AI Usage Policy Statement for the b2b Newsroom

09-18-2024

AI Usage Policy Statement for the b2b Newsroom

Introduction


This policy outlines our approach to using artificial intelligence (AI) in content creation processes. It is designed to ensure transparency, maintain editorial integrity, and uphold the highest standards of journalism while leveraging AI technologies to enhance our newsroom capabilities.

AI Usage Tiers


We categorize AI usage in our content creation process into five tiers:
  1. No AI Usage (0%): Content created entirely by human journalists without any AI assistance.
  2. Minimal AI Usage (<10%): AI used for minor tasks such as headline optimization or SEO improvements.
  3. Light AI Usage (11-35%): AI assists with research, fact-finding, data analysis, and light content optimization. The majority of the content is still human-generated.
  4. Moderate AI Usage (36-50%): AI plays a significant role in content creation, such as generating first drafts or providing extensive background information. Human editors substantially revise and verify the content.
  5. Extensive AI Usage (>50%): AI is the primary content creator, with human oversight and editing.

Quality Control Measures


  • For content in Tiers 4 and 5, at least one (1) designated in-house editor, trained to identify factual inaccuracies, hallucinations, and other discrepancies, must review the content before publication.
  • Tiers 4 and 5 should ONLY be used when AI is provided ample human-vetted source material. Control and designation of source material maintains source control and an audit trail in case of a need for corrections and updates. Designated source material also streamlines quality control, making it easy to cross check quotes, facts and other claims in the generated content.

Appropriate Use Cases for Tiers 4 and 5


AI-generated content (Tiers 4 and 5) is appropriate for:
  1. Audio transcription summaries
  2. Converting press releases and corporate announcements into brief news articles
  3. Creating brief news articles based on human-provided source material and strict guidelines

Restricted Use Cases for Tiers 4 and 5


AI should not be extensively used (Tiers 4 and 5) for:
  1. Op-eds, commentaries, or opinion pieces
  2. Thought leadership articles
  3. Content that could significantly impact the subject (both positively or negatively)

Attribution Guidelines


To maintain transparency with our audience, we use the following attribution statements:
  1. For Tiers 2-3: "This article was written by human journalists with minor assistance from AI tools."
  2. For Tiers 4-5: "This article was generated with assistance from AI, under human oversight and editing."
    • For Content Vetted via Edisource Quality Control Check: “This article was written with assistance from AI, and vetted in via a 3-person editorial check to cross-check all claims, facts, quotes and other statements.”
    • Audit Trail: For Tier 4 and 5 content, a comprehensive audit trail is generated so that an archive of all original source material and the unaltered AI response is preserved.

Ethical Considerations


  1. Transparency: We are committed to being open about our use of AI in content creation.
  2. Accuracy: All AI-generated content undergoes human verification to ensure factual accuracy.
  3. Fairness: We strive to prevent and eliminate bias in AI-generated content through careful oversight and diverse human input.
  4. Privacy: We respect data privacy laws and ethical standards when using AI for data analysis or content creation.
  5. Accountability: Human editors and journalists remain accountable for all published content, regardless of AI involvement.

Continuous Evaluation


This policy will be regularly reviewed and updated to reflect advancements in AI technology and evolving industry standards.

Conclusion


By adhering to this policy, we aim to harness the benefits of AI in our newsroom while maintaining the trust of our audience and the integrity of our journalism.