×
The Elusive Dream of Artificial General Intelligence/ai-insights/the-elusive-dream-of-artificial-general-intelligence

The Elusive Dream of Artificial General Intelligence

Jul 31, 2024

The Elusive Dream of Artificial General Intelligence

Throughout human history, there have been watershed moments that have fundamentally altered not only our society at large, but also have changed the very essence of our species’ trajectory in the development of our civilization. From hunter gatherers, we are now hackers and Artificial Intelligence and deep learning researchers. One can actually count the inflection points in the annals of history and see how we progressed, sometimes slowly and sometimes rapidly, into the era of information technology that we live in today. But we weren’t stopping - pushing the boundaries until we taught a computer how to think. Then we came to rely on them for specific tasks that required intelligence and trained them with data. Today, there are companies whose ambitions are to ingest even more data to make their AI models as accurate as possible. But not just for the current generation of bulky LLMs or showcase robotics. They are after, what has been called the Holy Grail of Artificial Intelligence – Artificial General Intelligence (AGI).

If It Is General, Why Is It So Elusive and Important?

This is probably the most asked question all over the world right now. The ‘General’ denotes AI models that are all encompassing in their intelligence. Artificial General Intelligence is expected to usher in a new era of possibilities and potentially new discoveries. The “General” in AGI means that it will mimic the human brain at the most granular level, including sensory inputs to compute, and form an exact replica of our brains. The current speculation is that AGI will surpass all human knowledge and compute exponentially within a minimal time frame, all of this at a faster pace than we can fathom as of now.

The human brain is by far the most complex piece of engineering, courtesy mother nature. The goal, or at least the definition of AGI is to mimic the same structure, functioning, analytical capabilities and awareness of the human conscience, within a device, (at this point we don’t know if it will be a device or an application framework) compute power and the level of intelligence equal to the highest intelligence parameters of human or even current AI technologies, and, eventually, surpass it (of course!). This may sound simple at first, but the terminology doesn’t accurately convey how the world would change if AGI is achieved. Artificial General Intelligence will surpass all other modern inventions by actually not restricting itself to the data and the programs that have built it, but its own cognitive capabilities. As Nick Bostrom, director of the Future of Humanity Institute, notes, "AGI would be a machine that can outperform humans in almost every cognitive task" (Bostrom, 2014).

Dear reader, its time ponder upon the consequences of that statement for a while. From the homo sapiens that eliminated the Neanderthals and invented fire, it has been the human brain that has been at the forefront of almost every development in society. AGI is expected to have the same capabilities, only in multiples and many times over. The implications of this are beyond what we can fathom at this point. According to experts, this will transcend human intelligence and cognition in every aspect at levels that we cannot, again, fathom right now.

But How Will AGI work?

Much like the human brain it seeks to emulate, at its core, AGI will comprise a robust and technical framework of interconnected artificial neural networks which will be multi modal and be capable of integrating various forms of information processing including learning and decision-making. Drawing closely from its human counterpart, its core will be a complex set of vastly complex neural networks that will easily surpass the narrow AI models prevalent in the current Large Language Models (LLMs)  and AI agents that we see today. AGI will think like the human brain and will incorporate several elements of deep learning, capable of extracting high levels of abstract outputs from raw data without any intervention, in real time. Unlike the AI models today which rely heavily on highly processed data, AGI will ingest data in real time, an important differentiator in the world of Artificial Intelligence. Another aspect to note is that AGI will need computing power not available yet. Therefore, researchers are using the power of quantum computing, with multiple processes running simultaneously, on their path towards achieving AGI.

Where do we stand now with AGI?

The quest to achieve AGI has been a long and complex process, with mostly dead ends and false starts. The current stage of AGI research has confronted data scientists both with a complex set of challenges and quite some progress. While we have not achieved AGI yet, companies across the world are heavily investing in AI as a path to progress to achieve AGI. We can infer how much companies are looking forward to achieving AGI, from that statement. The ability to learn in real time, not just ingested data, using unsupervised learning, reinforcement learning and more advanced meta level learning will be essential on the path to achieving AGI. According to data scientists, AGI will require computing power of more than 10˄16 teraflops of computing power to function, making today’s AI models’ computing seem like child’s play.  

Important Statistics

It is difficult to determine an exact timeline to achieve AGI, but some actual statistics shed light on how earnestly companies are pursuing this technology. Consider these facts:

  • A survey of AI researchers suggests a 50% probability of AGI being achieved by 2060, with a 10% chance of it occurring as early as 2030.
  • The AGI job market is projected to grow by 72% in the next five years, creating an estimated 1.5 million new positions globally.
  • 63% of Fortune 500 companies have established dedicated AGI research divisions or partnerships.
  • The number of academic papers published on AGI-related topics has increased by 300% over the past decade.

In Conclusion

In conclusion, AGI remains the elusive dream of computer science, a prospect both thrilling and terrifying at the same time. As we continue to push the boundaries of what is possible, it's crucial that we do so with caution and foresight, and with skilled professionals at the front lines. So, begin your journey today, with a professional certification in AI and you might just be the one to create the next great leap forward. Take up a Generative AI certifications, like the prestigious USAII programs and join the ranks of those who are shaping the very future of human civilization.