Can deep learning ever make AI human? Science fiction has been suggesting a lot of material for our imagination for centuries, but today - in part - there is less and less to imagine and more to discover.
Before answering the question then let's find out what are neural networks. Passing through the machine learning algorithm traditional, we understand how deep learning differs and why it is the key to training AI to develop capabilities increasingly similar to those of the human brain.
Content index
Neural networks, machine learning and deep learning
Let's contextualize a bit.
- Le neural networks are models designed to simulate the behavior of the human brain, they therefore process information in a similar way to how our neurons do. To find out more go to the article dedicated to neural networks.
- Il machine learning algorithm it allows AI to learn and evolve with experience, without being explicitly programmed.
- Il deep learning is a subset of machine learning that uses deep neural networks, enabling applications such as machine translation, image recognition and natural language processing.
How an AI learns
In the context of deep learning, AI learns through a process of data-driven training. During training, the neural network is exposed to a huge amount of data. So learn to recognize patterns and make predictions.
As aforementioned, learning happens via deep neural networks composed of many layers of artificial neurons connected to each other.
Each of these connections has a "weight" which determines theimportance of information that passes through it. During training, the model optimizes these weights, progressively reducing the error between the predictions made and the expected results.
Deep neural networks then they learn to extract and combine increasingly complex features from data.
In the first layers, the model recognizes simple patterns. As it advances through the layers, the AI combines this information to recognize more abstract concepts. For example, in an image recognition system, early layers might identify edges and colors, while later layers can recognize complex shapes and objects.
This abstraction process is what makes deep learning so powerful in tasks like computer vision or natural language understanding.
So can he learn the abstractive capacity that makes us human?
Not exactly. In fact, the human capacity for abstraction can also imagine things you have never seen. Maybe, things that don't even exist yet in the exact form in which they are imagined. AI, on the other hand, cannot even do the same with deep learning. It is still far from fully replicating human thinking, which involves unconscious processes and non-neural brain cells.
So no: AI does not become human with deep learning. At least not today and estimating published technologies.
Meanwhile il MIT has developed a new hybrid artificial intelligence architecture is tries to imitate the human ability to reason. They created three “libraries” that improve contextual understanding and abstraction in natural language.
Georgia Tech developed a neural network is tries to emulate human decision making. Unlike traditional neural networks, this model integrates a human-like evidence accumulation process, making decisions with a level of confidence that varies based on the evidence.
What else will happen?
