Artificial intelligence (AI) has recently achieved a major breakthrough by surpassing a fundamental human capability. This advancement begs the question: Will machines eventually reach and possibly exceed human intelligence?
Human intelligence encompasses a variety of skills, one of which is the capacity to learn the meaning of a word and apply it to other linguistic concepts. We possess the ability to abstract ideas and identify objects based on their shapes, regardless of their color or material composition. Even when observing cloud formations, we can discern patterns and draw meaningful conclusions.
This aspect of human intelligence is known as composite generalization. Cognitive scientists Jerry Fodor and Zenon Pylyshyn previously proposed that artificial neural networks could acquire this ability; however, progress in this field has been limited since the 1980s.
Researchers from New York University and Pompeu Fabra University in Spain have been diligently working in this area for some time and have recently developed a new technique aimed precisely at achieving this capacity.
Their study, which was published in the prestigious scientific journal Nature, reveals how ChatGPT-like tools can achieve compositional generalizations through a technique called “Meta-aprendizado para composicionalidade” (MLC).
To explore the effectiveness of this technique, extensive field tests were conducted. The experiments demonstrated that artificial intelligence has not only achieved parity with human intelligence but even surpassed it in some cases.
This remarkable feat was accomplished through experiential learning rather than traditional learning methods. The system is presented with a word and then prompted to apply it in a different context. For example, the AI system is given the word “falar” (to speak) and instructed to generate contexts such as “falar muito” (to speak a lot), “falar pouco” (to speak a little), “falar baixo” (to speak softly), and “falar alto” (to speak loudly).
As AI continues to advance, it will acquire the understanding of idiomatic expressions like “falar abobrinha” (to speak nonsense) and “falar besteira” (to talk rubbish) in both their literal and figurative senses. This capability will enable AI-powered language systems to cater to a much broader audience.
These recent developments hold immense promise for the field of programming. With the ability to comprehend more complex commands, computers will be able to receive, understand, and execute intricate instructions with precision and accuracy.
In conclusion, artificial intelligence has made significant strides by surpassing human intelligence in certain domains. With ongoing research and advancements, it is increasingly plausible that machines will continue to evolve and potentially outperform humans in various cognitive capabilities. The prospect of machines reaching or even surpassing human intelligence sparks both excitement and skepticism, but it remains an intriguing area of exploration for researchers around the world.
