Artificial General Intelligence: What Is Its Potential?

In December 1955, Herbert Simon and Allen Newell, two American computer science researchers, created the first running artificial intelligence (AI). The program, the Logic Theorist, was an AI model that proved mathematical theorems using a heuristic deductive methodology that worked backwards and made substitutions. Since then, AI has advanced significantly and become a bigger part of society. AI can write music mimicking certain composers, make investments, code complex interactive 3D simulations in minutes. But how does it work?

Currently, most AI programs function using processing algorithms to analyze enormous amounts of data. They examine information patterns and learn from them. AI can study large amounts of data and carry out complex tasks in a short amount of time, allowing it to advance quickly.

However, one of AI’s biggest weaknesses is its lack of common sense. AI models do not have any understanding of the responses they generate. They reach conclusions via pattern recognition. Models use data to predict what response is most probable to be “correct.” For instance, AI models that create music understand neither the lyrics they write nor the notes they produce. Instead, they form melodies using probability to predict which notes and lyrics are most likely to come next according to past music compositions.

AI’s ignorance has caused many issues to arise; models may provide illogical responses or incorrect information. This lack of understanding draws a clear distinction between human and artificial intelligence. However, Artificial General Intelligence (AGI) may blur this distinction.

AGI is a futuristic and theoretical model of AI that can think, learn, behave, and, most importantly, understand concepts as humans can. AGI would consume data as they are discovered instead of drawing from fixed databases as current models do, produce original ideas, and match or exceed human capabilities like reasoning and content creation in all subjects. Although achieving AGI is the end goal of many leading AI companies such as OpenAI, creator of ChatGPT, and Deepmind, creator of Genie, scientists are far from creating a working model of it. In fact, while some scientists like OpenAI co-founder Elon Musk believe that AGI can be achieved in the next few decades, others such as Microsoft co-founder Paul Allen don’t believe it is attainable at all.

AGI’s biggest limiting factor is the extreme amount of computational power needed to develop it. Currently, data scientists estimate AGI would require 10¹⁶ teraflops of computing power, far beyond what our current world can sustain. In comparison, GPT-4 requires just 560 teraflops.

Even if AGI is possible, there is still debate about whether developing it would be ethical. Advocates argue that AGI would aid humans in finding working solutions to the world’s biggest problems, such as climate change and poverty, emphasizing that AGI could solve problems humans currently cannot. They highlight how AGI can quicken medical advancement, making diagnostics more accurate and treatments more effective. Supporters also bring up the possibility for AGI to bridge learning gaps and provide students with additional support.

On the other hand, there are several potential dangers to AGI. Historically, powerful technological outbreaks tend to be controlled by wealthy people. This pattern creates worry among the general population that AGI will not improve the quality of life of all people, but will instead focus on benefiting those at the top while widening the gap between socioeconomic classes. Contrary to the pro-AGI argument about improving efficiency, those against AGI believe that it will not improve productivity but instead cause large numbers of people of lower socioeconomic standing to lose their jobs. Additionally, since AGI would be able to think and innovate like humans, it could devise plans that could either intentionally or unintentionally harm people. In proposing regulatory oversight over artificial intelligence, Musk states, “[W]ith artificial intelligence we’re summoning the demon.”

Another ethical dilemma is about whether or not AGI deserves rights. Some people believe that since AI is artificial, it should never be given human rights. Others believe that with AGI, there would be no difference between humans and AI except the body they are held in. Gray Scott, techno-philosopher, futurist, and expert in emerging technology, says, “The real question is, when will we draft an artificial intelligence bill of rights? What will that consist of? And who will get to decide that?” Despite this controversy, scientists continue to make advancements towards a working model of AGI. What is the future of AGI? Only time will tell.