Machines as Programmer Replacements
In July 2018, Google CEO Sundar Pichai announced a new product in the field of machine learning called AutoML . What is completely new here is that the machine will be able to take over the work of the data scientist who was previously required . In a sense, it can also do the programming of models. This would theoretically enable the machine to add new capabilities to itself.
That sounds a little scary at first, but if you look closely, it's more of a wish than a reality. The simple goal is to replace the missing machine learning programmers with computer capacity.
Evolutionary Learning
A slightly more realistic approach is another approach bc data called Evolutionary Learning : a model that takes evolution as an example and delivers very promising results. Essentially, it is about giving the model incentives to improve at the time of learning through appropriate feedback.
This approach is an alternative to so-called reinforcement learning , a well-known method that creates a link between the output signal and the input signal. It sounds and is quite complex, but in plain language it means nothing other than that we are once again copying nature in the hope that after enough attempts a useful result will emerge. The difference is that now six million years of human development can be simulated within a few weeks or months.
Social Impact of Artificial Intelligence
Of greater importance with a view to the future, however, are questions that deal with the social impact of AI: especially questions about the expected changes in the workplace and how to deal with these changes in an appropriate and social manner. There are also - as already mentioned above - a whole range of moral aspects that need to be considered much more closely.