Spotting AI BS - Humans do it best!

In a world increasingly dominated by artificial intelligence hype and facts, there’s one area where AI seems to get wrong - producing quality bullshit. AI BS is really bad at all levels. For a human user, probably this is one of the most hated attribute of AI - especially when i challenges human creativity and jobs.  But what exactly is the nature of AI bullshit? It’s the flood of plausible-sounding but ultimately meaningless responses generated by AI models. These answers often masquerade as insightful commentary but are, in reality, just a bunch of buzzwords and jargon strung together to create an illusion of intelligence or correctness.  This article tries to understand AI bullshit, why humans are so adept at detecting it, and what we can do to mitigate this problem.

What is AI Bullshit - in an academic context?

AI bullshit can take many forms, each with its unique flavor of emptiness. Here are some common types:

1. Training Model Bias: AI systems learn from vast datasets, which are often riddled with biases. When these biases are reflected in AI’s outputs, it can lead to responses that are skewed, misleading, or even harmful. This type of bullshit isn’t just annoying; it can reinforce negative stereotypes and misinformation.

2. Appeal to Prompter Bias: AI often tailors its responses based on what it perceives the user wants to hear. Instead of providing objective insights, it ends up echoing the user's biases or expectations, leading to answers that feel comforting but lack substance.

3. Acrobatic Lingo and Jargon: Ever encountered an AI response loaded with terms like “tapestry,” “dive deep,” or “delve into”? This is AI’s attempt to sound profound by using complex language. However, this often results in verbose but shallow content that fails to deliver real insights.

4. AI Fake Creativity: One of the more deceptive forms of AI bullshit is fake creativity. AI might generate ideas that seem novel or innovative at first glance, but upon closer inspection, they turn out to be random or nonsensical combinations of existing concepts—essentially an exercise in sounding smart without actually being so. This is called Hallucination. 

5. Overgeneralization: AI often makes sweeping generalizations that sound logical but are overly simplistic. This is particularly problematic in complex fields where nuance is crucial, as it can lead to misleading conclusions.

Why Humans Don’t Like AI BS

Humans have a natural aversion to bullshit, or in a fair context - a taste for it, especially when it comes from something that’s supposed to be smarter than us - leading to a mike drop moment where all listeners play along. Form a human angle, here’s why AI bullshit is particularly irksome - mostly are not scientific but rather an oversimplification of the "general user perception". 

• Feeling Sidelined: When AI generates meaningless responses, it can feel like we’re being overshadowed by something that doesn’t even possess genuine intelligence. This is particularly frustrating in professional environments where decision-making relies on nuanced understanding.

• The Idiot Paradox: No one likes the idea of losing their job to a machine, especially if that machine is just regurgitating nonsense. It’s one thing to be outdone by a super-intelligent AI, but quite another to be replaced by what amounts to a glorified parroting device.

• Erosion of Trust: Over time, repeated exposure to AI-generated bullshit can erode trust in AI systems. If users feel that they can’t rely on AI for meaningful insights, they’re less likely to integrate it into their decision-making processes.

How to Spot AI Bullshit

Spotting AI bullshit is about knowing what to look for - a pattern of cringe to say the least. Here are some tell-tale signs:

1. Overuse of Buzzwords: If an AI response is packed with jargon but lacks clear, actionable content, that’s a red flag.

2. Lack of Specificity: Vague answers that don’t address the specific question or problem at hand are often a sign that the AI is out of its depth.

3. Echoing Prompter’s Bias: When AI seems to agree too readily with the user’s assumptions or biases, it’s likely not thinking critically.

4. Repetition of Phrases: AI often repeats certain phrases or ideas to create the illusion of depth. If you notice the same concept being rephrased multiple times, it’s likely fluff.

5. Inconsistent Logic: AI bullshit often falls apart under scrutiny. If an AI’s reasoning doesn’t hold up when you dig deeper, it’s a sign that the response is more style than substance.

