At FACTNet and its affiliated websites (the Universe Institute, Job One For Humanity, the Way of the Universe, Universe Spirit, and Universe College), original human thinking, credible sources, critical review, and carefully controlled AI assistance are combined. At FACTNet and at its affiliated websites, artificial intelligence is a research assistant—not the researcher, author, or final decision-maker.

Our researchers determine the questions, conduct original research, evaluate the evidence, develop the interpretations, write the initial drafts, and approve every conclusion we publish. We may use AI at selected stages to help locate credible information, improve clarity, identify missing considerations, and test our conclusions. However, our people remain responsible for the reasoning, accuracy, integrity, and final content of every publication.

Our Research and Publication Process

1. We define the question and conduct original research

Every project begins with people. Our researchers identify the problem, establish the purpose and scope of the inquiry, and contribute their own knowledge, experience, analysis, and original research.

Depending on the project, we may also use AI-assisted searches to help locate credible books, scholarly studies, reports, data, and other relevant sources. AI can help us search more broadly and efficiently, but it does not determine which sources or claims we accept.

2. We critically review the information

Our researchers examine the information gathered and decide what is credible, accurate, relevant, and useful. We also identify material that may be incomplete, poorly supported, misleading, outdated, or inaccurate.

We do not assume that information is reliable merely because an AI system found or summarized it. Important claims are checked against the original sources whenever practical, and human judgment determines what is included.

3. We integrate original analysis and look for what may be missing

We combine the strongest existing research with our own original work and analysis. We then examine the developing findings from multiple perspectives, asking:

• • What important contexts may have been overlooked?
  • What underlying processes may not yet be visible?
  • What relationships, interactions, or dependencies may have been missed?
  • What past, present, or emerging transformations could change the meaning of the evidence?

This step helps us move beyond simply collecting information toward developing a more complete and useful understanding of the subject.

4. Our researchers assemble and write the first draft

People organize the evidence, determine the structure of the argument, interpret the findings, and write the initial draft. The central ideas, judgments, explanations, and conclusions come from our research team—not from an automated system.

5. We may use AI to improve clarity and accessibility

After the first draft is written, we may ask AI to suggest clearer grammar, phrasing, organization, or explanations. Our usual goal is to make complex subjects understandable to readers with approximately a first-year college education, without oversimplifying the ideas or weakening their accuracy.

When useful, we may also ask AI to identify specialized or unfamiliar terms for a glossary and to propose frequently asked questions about concepts readers may misunderstand. Our researchers review, revise, and approve all such material before it is used.

6. We document the sources

Before publication, we assemble and review the bibliography or reference list. This allows readers to examine the sources behind important claims and helps make our research process more transparent.

7. We red-team the final draft

As a final quality-control step, we may ask AI to examine the document adversarially. This process is often called red teaming.

We ask it to identify credible contrary evidence, competing interpretations, weaknesses in our reasoning, unsupported assumptions, missing qualifications, and strong objections to the positions we present. Our researchers then review these challenges and revisit the evidence.

When a conclusion is well supported, we may strengthen its explanation and documentation. When it is not adequately supported, we qualify, revise, or remove it.

The purpose of red teaming is not to make every viewpoint appear equally valid. It is to ensure that our conclusions can withstand serious, evidence-based criticism.

Our Commitment

FACTNet uses AI to extend the reach and efficiency of responsible human research—not to replace human thinking or accountability.

Our work is a highly interactive, human-led process. Our researchers:

• originate the questions and much of the research;
• provide original thinking and analysis;
• evaluate the credibility and usefulness of information;
• write and organize the initial drafts;
• review every AI-assisted suggestion;
• test conclusions against contrary evidence; and
• make every final editorial and publication decision.

AI can help us search, compare, clarify, and challenge. People lead the inquiry, interpret the evidence, accept responsibility for the conclusions, and decide what FACTNet ultimately publishes.

In short: AI assists the process. Human researchers lead the work and remain accountable for the result.