{"id":83042,"date":"2022-02-17T09:00:00","date_gmt":"2022-02-17T08:00:00","guid":{"rendered":"https:\/\/www.hiig.de\/?p=83042"},"modified":"2023-03-28T14:02:45","modified_gmt":"2023-03-28T12:02:45","slug":"explainable-ai","status":"publish","type":"post","link":"https:\/\/www.hiig.de\/en\/explainable-ai\/","title":{"rendered":"Why explainable AI needs such a thing as Society"},"content":{"rendered":"\n

Have you ever asked yourself what the basis of your search engine autocompletions is? For example, when it was suggested that you search for what it feels like to have heartburn, whilst your intended search seemed to have nothing to do with it at all. There is not yet a standard for explaining such automated decisions. Moreover, today’s explainable AI (XAI) frameworks focus strongly on individual interests, while a societal perspective falls short. This article will give an introduction to communication in XAI and introduce the figure of <\/strong>the public advocate<\/em><\/strong> as a possibility to include collective interests in XAI frameworks. <\/strong><\/p>\n\n\n\n

The article is based on the thoughts of Dr. Theresa Z\u00fcger, Dr. Hadi Asghari, Johannes Baeck and Judith Fa\u00dfbender during the XAI Clinic in Autumn 2021<\/a>.<\/p>\n\n\n\n

Missing or insufficient explainability for lay people and society<\/strong> <\/strong><\/h2>\n\n\n\n

Have you ever asked yourself what the basis of your search engine autocompletions is? For example at that time when you typed \u201chow does\u201d and your search engine suggested \u201chow does…it feel to die\u201d, \u201chow does…it feel to love\u201d, \u201chow does…it feel to have heartburn\u201d but you actually wanted to continue typing \u201chow does\u2026 a2<\/sup> relate to b2<\/sup> in Pythagoras\u2019 theorem\u201d. If explanations for automated decisions were a standard, you would have been able to get an explanation of the inner workings of that search engine fairly easily. Due to a mixture of technical feasibility, communicational challenges and strategic avoidance, such a standard does not exist yet. Whilst a number of major providers and deployers of AI-models have published takes on Explainable AI (XAI) \u2013 most prominently IBM<\/a>, Google<\/a> and Facebook \u2013 none of these efforts offer effective explanations for a lay audience. In some cases, lay people are simply not the target group, in others, the explanations are insufficient. Moreover, collective interests are not taken into account sufficiently when it comes to how to explain automated decisions; the focus lies predominantly on individual or private interests.<\/p>\n\n\n\n

This article will focus on how explanations for automated decisions need to differ in regards to the audience that is being addressed \u2013 in other words, on target group specific communication of automated decisions. In light of the neglected societal perspective, I will introduce the figure of the public advocate<\/em> as a possibility to include collective interests in XAI frameworks. <\/p>\n\n\n\n

Technical elements of AI-systems to explain <\/strong><\/h2>\n\n\n\n

The technological complexity of AI-systems makes the traceability of automated decisions difficult. This is due to models with multiple layers, nonlinearities and untidy, large data sets, amongst other reasons. As a reaction to this problem, there have been increasing efforts to develop so-called white-box algorithms or to use more simple model architectures which produce traceable decisions, such as decision trees.<\/p>\n\n\n\n

But even if each element of an AI-system is explainable, a complete explanation for an automated decision would consist of a fairly large number of elements. To give an idea of these elements, let me share a dry yet helpful overview of possible elements (based on Liao et al. (2020)<\/a>):<\/p>\n\n\n\n

(1.) The global model<\/em> which refers to the functionalities of the system that has been trained, this includes which training data has been used, and which architecture (i.e. a convolutional neural network, linear regression, etc.). Global means that the functionality of the system is not case-specific (2.) The local<\/em> decision, which concerns a decision in a specific case. (3.) The input <\/em>data which refers to the specific data a local decision is made on. (4.) The output<\/em> refers to the format and the utilisation of the output the system gives (5.) A counterfactual explanation<\/em>, which shows how different the input would have to be in order to get a different output; such as (6.) the performance<\/em> of the system.<\/p>\n\n\n\n

The challenge of target group specific communication <\/strong><\/h2>\n\n\n\n

If what you\u2019ve read up to now has either bored or overwhelmed you, it could either mean that you are not the target group for this blog post or that I have missed the sweet spot between what you, as part of my target group, knew already and what you expect from this article. Target group specific communication and hitting that sweet spot is a struggle when explaining automated decisions as well.<\/p>\n\n\n\n

To give you a schematic, but better, explanation, here are the elements listed above, applied to the search engine example from the beginning of this blog post: <\/p>\n\n\n\n