6. Superficial Creativity: If an AI suggestion sounds too “out there” without any grounding in reality, it’s probably fake creativity at work.

7. Over-Polished Responses: AI-generated content that sounds overly polished or rehearsed might be trying to mask a lack of genuine understanding.

8. Ignoring Counterarguments: AI bullshit often fails to consider alternative perspectives or counterarguments, presenting its response as the only viable solution.

9. Shifting the Goalposts: If an AI keeps changing the topic or redefining the problem to avoid addressing the core issue, it’s a classic bullshit tactic.

10. Overconfidence: AI can sometimes present speculative answers with unwarranted certainty. This overconfidence is a clear sign of bullshit, especially in complex or uncertain scenarios.

Examples of AI BS

What Needs to Happen to Fix AI BS

To address the issue of AI bullshit, several steps need to be taken:

1. Enhanced Dataset Curation: AI needs to be trained on datasets that are not only large but also carefully curated to minimize bias and maximize relevance.

2. Algorithmic Transparency: Users should be able to understand how AI models generate their responses. Transparency in AI processes will help users identify potential bullshit and treat AI outputs with the appropriate level of skepticism.

3. Human-AI Collaboration: Rather than relying on AI alone, human oversight is essential to catch and correct bullshit before it causes harm. This means Guardrails at prompts (UI/UX), the algo, and the backend.

4. Ethical AI Development: Developers must prioritize ethical considerations in AI design, ensuring that their models don’t perpetuate misinformation or generate misleading content.

5. Improved AI Creativity: AI needs to be better equipped to generate genuinely creative ideas. This might involve integrating more sophisticated models of human creativity or incorporating real-world feedback into AI training.

6. Public Education: Educating users about the limitations of AI and the signs of AI bullshit is crucial. The more knowledgeable the public, the less likely they are to be fooled by AI-generated nonsense.

7. Regular Audits of AI Systems: Conducting regular audits of AI outputs can help identify patterns of bullshit and implement corrective measures before the AI is deployed in high-stakes environments.

8. Encouraging Critical Thinking: Both AI developers and users need to cultivate critical thinking skills to better analyze and evaluate AI-generated content.

9. Feedback Loops: Incorporating user feedback into AI systems can help refine responses and reduce the incidence of bullshit over time.

10. Promotion of Ethical Standards: Establishing and promoting ethical standards for AI usage can help mitigate the spread of AI-generated bullshit, particularly in fields like journalism, science, and education.

Humans Do BS Best, But AI Is Catching Up

Now let us go back to our topic, sometimes a healthy dose of BS is needed; here, humans have a distinct advantage. Humans have been perfecting the art of bullshit for centuries—whether it’s in politics, business, or everyday life. However, AI is rapidly catching up. The difference is that while human bullshit can be strategic, AI bullshit is often unintentional, a byproduct of its design and training.

Humans have the advantage of understanding context, nuance, and intent, which allows them to bullshit with purpose. AI, on the other hand, lacks this deeper understanding and often produces bullshit simply because it doesn’t know any better. This difference is crucial, as it highlights the limitations of AI and the need for ongoing human oversight.

While AI has the potential to revolutionize many aspects of our lives, it’s not without its shortcomings—chief among them being its propensity to generate bullshit. By recognizing the signs of AI bullshit and taking proactive steps to address them, we can ensure that AI serves as a valuable tool rather than a source of misinformation. After all, humans may be the original masters of bullshit, but that doesn’t mean we should let AI take over the throne.


At the Value Creation Innovation Institute (VCII), we understand the importance of fostering innovation in today’s fast-paced business environment. Visit VCII today to learn more about how this framework can revolutionize your business.

#AI #ArtificialIntelligence #AIEthics #TechInnovation #AIUseCases #DigitalAge #MachineLearning #AIandBusiness #LLM #AIResponsibility #TechTrends #AIChallenges #BusinessStrategy