Issue 1 Explainable Artificial Intelligence

A Deep Dive Into XAI

Table of Contents

  1. Summary
  2. What is XAI?
    1. How We Got Here
    2. The Argument for XAI
    3. The Difficulties of XAI
    4. The Argument Against XAI
  3. Does AI Need to be Explainable?
    1. What is an Explanation?
    2. Are Human Decisions Explainable?
    3. Types of Explanations
    4. Attributes of a Good Explanation
    5. Places Where Explanations are Most Important
  4. Methods to Achieve Explainability
    1. Possibility 1 – “It’s a black box, trust us.”
    2. Possibility 2 – “We can’t explain it, but here are stats about how well it works.”
    3. Possibility 3 – Surrogate Models
    4. Possibility 4 – “Here are the inputs we use.”
    5. Possibility 5 – Partial Explanation
    6. Possibility 6 – Sensitivity Analysis
    7. Possibility 7 – Interpretive Explainability
    8. Possibility 8 – Certification of Models
    9. Possibility 9 – Or Some Partial Combination of the Factors Above
  5. Alternatives to Explainability
    1. Alternative 1 – Keep Models Simple
    2. Alternative 2 – Only Use AI in Unimportant Arenas
    3. Alternative 3 – Give Up
  6. Analysis: Will Consumers Demand XAI?
  7. Regulation of AI


In the 1980s, Bloom County was arguably the most popular comic strip on the planet. In one sequence, the penguin Opus is running for elected office, and the local computer nerd, Oliver Wendell Jones, uses AI to analyze polling data and determines that the ideal image that voters want in a candidate is “chocolate éclair.” But, as we see the next day, Jones has made an error in his calculations:

The root problem here is that of AI explainability. Had Opus demanded an explanation on why chocolate éclair was the suggestion of the AI, the error would have been discovered. But Opus didn’t, for he trusted the system, and the system was wrong. The foibles and pitfalls surrounding this situation is the topic of this, the first issue of Deep Dive into AI.

Issue #1 – Explainable Artificial Intelligence

As artificial intelligence becomes more powerful and is used in more places of greater importance, the question of why an artificial intelligence (AI) makes the recommendation or choice that it does becomes ever more relevant.

The challenges of creating explainable artificial intelligence (XAI) are numerous and potentially insurmountable. Yet, social and legislative burdens are being placed on companies to provide XAI. How will this all unfold? Let’s dive in.