Neuroscientists at Trinity College say that the babies can help unlock the next generation of artificial intelligence (AI). The article, published in the scientific journal Nature Machine Intelligence, describes the principles of how babies absorb information and how it can be replicated for application in AI. So, check out more information on how a baby's learning can advance artificial intelligence!
Read more: Google AI 'Child' May Slip Away And Do Bad Things, Insider Claims
see more
Replaced by ChatGPT at work, woman spends three months…
Towards artificial intelligence: Apple plans to integrate chatbot in…
The advancement in AI machines
Trinity College scientist Lorijn Zaadnoordijk explains that all the amazing developments in artificial intelligence have been performed because of machine learning that uses a large amount of data to train models of neural networks artificial.
However, the scientist went on to say that progress in many areas has been slowed down as the foundations of data for the improvement of machine learning needs to be managed and fed by minds human.
However, with his theory, he claims that learning can be done more efficiently, since this is not how babies learn. This happens as babies experience the world around them, sometimes even seeing it once.
Three important factors for an AI to have quality
The article identifies three crucial factors for an AI to achieve the quality and speed of a baby's learning.
- The first is that children's information processing is guided and limited;
- The second is that they learn through diverse and multimodal inputs;
- Finally, babies' input is shaped by development and active learning.
The idea of the research is to explore which concepts have not yet been properly applied in the development of artificial intelligence and improve them to produce a system that can learn without being supervised.
Implement the child learning process
According to the study, to apply children's learning processes in AI, they need to establish their preferences from the beginning to be able to shape learning.
They also need to be provided with richer data that represents the world, not just images and spreadsheets. Thus, it is important to understand what the surroundings, sounds, smells and taste of things are like.
Finally, machines also require development trajectories defined by researchers. Just as children experience different stimuli over time, it is important to replicate this behavior by giving computers different experiences and networks as they “grow up”.