.org/abs/2311.02462(opens in new window).
Immanuel Kant's Humanity (or second) Formulation of the Categorical Imperative states:
"So act that you use humanity, whether in your own person or in the person of any other, always at the same time as an end, never merely as a means." [1, p.80]
Johnson and Cureton note that this formulation does not completely rule out using people as a means to an end, only that it rules out "engaging in this pervasive use of humanity in such a way that we treat it as a mere means to our ends." [2] They also point out that the "humanity" Kant refers to is not necessarily individual human beings, but "collection of features that make us distinctively human, and these include capacities to engage in self-directed rational behavior and to adopt and pursue our own ends, and any other rational capacities necessarily connected with these." Johnson and Cureton use an example of a horse and a taxi driver as an explanatory analogy, noting that the difference between the two is not whether we may use one and not the other as a means of transportation, but that we must take into consideration the humanity of the taxi driver, particularly the rational capacity to pursue a line of work that involves receiving compensation for providing transportation to others.
For this forum, consider the development of AGI itself. Among the potential risks of such development are the possibility of an AGI replacing human labor without providing an alternative opportunity for human workers or the possibility of an advanced AGI interpreting humanity, as a whole, as a means to an end that might involve significant risks to human life. As AGI systems become more sophisticated, the question of their moral status – whether they deserve ethical consideration similar to humans or animals – becomes increasingly relevant. This involves exploring what qualities confer moral standing and whether advanced AGIs might possess those qualities.
Considering the potential risks to humanity, some might suggest that all efforts to develop AGI should be stopped permanently. However, if we consider the possibility that an AGI might achieve consciousness and moral agency, would the action of halting development be treating the potential for such a future AGI as a means to the end of avoiding the risks to current humanity? Clearly answer this question in the affirmative or negative and discuss the ethical responsibility of developers to consider the long-term consequences of preventing the development of another form of moral agent.
Your initial post must clearly state your response to the given prompt and should incorporate a clear and convincing ethical argument for your stated position that is grounded in an appropriate ethical framework (e.g., utilitarianism, virtue ethics, care ethics). Your argument should provide the reader with verifiable, real world evidence in support of the important aspects of your position on the question. The best responses will anticipate and address a strong objection to your position from a different ethical framework. Note that professional codes of ethics are not evidence of anything other than an organization's aspirations for its members, and are generally not considered strong evidence for anything more than that. Be sure your primary focus is on the ethical dimension, rather than legal, contractual, or similar justifications.