Core Philosophy

Evidence Over Speculation

Our conclusions are driven by verifiable data, but we explicitly acknowledge the limits of evidence-based analysis. Many important questions about AI involve values, ethics, and uncertain predictions that can’t be resolved through data alone.

Intellectual Humility

  • We admit when we don’t know something
  • We correct errors promptly and transparently
  • We acknowledge competing interpretations of evidence
  • We distinguish between confidence levels in our claims

Our Perspective and Its Limitations

We do not claim a purely objective “view from nowhere.” That is an impossible standard. Our commitment is not to an abstract neutrality, but to transparency about our own perspective and methodology.

Our analysis is guided by:

  • A Sociotechnical View: We believe AI cannot be understood through technical metrics alone, but must be analyzed in its social, economic, and political context.
  • A European Regulatory Lens: Our initial focus is grounded in the EU’s approach to rights-based technology regulation.
  • An Acknowledgment of Values: We recognize that interpreting evidence involves value judgments. Our core value is ensuring that technological progress does not come at the cost of public accountability and individual rights.

Content Standards

What We Cover

  • Explainers: Foundational concepts that shape AI discourse
  • Investigations: Claims that can be fact-checked or analyzed with available evidence
  • Analysis: Current events with broader implications for AI policy and society
  • Response pieces: Counter-arguments and alternative perspectives to our published content

What We Don’t Cover

  • Pure opinion pieces without evidence base
  • Breaking news without sufficient context
  • Technical tutorials (plenty of good resources exist)
  • Content primarily aimed at AI researchers (not our audience)

Evidence Standards

Source Hierarchy

  1. Primary sources: Research papers, official documents, first-hand accounts
  2. Expert analysis: Credentialed specialists commenting within their expertise
  3. Quality journalism: Well-sourced reporting from established outlets
  4. Industry sources: Vendor claims, marketing materials (treated skeptically)
  5. Social media: Only for documenting public statements, never as evidence

Verification Process

  • All major factual claims must have citations
  • Numerical data verified from original sources when possible
  • Expert quotes confirmed directly when feasible
  • Correction policy: errors fixed within 24 hours with prominent notation

Bias Management

Bias Mitigation Strategies

  • Devil’s advocate requirement: Every major piece must seriously engage with strongest counter-arguments
  • Perspective diversity: Actively seek contributors from affected communities
  • Red team reviews: Major investigations reviewed by subject matter experts who disagree with conclusions
  • Community feedback: Public comment periods for controversial analyses

Handling Uncertainty

Confidence Levels

We explicitly label our confidence in major claims:

  • High confidence: Multiple independent sources, peer review, established consensus
  • Medium confidence: Good evidence but some gaps or disagreement among experts
  • Low confidence: Limited evidence, early-stage research, or significant expert disagreement
  • Speculation: Clearly labeled when we venture beyond available evidence

“What We Don’t Know” Sections

Every major investigation includes acknowledgment of:

  • Key data that’s unavailable or unreliable
  • Important perspectives we may have missed
  • Assumptions we’re making that might be wrong
  • Questions our analysis can’t answer

Controversial Topics

Balanced Coverage Doesn’t Mean False Balance

  • We don’t give equal weight to all viewpoints regardless of evidence
  • We distinguish between legitimate scientific disagreement and fringe positions
  • We provide context for minority positions (why they’re minority positions)

Engaging with Criticism

  • We welcome substantive criticism and publish response pieces
  • We distinguish between good-faith criticism and bad-faith attacks
  • We don’t platform known bad actors regardless of “balance”
  • We engage seriously with criticism even when we disagree

Ethical Guidelines

Harm Reduction

  • We consider potential harms from our coverage, especially to vulnerable communities
  • We don’t amplify dangerous misinformation even when fact-checking it
  • We provide context that helps readers make informed decisions
  • We avoid sensationalism that might distort public understanding

These principles are living guidelines, updated based on experience and community feedback.