Currently, we can build the Artificial Intelligence (AI) approaches that respectively explain their actions within the use of goal trees 1. By moving up and down across the tree, it keeps tracking the last and next movements. Therefore, giving the ability to the machine for "explain" the actions.

Explainability regarding human levels, requires some cognitive effort, such as self-awareness, memory retrieval, a theory of mind and so on 2. Humans are adept at selecting several causes from an infinite number of causes to be the explanation. However, this selection is influenced by certain cognitive biases. The idea of explanation selection is not new in eXplainable Artificial Intelligence (XAI) [3, 4]. But, as far as we are aware, there are currently no studies that look at the cognitive biases of humans as a way to select explanations from a set of causes.

Despite a clear definition and description of the XAI field, several questions remain present. The issues are summarized in just one sentence and listed as follows.

That said, our question is:

How can we create and build XAI?


1 Hadoux, Emmanuel, and Anthony Hunter. Strategic Sequences of Arguments for Persuasion Using Decision Trees. AAAI. 2017.

2 Miller, T., 2018. Explanation in artificial intelligence: Insights from the social sciences. Artificial Intelligence.

3 Gunning, D., 2017. Explainable artificial intelligence (XAI). Defense Advanced Research Projects Agency (DARPA), nd Web.

4 Samek, W., Wiegand, T. and Müller, K.R., 2017. Explainable artificial intelligence: Understanding, visualizing and interpreting deep learning models. arXiv preprint arXiv:1708.08296.


If the Artificial Intelligence will use natural language as an intermediate representation this is equal to a shared description between man and machine. Man will understand a sentence like “obstacle ahead”, while the machine will understand it too. The desktop metaphor better known as GUI is an early example of a man-machine interface which is working in both directions.

From a philosophical standpoint we can discuss about such kind of abstraction layer is anymore an Artificial Intelligence because it is not acting by it's own but translates commands from one direction to the other. The layer takes numerical values from the machine and converts it into natural language and it is doing the backward step too. Sometimes such a layer is called a transparent AI, because it is working similar to a computer mouse.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.