Public preview

Fact-Grounding Strategies

Every factual AI claim = ask for a source. No source = don't use.

8 min Intermediate

Goal

Distinguish factual claims (need verification) from analyses (can accept).

Steps

1

Read AI output and highlight: red = factual claim (numbers, dates, names, quotes), blue = analysis/opinion.

3 more steps locked

Expected Outcome

Immediate

You discover 10-30% of AI "facts" are wrong or fabricated.

Evidence base

School: AI Reliability / Hallucination Research

Founders: Stuart Russell · Anthropic Safety Team (2023)

Russell's "Human Compatible" warns trusting AI output. Anthropic publishes hallucination rates 5-15% even for advanced models.

Frequently asked questions

How long does "Fact-Grounding Strategies" take?

This exercise takes about 8 minutes, practiced solo in a Written reflection format.

When will I notice the effect of "Fact-Grounding Strategies"?

You discover 10-30% of AI "facts" are wrong or fabricated. In the short term: You develop surface-skill distinguishing AI "fact" vs "analysis".

Is "Fact-Grounding Strategies" evidence-based?

Yes — it draws on AI Reliability / Hallucination Research (Stuart Russell, Anthropic Safety Team (2023)). Russell's "Human Compatible" warns trusting AI output. Anthropic publishes hallucination rates 5-15% even for advanced models.

Is "Fact-Grounding Strategies" suitable for beginners?

Its difficulty level is: Intermediate. No external tools required — it can be practiced directly inside the app.

Subscribe for full access

Subscribe to Plus for full access to this practice and the whole catalogue, or try first — free, no sign-in, no commitment.

Interactive practice locked · 3 more steps locked

7 complete practices (one per domain) are free forever, no sign-in.