All of us, in one way or another, have witnessed how the impressive advances of artificial intelligence have replaced some, if not many, of the complex everyday tasks of human beings, as well as discipline-specific tasks in fields such as medicine, engineering, construction, communications, among others. Although the progress is obvious, the question still arises: Why are simple tasks difficult for robots, but easy for humans? The answer to this question is found in the Moravec paradox, the protagonist of this article.
What is Moravec’s paradox and when did it originate?
Although current advances have shown us that artificial intelligence could surpass us in different fields and even generate skills unknown to us, it is very rarely capable of copying the simplest or most ordinary skills of human intelligence; This is due to the famous Moravec paradox, which is believed to be the key why it has not been possible to build a brilliant robotic system.
Broadly speaking, this paradox states that both artificial intelligence and robotics may require fewer computational processes for reasoned thought. However, for simpler methods – such as picking something up from the floor or tying your shoes – the computational effort could be immense. So, Moravec’s paradox mentions:
It is relatively easy to get computers to show abilities similar to those of an adult human in an intelligence test or when playing checkers, and very difficult to get them to acquire the perceptual and motor skills of a one-year-old baby .
The argument on which this statement works points out that human beings when developing artificial intelligence, apply a series of reverse engineering, which is complemented by the consideration that the effort required to copy any skill ends up being directly proportional to age. of its appearance in human genealogy.
The origin of Moravec’s paradox
Some of the characteristics of Moravec’s paradox have already been defined, but who created it? This aforementioned duality of computers gets its name from the surname of its creator: the Austrian engineer Hans Moravec. In addition to the fact that throughout his career he presented important advances in artificial vision techniques and how technology impacts the society in which it is inserted, one of his most important research was related to this paradox.
The formulation of Moravec’s paradox occurred in the second half of the eighties and with the participation of other specialists in the field, among whom it is worth mentioning Rodney Brooks – the creator of iRobot – and Marvin Minsky – co-founder of the Laboratory of Artificial Intelligence from the Massachusetts Institute of Technology─.
High and low-level abilities in Moravec’s paradox
Now, another of the key premises of Moravec’s paradox is the existence of a hierarchy of intelligence that functions like an inverted pyramid. At the top of this pyramid are high-level skills ─linked to solving equations and strategy─; in the middle, intermediate skills ─such as natural language─ and at the base, primitive skills ─such as picking up things or walking─. In this section, we will talk about the first and the last.
High-level skills: Why do machines excel?
As mentioned above, at the top of the pyramid, which gives meaning to the hierarchy of intelligence, are high-level skills, those in which artificial intelligence and robotics excel. These skills are linked to solving equations, mathematical problems, or creating strategies; But why does this happen?
To find the answer, we must appeal to the fact that artificial intelligence systems are made up of processing areas ( such as machine learning ) implemented through programming. This allows you to use algorithms, neural networks, and other technological tools to simulate intellectual and reasoning processes.
Now, it is worth highlighting that this is achieved because they are skills that, due to their complexity, man has sought to develop fully consciously, so he understands their functioning and what is required for them to occur properly; Therefore, he can transmit it to AI systems.
Low-level skills: Why do machines struggle?
In the case of low-level skills or those that are more primitive – visual recognition, locomotion, tactile perception, flexibility, among others – artificial intelligence faces greater difficulty, because they are completely intrinsic to human beings, that is, they are thousands of years old. ancient and have evolved at the same pace as man. Furthermore, since they are such ordinary skills, many of them are performed without being fully aware of how they are carried out, making it even more complex to transmit these skills to a machine.
How to overcome Moravec’s paradox?
Despite the above and the high relevance that the Moravec paradox had at the time of its appearance and that it has returned to in recent years, it is important to highlight that it was developed at a time when artificial intelligence and robotics were not presented advances that were so promising, or at least so frequent. However, over time and technological innovation, new measures have been sought, identified, and implemented to make smart machines work efficiently in the real world.
Thus, among the most relevant actions to confront the paradox is the implementation of an interdisciplinary approach that not only integrates artificial intelligence tools but also works hand in hand with the main advances in neuroscience and engineering.
The above and thanks to the different scientific research that surrounds the field, AI and robotics are supported by deep learning algorithms and neural networks that allow them to better understand the context and the subtleties that the human being. able to notice This is because, by combining data from different types of sensors – visual, tactile, auditory, among others – intelligence is allowed to somehow assimilate how human beings use their senses. And this is the most viable way to overcome Moravec’s paradox.
Future of AI and Moravec’s paradox
There is no doubt that the future of artificial intelligence is completely related to the expansion of this technology to different areas of human daily life. This means that not only robots with artificial intelligence that are capable of solving a mathematical problem or winning a chess championship are required, but healthcare or domestic robots, capable of performing high-precision surgeries, are increasingly common. or entertainment artificial intelligence that creates completely interactive experiences with humans. Therefore, new perspectives are essential.
New perspectives in AI research
So, currently, research for its development focuses on data processing, analysis, and decision-making, which should eventually lead to the development of at least an essential ability to understand the environment and act. consequently, without relying solely on statistical predictions.
Compared to the above and as mentioned in previous sections, much more advanced sensors have been developed that provide detailed and contextual information about the environment. These sensors work through reinforcement learning techniques that allow machines to learn through experimentation and feedback, just as humans do when they develop cognitive and motor skills.
In this way, those intuition capacities are promoted, so typical of the human being, and that are linked to abstract conditions such as creativity, ingenuity, and insight, among others.
Future challenges of AI and Moravec’s paradox
One of the main challenges of current artificial intelligence, and what will surely be the same challenge in the future, is that these systems are not replacements, which would imply the non-existence or near possibility of general artificial intelligence systems.
Despite great technological developments, current AI systems remain fundamentally pattern recognition systems, trained from the inference of different types of rules. However, this does not mean that AI is or will be able to understand the world around it, as we can. The above is revealed, for example, in the few steps taken towards the achievement of the so-called artificial empathy, which resembles human empathy in very few ways.
Furthermore, it should be mentioned that, according to the above, Moravec’s paradox has not changed much. The skills developed on a large scale in the field have not surpassed anything other than language, which is synthetic and, for many, “easy” to approach. Evolution, then, continues and will continue to be the biggest problem when thinking about the creation of a fully intelligent artificial intelligence, despite the redundancy.
Final words
Although Moravec’s paradox continues to exist and exponentially impact this technological field, the continuous evolution of AI and robotics cannot be ignored. In this way, the paradox overcomes the barriers of being only a technical enigma, to act as a continuous reflection on the very nature of intelligence and the limitations or challenges that machines could pose for it.
This is how we end our article! But we don’t want to leave without asking you: did you already know about Moravec’s paradox? What do you think is the most interesting thing about her? We read you!
