The race in artificial intelligence has generated expectations and concerns that this technology will be able to reach human cognitive abilities (AGI). How far are we from that moment? Should we be afraid?
Rapid developments in artificial intelligence have made countless headlines this year thanks to the popular chatbot ChatGPT, search engines releasing their own AI tools, and people using some apps on a daily basis to create images, audio, and video from scratch and in seconds.
But some companies and experts in this field are providing more and more more attention to what is known as artificial general intelligence or AGIwhich some hope will herald the arrival of systems that are as intelligent as humans.
It is an ambitious goal long considered with a mixture of fear and very positive expectations due to a possible danger of it getting out of hand. Experts in the field of AI have different opinions about how close it is to achieving it. Some believe that this is a matter of a few years, while others think that it could take several decades.
Perhaps the key here lies in the very process of this technology. If breakthroughs continue to happen at such a rapid pace, it’s likely to come sooner than you think.
However, it appears that there is currently a general madness about its dangers, which although they exist, would be overstated. Some even go so far as to say that the GPT-5 version of OpenAI will be artificial general intelligence, or AGI.
It is true that a key concern is the displacement of human workers as machines become capable of performing the same tasks. Another important issue to consider is the potential misuse or unintended consequences of this evolution, especially if it falls into the wrong hands.
“There is both enthusiasm and concern around AI. Significant efforts are being made in research and development to maximize the benefits of AI and minimize its drawbacks. The direction it takes in the future will depend on how these challenges are managed and of how technology is applied in different sectors of the economy”, Josué Pérez Suay, specialist in Artificial Intelligence and ChatGPT, explains to Computer Hoy.
The real complications of reaching AGI, what is there really to be afraid of?
While technological advancement can provide great benefits for humanity, it is essential to ensure that it aligns with values and goals so that everyone can benefit from its advances without compromising a shared future.
As explained before, AGI applications are supposed to replicate human consciousness and cognitive abilities seamlessly. However, the developers of this technology cannot design neural networks that can recreate artificial consciousness in a laboratory, at least not yet. Human consciousness is too abstract and asymmetrical.
Another problem that can be observed is that AI, for all its incredible capabilities, still runs on carefully crafted algorithms and computer programs that, on the one hand, they can fail and stop working as it happens with the rest of the tools and, on the other hand, each decision is supported by analysis and processing of human data.
Here the fear should really lie in how certain governments or security bodies or even certain companies make use of this new technology rather than whether it is capable of reaching human cognitive abilities.
“Some people, the military, independent state actors, terrorist groups, will want to marry that advanced technology with warfare capability. And that’s not something we want to do without thinking very, very carefully,” explains Christof Koch, chief scientist at the Mindscope program at the Allen Institute in Seattle, in an interview for IEEE Spectrum.
As a last resort, the road to AGI will require great technological advances and ethical considerations — and good regulation — as progress nears the creation of systems with human-like intelligence.
“It’s all a matter of perspective. On the one hand, AI has incredible potential to boost productivity and efficiency, and to unlock new economic opportunities. On the other hand, there are legitimate risks and challenges that need to be addressed, and it’s important that we are aware of them and that we take measures to mitigate them”, adds the expert interviewed.
Simply put, and while progress has been made in AI research, achieving AGI is still a truly complex feat due to issues such as data limitations, hardware restrictions, and ethical considerations. However, experts are optimistic about its potential to revolutionize the industry and change the world.
Whether or not it is just around the corner remains to be seen. “AGI remains a long-term goal and many more advances in various aspects of AI are likely to be needed to achieve it,” concludes Josué Pérez Suay.