AI Designs GUI: First Steps

In the last year, news has been multiplying that artificial intelligence has learned to draw pictures from the text description. Microsoft's DALL-E neural network is widely known for this purpose; it was released in May 2021. And what processes can be automated using such technologies? The most interesting utilitarian direction seems to be drawing user interfaces and generating captions for drawings. Many large companies in software development use the services of outsourcing services like Grow Your Staff. India is the undisputed leader among outsourcing providers.

Artificial intelligence

Artificial intelligence is a set of algorithms designed to make certain tasks easier for a person, especially difficult ones for him. For example, to calculate trends. Some tasks that are solved by a person almost involuntarily, for example, recognition of images or emotions, are very difficult or practically unsolvable for modern AI.

Today, artificial intelligence is increasingly understood as machine learning, deep learning, and reinforcement learning. All these paradigms are reduced to a logical conclusion based on the provided set of data by the underlying algorithms. But in fact, such systems simply mimic the functions of the human brain, in particular, learning and problem-solving. AI is programmed by people, and all people are biased, and therefore AI can also only be objective to some extent. AI has no consciousness, so it doesn't really "understand" anything. At the same time, one of the most important directions in the development of artificial intelligence is the development of human-machine interactions.

AI and UI

Artificial intelligence is especially good for complex, routine, and well-defined tasks. The human intellect, on the other hand, tends to do things that are simple and/or interesting. A person learns from experience and brings some acquired experience to automatism. Artificial intelligence learns from sets of data precisely because it is incapable of gaining experience.

An outstanding example of such automaticity is the work with the user interface. Working with the user interface is, first of all, communication: the exchange of data between a person and a computer.

Interhuman communication has been studied for many years within the framework of communication theory, and communication between machines has not yet generated a full-fledged theory and is at the stage of formulating machine ontologies. The intermediate discipline is the study of communication between man and machine, and such communication always takes place through one or another user interface. The most important part of human-machine communication is the design, implementation, and evaluation of user interfaces. Currently, interest in user interfaces is only growing, the interfaces themselves are becoming both simpler and more complex. The simplicity of an interface is called "intuitiveness," and the price of that "simplicity" is to abstract away the growing complexity that goes under the hood.

Conclusion

It is likely that when the technology for creating user interfaces is fully developed, it will be based on pattern recognition and annotation, reinforcement learning, and multi-world interpretation of user experience. The practical benefits of such technology would be enormous. A powerful artificial intelligence that received a photo of a potential UI as input could compose the layout of the interface, layout template code for its specific implementations, and also provide for extensibility and immediately exclude entire classes of user errors. After a short use of the device, its interface could well adapt itself to the owner. Not only testing of interfaces would be simplified, but also the analysis of errors, and the programmer would be spared from performing many routine tasks.