Artificial Intelligence has been the hype of the modern world of technology for many reasons. One of which, is because it's able to be trained to become smarter.
But the thing is, the term is still confusing to some people. And making things worse, the word AI often comes with even more terms that make it even more difficult for the general public to understand.
Artificial General Intelligence (AGI), for example.
The term is often used to describe hypothetical machines that can perform any intellectual task that a human can. But before humans can ever create an AGI, there’s no consensus on what the word really means. Some AI experts believe that the term is a misnomer, while others argue that the technology could never be realized.
As a result, the terms are not only confusing, as they can also mislead people.
But according to Elizabeth Bramson-Boudreau, the CEO and publisher of MIT Technology Review, the better term to describe an AGI, is a "multi-skilled AI."
This term should make an AGI a more comprehensible concept.
Since the beginning researchers worked on AI, they have longed understand that AI is only as good as the data it has been trained on.
But no matter how good the training materials are, as long as the AI is an Artificial Narrow Intelligence (ANI), the AI would remain the "dumbest" and the least capable. This is because the AI can only do one task.
And here, speaking to The Next Web, Bramson-Boudreau suggested that to turn an ANI to an AGI, or also known as a Strong AI, a multi-skilled approach is to improve the technology by expanding its senses.
Multi-skilled AI systems should be able to combine senses and language to broaden their understanding of the world.
This is similar to how children are taught to learn through perception and talking.
“It goes beyond image or language recognition and allows multiple tasks to be done,” she said.
For example, an AI system would learn in multiple modalities, such as the textual and visual domains, to expand their capabilities. Robots, for example, could integrate visual, audio, and tactile data to execute a wider range of tasks.
As a result of this, theoretically, a more flexible intelligence technology could be created.
She prefers to call an AGI a multi-skilled AI because the AI learns through observation and imitation like humans.
According to her, the name is more in line with reality.
For her, this approach is much more realistic and offers security for researchers and the discipline.
She then continued to explain that the developments of AI have progressed rapidly, and a number of projects have shown some potentials.
But there are risks.
While the idea of having a multi-skilled AI that can learn through general observation and not just knowledge of previous databases or external networks can certainly help humanity to become a lot smarter too, Bramson-Boudreau is concerned about its susceptibility to data biases, environmental impact, and potential use in autonomous warfare.
But still, she’s optimistic that the dangers can be addressed, as long as they can be foreseen so any potential negative consequences can be prevented from ever happening.