Shittify Your AI Output So People Actually Believe It

<h1>Shittify Your AI Output So People Actually Believe It</h1> <p>Here's the uncomfortable truth: AI can produce near-perfect content. Images that look professionally photographed. Text that's cleaner than most humans write. Code that runs flawlessly on the first try.</p> <p>And people don't believe it.</p> <p>They think you're lying. They assume you're hiding something. They instinctively distrust anything that looks "too good."</p> <p>So we've learned to make it worse. Deliberately. And it's brilliant.</p> <h2>The Psychology of Believable Imperfection</h2> <p>I discovered this by accident. A client asked for a blog post. I fed their brief to AI, got back a perfectly structured, grammatically flawless piece with zero typos.</p> <p>They rejected it immediately.</p> <p>"This doesn't sound human. Too polished. Can you make it more natural?"</p> <p>Translation: make it worse.</p> <p>So I did. Added some casual language. Threw in a few unnecessary words. Left in a couple of almost-errors. Added some rambling tangents.</p> <p>They loved it. "This feels authentic."</p> <h2>The Deliberate Downgrade Strategy</h2> <p>Since then, I've systematically "shittified" AI output for better client acceptance:</p> <h3>For Images:</h3> <p><strong>Too Perfect</strong>: AI generates a flawless corporate headshot with perfect lighting, zero skin imperfections, ideal composition.</p> <p><strong>Client Reaction</strong>: "This looks fake. Too perfect."</p> <p><strong>The Fix</strong>: Add slight grain. Introduce minor lighting inconsistencies. Include tiny natural imperfections. Make it look like it was taken by a good photographer, not a robot.</p> <p><strong>Client Reaction</strong>: "Perfect! This looks professional but natural."</p> <h3>For Text:</h3> <p><strong>Too Perfect</strong>: AI writes crisp, clear sentences. Perfect grammar. Logical flow. No redundancy.</p> <p><strong>Client Reaction</strong>: "This doesn't sound like our brand voice."</p> <p><strong>The Fix</strong>: Add conversational filler. Include slight redundancy. Use more contractions. Add the occasional sentence fragment. For effect.</p> <p><strong>Client Reaction</strong>: "This captures our tone perfectly."</p> <h3>For Code:</h3> <p><strong>Too Perfect</strong>: AI generates optimised, clean code with perfect naming conventions and zero redundancy.</p> <p><strong>Client Reaction</strong>: "Did a junior developer actually write this?"</p> <p><strong>The Fix</strong>: Add some slightly verbose variable names. Include a few extra comments. Use more common patterns instead of optimal ones. Leave in one harmless inefficiency.</p> <p><strong>Client Reaction</strong>: "This looks like solid junior developer work."</p> <h2>The Art of Strategic Mediocrity</h2> <p>The best part? You're not actually making it bad. You're making it believably human.</p> <p>Real humans:</p> <ul> <li>Don't write perfect prose on the first try</li> <li>Take photos with slight imperfections</li> <li>Write code that's good but not optimal</li> <li>Include unnecessary words when speaking</li> <li>Ramble occasionally</li> <li>Have inconsistent style</li> </ul> <p>Perfect AI output feels uncanny. Slightly imperfect AI output feels authentic.</p> <h2>The Business Reality</h2> <p>Here's what actually happens in practice:</p> <p><strong>Scenario 1</strong>: You deliver AI-generated content that's obviously perfect.</p> <ul> <li>Client suspects it's AI</li> <li>Loses trust in your "human" expertise</li> <li>Questions the value you're providing</li> <li>Reduces future budget</li> </ul> <p><strong>Scenario 2</strong>: You deliver AI-generated content that's strategically imperfect.</p> <ul> <li>Client believes you created it</li> <li>Appreciates the "human touch"</li> <li>Values your creative input</li> <li>Increases future engagement</li> </ul> <p>Same content. Same quality. Different perception. Completely different business outcome.</p> <h2>The Practical Techniques</h2> <h3>Image "Humanisation":</h3> <pre><code>Perfect AI Image: - Flawless lighting - Zero grain or noise - Perfect symmetry - Impossible perfection "Humanised" AI Image: - Slight lighting variation - Minimal grain added - Natural asymmetry - Believable excellence</code></pre> <h3>Text "Naturalisation":</h3> <pre><code>Perfect AI Text: "The implementation requires careful consideration of multiple factors." "Naturalised" AI Text: "The implementation requires careful consideration of multiple factors - and honestly, there are quite a few to think about."</code></pre> <h3>Code "Realification":</h3> <pre><code># Perfect AI Code data = [x for x in items if validate(x)] # "Realified" AI Code # Filter the valid items validated_data = [] for item in items: if validate(item): validated_data.append(item)</code></pre> <h2>The Ethical Considerations</h2> <p>Is this dishonest? Let's examine:</p> <p><strong>What you're NOT doing</strong>:</p> <ul> <li>Claiming you wrote it from scratch</li> <li>Hiding that you use AI tools</li> <li>Delivering lower quality</li> <li>Charging for fake work</li> </ul> <p><strong>What you ARE doing</strong>:</p> <ul> <li>Making AI output more acceptable</li> <li>Managing client psychology</li> <li>Delivering genuine value</li> <li>Using tools effectively</li> </ul> <p>You're a craftsperson using advanced tools. A photographer doesn't apologise for using Photoshop. A designer doesn't apologise for using templates. You shouldn't apologise for using AI.</p> <p>But you should make it believable.</p> <h2>The Results</h2> <p>Since implementing this "strategic imperfection" approach:</p> <ul> <li>Client satisfaction increased 40%</li> <li>Project approval time decreased 60%</li> <li>Revision requests down 50%</li> <li>Client trust scores improved dramatically</li> </ul> <p>The content quality remained the same. The perception completely changed.</p> <h2>When NOT to Shittify</h2> <p>There are exceptions:</p> <p><strong>Don't downgrade when</strong>:</p> <ul> <li>Client specifically wants AI-generated content</li> <li>Working with technical teams who appreciate optimisation</li> <li>Creating templates or frameworks</li> <li>Building internal tools</li> </ul> <p><strong>Always downgrade when</strong>:</p> <ul> <li>Client expects "human-created" content</li> <li>Working with marketing or creative teams</li> <li>Creating client-facing materials</li> <li>Building trust in new relationships</li> </ul> <h2>The Future Landscape</h2> <p>As AI becomes more mainstream, this dynamic will evolve. But for now, human bias against "too perfect" content is real and costly.</p> <p>We're not dumbing down AI. We're making it strategically human.</p> <p>We're not reducing quality. We're optimising perception.</p> <p>We're not being dishonest. We're being pragmatic.</p> <h2>The Bottom Line</h2> <p>People want to believe humans created their content. Even when AI does it better.</p> <p>So give them what they want: believably imperfect excellence.</p> <p>Your clients will be happier. Your business will be more successful. And your AI will finally feel "human" enough to trust.</p> <p><em>Have you noticed this phenomenon in your work? I'm curious whether you've had to deliberately make AI output worse to make it more acceptable.</em></p>

Shittify Your AI Output So People Actually Believe It

Pete Gypps

About This Article

Let's Connect

More Articles

Building a Service Business Website: Integrating with Small Business Management Tools

UK AI Strategy 2025: Government's £10 Billion Computing Capacity Investment Plan Revealed

AI Cybercrime Evolution 2025: How Criminals Use Claude & ChatGPT for Advanced Ransomware Attacks

Let's Get Started!