Why should we cease to confuse the human and machine intelligence

We are already accustomed to hear such phrases as "machine learning" and "artificial intelligence". We think that someone was able to reproduce the human mind inside a computer. This, of course, is not true. But part of the reason for which this idea is so widespread, due to the fact that the metaphor of human learning and intelligence has been very helpful for explaining machine learning and artificial intelligence. Some AI researchers in close contact with the community of neuroscientists, and inspiration goes both ways.

, however, that the metaphor can be an obstacle for people who are trying to explain the machine learning to those who know him less. One of the biggest risks combination of human and machine intelligence is that we are beginning to transfer too many rights machinery. But for those of us who work with the software, it is important to remember that it is an intelligent agent is a person - a person who builds these systems in the end.

It should hold the key differences between machine and human intelligence. Despite that similarity, of course, we are, looking at the differences we could better understand how artificial intelligence and how we build and use it as efficiently as possible.

Neural Networks

The central place in the metaphor that links human and machine learning is the concept of neural networks. The biggest difference between the human brain and artificial neural network - this is the scale of the neural networks of the brain. It is important to not only the number of neurons in the brain (which is estimated in billions), but also a surprising number of connections between them. But the problem goes deeper than simply scale questions. The human brain is qualitatively different from the artificial neural network in two other important reasons: compounds that nourish it, analog, not digital, and the neurons themselves are non-uniform and non-uniform (as opposed to artificial neural networks).

That's why the brain is so complex. Even the most sophisticated artificial neural network, although it is sometimes difficult to understand, is the underlying architecture and principles that guide. At least, we would like, so we are committed to this.

Even the most sophisticated neural network artificial intelligence designed for a specific purpose and to achieve a certain result. But the human brain does not have the same degree of focus in the project. Yes, he has the principles of self-preservation, and so on, but it still requires us to critical thinking and creativity, which still can not be programmed.

The beautiful simplicity of the AI ​​

The irony is that the artificial intelligence system is much simpler than a human brain, allowing the AI ​​to cope with much more computational complexity than we can.

Artificial intelligence neural network can store much more data and information than the human brain, mainly due to the type of data stored and processed by the neural network. They are discrete and specific as the contents of the Excel spreadsheet.

In the human brain data does not have the same properties of the discrete. Therefore, even if the artificial neural network can process specific data, it can not process the information in a rich and multidimensional manner, as does the human brain. This is a key difference between the designed system and the human brain. Despite years of research, the human brain is still unclear in many respects. This is due to the fact that the analog synaptic connections between neurons is almost impermeable for digital connections to an artificial neural network.

The speed and scale of the

Consider what this means in practice. The relative simplicity of the AI ​​allows you to quickly perform a complex task, and very well. The human brain just can not process data at a rate if, for example, converts speech to text or handles a huge range of cancer reports.

what matters is that it splits the data and information on the tiny components for the AI ​​in these contexts. For example, he can beat sounds on the phonetic part, which will then be converted into full sentences, or split the image into pieces to understand the rules by which great pictures are produced.

People often make the like, and are reminiscent of human machine learning; as well as algorithms, people break data or information into small pieces to handle it.

But there is a reason for this similarity. breakdown of the process is developed in each neural network engineer man. Moreover, the design process is usually based on the problem of the premise. How artificial intelligence system splits the data set is its own way of "understanding." Even when running a very complex algorithm, the parameters of how to teach the AI ​​- it breaks up the data to process - established from the outset.

The human intellect: problem identification

The human intellect should not be such a set of constraints, that's what makes us much more effective in solving problems. It is the ability of people to "create" the problem allows us to solve them well. In our approach to problem solving is an element of the contextual understanding and decision-making. AI can and could unpack the problem or find new ways to solve them, but it can not determine the problem, which is trying to solve.

In recent years, it becomes the subject of attention algorithmic insensitivity. A growing number of scandals associated with the bias AI systems. Of course, this is directly related to the prejudices of those who make the algorithms, but the reasons where there are algorithms in these biases, can identify only people.

The human and machine intelligence to complement each other

We must remember that artificial intelligence and machine learning - is not just algorithms that "got out of hand" and are out of our control. They create, design and create us. This imposes on us the responsibility for our future - it will be what we make it ourselves.

We are willing to accept an artificial intelligence? Tell us in our chatting in a telegram.