AI Validity Checks
At Technology Adoption Barriers (TABS), ensuring the factual accuracy and academic rigor of our content is paramount. To achieve this at scale, we employ advanced Large Language Models (LLMs) to conduct exhaustive, read-only validity checks across our entire public-facing site.
These AI Validity Checks are designed to audit our academic models, organizational frameworks, bibliographies, and general site copy. By leveraging the vast internal knowledge bases of models like Gemini 3.1 Pro, we can cross-reference our claims, dates, and author attributions against established literature with unprecedented speed and accuracy.
The Nature of the Review
The validity review process is strictly read-only. The AI is instructed to act as an independent auditor, scanning the codebase and extracting factual claims without making any modifications. The primary goal is to identify:
- False statements or misrepresentations of academic theories.
- Incorrect attributions of models to authors.
- Inaccurate publication dates or historical timelines.
- Formatting errors in APA citations and bibliographies.
The Prompt
To initiate these reviews, we use a highly specific prompt designed to constrain the AI's behavior and focus its analytical capabilities:
"I want to do a full validity check on the public content on the site. I am looking for false statements and wrong academic findings or other errors. Don't make any changes; your goal is to find and present the factual errors in the content. Then we will address them one by one."
Model Reviews
Below, you will find detailed reports from our AI validity checks, categorized by the model that performed the audit.