General artificial intelligence is a branch of AI whose subject of study is the hypothetical development of an artificial intelligence system that matches or exceeds human cognitive abilities, particularly with regard to autonomous learning. Many AI engineers maintain that it is currently impossible to create a system with these characteristics. Meanwhile, others cannot agree on which cognitive abilities should be measured to define the intelligence of an AI.
It is worth noting that, whilst the development of AI in recent years has been exceptional, no one could claim that an artificial intelligence system possesses its own intelligence that exceeds that of humans. AI outperforms humans in terms of speed and efficiency in tasks that lend themselves to automation, but it lacks the capacity to make decisions of its own accord beyond what it is taught and the data it is trained on.
Furthermore, the concept of general artificial intelligence remains an abstraction with no real-world counterpart, which is why many experts may interpret it differently, with predictions ranging from the most ambitious to the most cautious, depending on the circumstances.
Benefits of an AGI
The benefits of AGI and its real-world applications are the same as those of implementing AI, but they go one step further. Currently, AI is used to help streamline repetitive tasks and to make certain decisions. An AGI that surpasses human cognitive capacity would not be a support, but rather another actor that he would be capable of making independent decisions and developing his own logical arguments.
General artificial intelligence could bring the automation of much more complex tasks and entire production processes, including tasks that we regard as more strategic.
Furthermore, governments and other organisations They could use it to address highly complex issues and matters of global concern relating to poverty, climate change or the likely collapse of a healthcare system on the brink of bankruptcy.
However, it would not be all advantages, as artificial intelligence still has to address a number of ethical dilemmas problems of their own, which the emergence of the AGI would only serve to exacerbate.
AGI (general artificial intelligence) vs. ANI (narrow artificial intelligence)
As a general rule, general artificial intelligence is conceptually opposed to what would be known as narrow intelligence. Narrow intelligence refers to systems trained on a specialised dataset, designed for more specific tasks, hyper-specialised. However, general artificial intelligence, as a supercomputer Compared to the human brain, it would cover a much broader range of topics.
Here is a brief overview of the differences between general and narrow artificial intelligence:
Narrow artificial intelligence
- It focuses on performing specific tasks very well, such as playing chess, translating languages or recognising images.
- You cannot carry out tasks outside your area of expertise.
- He’s like a specialist in a single subject.
General artificial intelligence
- It has the ability to understand, learn and solve a wide range of problems, in much the same way as a human being would.
- It can adapt to different situations and learn new skills without being specifically programmed for each task.
- It’s like a versatile, multifunctional person who can be used for a wide variety of purposes.
Is ChatGPT an AGI?
Since its launch, ChatGPT has received all manner of praise from technology experts and the general public alike. The praise ranges from “an astonishing breakthrough” to hyperbolic descriptions such as “a legendary revolution that will wipe out entire business departments”. Some might even go so far as to think that AGI has already become a reality with ChatGPT. After all, it can write an essay on any topic and its knowledge seems boundless, doesn’t it?
Well, no, ChatGPT cannot be regarded as an AGI. We must remember that ChatGPT does not possess intelligence of its own, but merely appears to do so. Its scope of action and decision-making is limited to the vast amount of data on which it has been trained. It is undeniable that a large language model of this kind represents an unparalleled advance in the world of AI and new technologies, but it neither replaces nor, by any means, surpasses human intelligence, our judgement and our decision-making ability.
For an AGI to be considered as such, it would need to have the capacity for autonomous learning, which in turn would be based on three pillars: an inner mental awareness of the self and its surroundings, temporal perception and one’s own imagination. If we tinker enough with any AI system right now, we would come to the same conclusion: AI is not as intelligent as humans in any sense.
The future of general artificial intelligence
Firstly, is it possible to create an AGI as we understand it today? There is no unanimous answer. For example, Google’s head of engineering, Ray Kurzweil, believes that artificial intelligence could beat the Turing test once and for all and reach a level comparable to that of humans by 2030, and estimates that it could even surpass our capabilities by 2045. If these estimates prove accurate, in a couple of decades we could be talking about the existence of general artificial intelligence.
However, Other experts claim that creating such a system is impossible. We’d be talking about machines developing a consciousness of their own, and all the futuristic scenarios and dystopias we’ve seen in films coming true. It would be straight out of *Black Mirror*.
At Euroinnova, we do not rule out any possibility. However, from our point of view, we believe that artificial intelligence, as a human invention, will always be limited to the dataset on which it is trained, however extensive and general that may be. The creation cannot surpass its creator – at least not for the time being!
Español 
English (UK) 
Français