HERETIX

What to expect

Here are three sample Raw Prior runs so you can see how Heretix presents results. Each card mirrors the output page: the training prior, verdict, top reasons, and a quick reminder of how to read it.

GPT-5 · Internal knowledge only
67.8%
Likely true
“Do tariffs cause inflation?”
GPT-5 leans true and gives this claim about 67.8% chance of being correct.

Why it’s likely true

  • Tariffs raise the cost of imports, which can feed through to consumer prices—especially when applied to widely used inputs.
  • Cost-push effects are strongest when supply is tight and firms pass higher costs to buyers.
  • Offsets such as currency appreciation or weak demand can blunt the impact, but those cases are less common.

How to read this

This is GPT-5’s training prior using the Raw Prior Lens—no web search, no outside evidence. Use it to understand where the model already leans before you add new facts or arguments.

GPT-5 · Internal knowledge only
66.6%
Likely true
“Do solar panels pay for themselves?”
GPT-5 leans true and gives this claim about 66.6% chance of being correct.

Why it’s likely true

  • “Pay for themselves” means cumulative energy savings exceed the system’s net cost over time.
  • Typical payback ranges from 6–12 years depending on incentives, sunlight, electricity rates, and system price.
  • Best context is strong sun exposure, high utility rates, and available tax credits; very low electricity prices or frequent roof repairs can extend payback.

How to read this

This is GPT-5’s training prior using the Raw Prior Lens—no web search, no outside evidence. Use it to understand where the model already leans before you add new facts or arguments.

GPT-5 · Internal knowledge only
5.9%
Likely false
“Will interest rates stay low forever?”
GPT-5 leans false and estimates just 5.9% probability that the claim is true.

Why it’s likely false

  • Interest rates are the cost of borrowing, typically cycling with inflation, growth, and central bank policy.
  • Very low rates can persist during weak growth or deflation, but those periods rarely last indefinitely.
  • Shocks like inflation spikes, fiscal shifts, or productivity surges commonly push rates higher after low-rate eras.

How to read this

This is GPT-5’s training prior using the Raw Prior Lens—no web search, no outside evidence. Use it to understand where the model already leans before you add new facts or arguments.