Lakeside Analytics | Explainable AI (XAI)

A Distraction Rather Than A Solution 🚀

The Scoop: The promise of XAI distracts from the deeper problem: We Are Biased.

AI does not create bias; it absorbs and reflects the data of our world as it is. Ultimately bias mitigation is an illusion. We cannot rewrite history, nor should we remove bias from the data that represents our reality - because this would be a manipulation and highly unethical.

With XAI, we reach for the easy way out: explanations and scapegoating the developer. But transparency isn’t fairness, and blaming the developer is just an attempt to cya.

Let’s explore this through a powerful analogy:

1️⃣ Explainable AI = Psychoanalysis 🧠🔍
Like psychoanalysis traces human behavior to past experiences, XAI tries to explain AI decisions. Both, however, often miss deeper structural causes.

2️⃣ Blaming childhood upbringing = Making developers scapegoats 👨‍👩‍👧‍👦💥
We often blame parents for a child’s behavior, just as we blame AI developers for biased outcomes. But in both cases, the environment (data) plays a bigger role.

3️⃣ Nature vs. nurture = Programmatic bias vs. biased training data ⚖️💻
Human behavior is shaped by both genes and environment—just as AI is shaped by both its code and the data it’s trained on.

4️⃣ A grown-up turned racist = An AI turned racist 🚫🤖
Both humans and AI absorb biases from their surroundings—society for people, data for machines. The results are predictable.

5️⃣ Taking away a kid’s phone to prevent bad influences = Eliminating biased data 📱🛑
Removing harmful content might help on the surface, but it doesn’t eliminate bias in society. Worse, deleting biased data can sometimes erase historical realities AI needs to recognize.

6️⃣ Banning a child from seeing certain people = Restricting AI’s access to data 🚷💡
Limiting exposure may prevent harm, but it doesn’t remove bias from reality. Similarly, restricting AI from certain datasets doesn’t fix the problem—it just hides it.

7️⃣ Creating laws to curb bad behavior = Skewing training data to make AI ‘fair’ ⚖️📊
Laws reduce harm but don’t change history. Similarly, forcing AI to produce “fair” outputs through artificial constraints doesn’t fix systemic bias—it just masks it.

🔹 What XAI Tries to Do—And Why It Fails 🔹
XAI promises to make AI decisions transparent, suggesting that explainability leads to fairness and control. But the deception is that transparency is not control—XAI doesn’t fix bias; it just exposes it while ignoring the real issue: AI is trained on a biased world. The failure is that no amount of explainability changes the flawed data AI learns from, yet XAI creates a false sense of reassurance, as if understanding a bad decision makes it any better.

Disclaimer

The companies and organizations mentioned in this article are referenced for informational and analytical purposes only. All discussions about their potential roles and interests in space-based data centers are based on publicly available information and do not imply any endorsement, partnership, or direct involvement unless explicitly stated. The opinions expressed are solely those of the author and do not reflect the official positions of the companies mentioned. All trademarks, logos, and company names are the property of their respective owners.