possomcohen.com

I’m Sorry to Burst Your Bubble: You Are Being Fooled About AI, and You Will Soon Feel Really Stupid

Excerpt:

Claude. AI Response

Claude. AI Response

Short answer: you can’t definitively tell—and neither can any tool—but this piece reads much more like a human-written article than pure AI output.

Here’s a grounded breakdown based on the text you shared:

---

🧠 1. Strong signs of human authorship

• Consistent voice and intent: The article has a clear argumentative stance, sustained over a long piece, with deliberate rhetoric (e.g., repeating structures, emphasis, tone shifts).
• Author self-awareness: Early on, David William Silva explicitly says he’s using a specific rhetorical device (“contrastive negation followed by a corrective assertion”) . That kind of meta-commentary is uncommon in raw AI output.
• Personal framing and motivation: He positions himself (“I am on a mission…”, “I have used AI many times…”) in a way that feels grounded in a real persona.

---

⚠️ 2. But there are AI-like traits

Some parts do resemble AI-assisted writing:

• Highly polished, repetitive rhetoric
  Phrases like:
  • “It is not X. It is Y.”
  • “That is not intelligence. That’s pattern-matching…”
  These are stylistic patterns AI often produces (though humans use them too).
• Overconfident generalizations
  Claims like “LLMs are a dead end” or sweeping statements about the industry are presented with strong certainty but limited nuance—common in both opinionated humans and AI-generated essays.
• One glitchy artifact
  The repeated word sequence:“helpshelpshelpshelps”
  That’s unusual. It could be:
  • a copy/paste error
  • editing artifact
  • or AI generation glitch
    But it’s not definitive either way.

---

🔬 3. Structural clues

• The article is too coherent over a long span to be raw, unedited AI output.
• It includes:
  • named figures (LeCun, Altman, Hinton)
  • narrative framing
  • persuasion strategy
    These suggest at least human direction and editing.

---

🧾 4. Most likely scenario

The most plausible explanation is:

Human-written (or heavily human-edited), possibly with some AI assistance.

That’s very common for Substack writers today:

• Draft ideas themselves
• Use AI for phrasing, cleanup, or expansion

---

🚫 5. Important reality check

Even the author claims:

AI detection is mathematically impossible

That’s actually close to the truth in practice:
There is no reliable way to prove authorship from text alone.

---

✅ Bottom line

• ❌ You cannot prove it was written by AI
• ❌ It does not look like purely AI-generated content
• ✅ It most likely reflects a human author’s voice, possibly AI-assisted

---

If you want, I can:

• compare it to known AI-generated essays
• or highlight specific sentences that feel most “AI-like” vs “human-like”

That’s usually the fastest way to build intuition.