The Opinion Advantage
Session 12.5 · ~5 min read
AI Defaults to Consensus
Ask any AI model a question that has a debatable answer. You will get a balanced response. "On one hand... on the other hand... ultimately, it depends on your specific situation." This is not thoughtfulness. It is RLHF-trained hedging. The model is optimized to avoid being wrong, and the safest way to avoid being wrong is to present all sides without committing to any of them.
Readers do not remember balanced takes. They remember positions. "I think X is wrong, and here is why" sticks in memory. "There are various perspectives on X" does not. In an environment where every AI can produce balanced overviews, the person who takes a clear position stands out by default.
The Opinion Advantage: Having informed opinions, supported by evidence and stated without hedging, is a competitive advantage in content markets saturated with AI-generated both-sides-ism. The courage to be wrong is more valuable than the safety of being vague.
The Opinion Spectrum
Not all opinions are equal. The spectrum runs from uninformed hot takes (worthless) to evidence-backed practitioner positions (invaluable).
'I think remote work
is better'
No evidence, no experience"] --> B["Common Opinion
'Remote work has
pros and cons'
Generic, AI-reproducible"] B --> C["Informed Opinion
'Remote work works for
async-first teams but fails
for collaborative design work'
Based on observation"] C --> D["Practitioner Opinion
'After managing 3 remote
teams and 2 hybrid teams
over 4 years, I found that
remote works only when you
restructure meetings entirely.
Here is how.'
Based on direct experience"] style A fill:#c47a5a,color:#111 style B fill:#8a8478,color:#ede9e3 style C fill:#c8a882,color:#111 style D fill:#6b8f71,color:#111
AI generates content in the "Common Opinion" zone. It can sometimes reach "Informed Opinion" if prompted carefully with the right context. It cannot reach "Practitioner Opinion" because that requires original experience that is not in the training data.
Five Properties of Strong Opinions
| Property | Description | Example | AI Can Generate? |
|---|---|---|---|
| Specific | Applies to a defined context, not all situations | "Code reviews are a waste of time for teams under 4 people" | No. AI hedges with "it depends on team size." |
| Evidence-backed | Supported by data, observation, or documented experience | "We stopped doing code reviews and our bug rate stayed flat for 6 months" | No. AI cannot fabricate your data. |
| Actionable | The reader can do something different after reading it | "Replace code reviews with pair programming and automated linting" | Partially. AI can suggest alternatives but not from experience. |
| Falsifiable | Could be proven wrong. Not a tautology. | "This approach works for teams shipping weekly. It fails for monthly release cycles." | No. AI avoids falsifiable claims. |
| Owned | The author takes personal responsibility for the claim | "I believe this. I have tested it. Here is what I found." | No. AI does not have beliefs. |
Building Your Opinion Portfolio
You need a collection of strong opinions relevant to your field. Not opinions about everything. Opinions about the things you know from direct experience. These opinions become the backbone of your content, the recurring themes that make your work recognizable and worth following.
Start by listing the things you believe that most people in your field would disagree with, or at least not say publicly. These are your highest-value opinions because they provoke thought, invite debate, and attract the audience that shares your perspective (your ideal audience).
Then list the things you believe because your experience showed you something the textbooks missed. These are your evidence-backed opinions. They are less provocative but more defensible.
Together, these two lists form your opinion portfolio. Every piece of content you produce should draw from this portfolio. Not every piece needs a controversial take. But every piece should have a clear position.
The "Would AI Say This?" Test
Before publishing any opinion-based content, run this test: prompt an AI model with the same topic. If your position matches the AI's output, you are writing consensus, not opinion. If your position differs, you have something worth publishing.
This test is not about being contrarian for its own sake. It is about ensuring that your content adds something that did not already exist in the AI-accessible information pool. Agreement with AI is not a problem if you support it with original evidence. Identical phrasing and framing with AI is a problem because it means you could be replaced by a prompt.
Further Reading
- What's Your Edge? Rethinking Expertise in the Age of AI, MIT Sloan Management Review
- How AI Killed Your Content Strategy and What to Do Next, Optimist
- AI-Generated Content in Transition: Between Progress and Fatigue, EY (2025)
- Content Strategy in a Saturated Market: Putting AI to Work, Matrix Marketing Group
Assignment
Write 5 opinionated statements about your field that AI would not generate because they are too specific, too controversial, or too grounded in personal experience. For each statement, write one supporting paragraph from your direct experience. Then prompt an AI model with the same topics and compare. If any of your opinions match the AI output, replace them with something sharper. These are your opinion assets.