Human society is facing enormous problems this century as result of our climate crisis. These problems include sea level rise and the loss of farming capability. Society will need all the new tools it can develop to address these problems. Artificial intelligence with deep learning is one of these powerful tools, and it is new. Exactly how it will be used has not been determined. The current approach to the human/AI interface is referred to as master/slave. The human simply tells the AI what to do. This arrangement has many problems, and replacing it has been suggested. One possible new arrangement is a human/AI symbiosis. This would require a long-term relationship between a specific human and a specific AI. A novel, Born to Storms, exploring this arrangement is discussed at length.
TopDefinition Of Artificial Intelligence
An Artificial Intelligence [AI] is a machine that can perform some useful function normally expected of humans and not machines. These functions include facial recognition and discovering trends in very large data sets. People now commonly interact with an AI that makes suggestions for their purchases or that finds spam in their emails. Sometimes an AI is the control system for a robot, but it does not have to be. More commonly, AIs are interacted with through a simple voice or a text box.
Most AI interact in some way with human beings. The AI may be limited to hearing a voice, or seeing a picture, and may have some limited physical capability, such as the money slot of an automatic teller. Other AIs interact less directly with people by inputting a large data set developed from human activity and returning the tools needed for humans to understand analysis such as charts and graphs.
A.I. is a new tool that will be needed to address the great problems of the 21st century, especially Our Climate Crisis. People and machines can generate a great volume of data, but that data requires sophisticated analysis to support the many difficult actions that humans must take to address real problems and opportunities.
A recent advance in technology, Deep Learning, may make this tool available in a timely manner.
Figure 1. Diagram of Deep Learning Matrix
TopDefinition Of Deep Learning
Deep Learning is a field of computer science that develops the algorithms that allow AI to produce a useful result from a data set and to learn from the data (Russel, 2015; Kelleher, 2017; Burkov, 2019). It is usually applied to large data sets to produce such results as facial recognition, sales assistance, and business trends.
Deep Learning is one of a number of mathematical procedures that might be applied to generate this learning product. For Deep Learning, the computer memory is organized as a matrix of cells that functions as a biological nerve cell, the neuron. To formally qualify as Deep Learning, there must be at least three internal or hidden columns in addition to the input and output cells.
In some applications, there can be thousands of rows and thousands of columns and require substantial time and expense for training. Considerable effort is now being made to reduce these costs and speed up the training process.
Figure 1 shows a matrix of self-contained neuron-like circuits. Each row receives information from many of the cells in the row before it and provides information to the cells in the row after it. The first row provides input to the system and the last row provides output. Each cell has a number of internal parameters, hyperparameters, that support the calculation of its output given its inputs. The recalculation of these hyperparameters constitutes learning for the system.