AI models can only access Open-Access scholarly materials because they do not have direct access to subscription-based databases or paywalled content (such as those provided by academic libraries). Here’s why this is a limitation:
Why AI Can Only Access Open Sources
- AI models are trained on publicly available data, which includes open-access journals, preprints, and freely accessible articles.
- They cannot retrieve or read content behind paywalls or in proprietary databases (e.g., JSTOR, ProQuest, EBSCO, Elsevier).
- This means AI-generated references often come from sources that are free online, which may include abstracts, summaries, or incomplete versions of scholarly work.
Why This Is a Weakness
- Incomplete Information
Abstracts provide only a high-level summary, not the full methodology, data, or nuanced discussion. Relying on them can lead to superficial analysis.
- Quality and Credibility Issues
Open-access sources vary in rigor. Some may not be peer-reviewed or may come from predatory journals, reducing the reliability of the literature review.
- Limited Scope
Subscription-based journals often contain the most authoritative and current research. Without access to these, AI-generated content may miss critical studies and perspectives, and results could potentially be out of date or provide obsolete information.
- Citation Accuracy Problems
AI may generate DOIs or links that point to abstracts rather than full-text versions, making it harder for readers to verify or follow up.
