I work with organizations to strengthen how AI-assisted outputs are evaluated, verified, and integrated into scientific and strategic decision-making.
AI accuracy is often framed as a problem of hallucinations, which are fabricated sources or incorrect claims. In practice, those errors are usually easy to catch.
The more consequential challenge is inaccuracy, where claims are accepted without fully examining the evidence that supports them, contradicts them, or complicates their interpretation. When conflicting findings are not surfaced and resolved, AI-assisted work can look coherent while quietly drifting away from defensible conclusions.
I train students and researchers to rigorously and comprehensively check evidence and citations, as well as aligning AI-assisted work with academic expectations.
Rigorous AI practice centers on systematically checking sources, citations, and claims, ensuring that every referenced piece of evidence can be traced, validated, and defended.
By clarifying how evidence is selected and verified, rigorous AI use helps students and researchers meet supervisor expectations and reduce revision friction.
When AI outputs are evaluated through explicit critical-thinking frameworks, users can focus on sound scientific judgment rather than editing language fluency.
As AI-assisted work moves across drafts, collaborators, or evaluators, rigorous verification practices help preserve methodological integrity and intent beyond the initial prompt or tool.
Why minimal AI works
AI-supported critical thinking is most effective when the method is simple, repeatable, and disciplined. I train teams to use a minimal AI approach through a small set of context-based questions that keep decision-making anchored to evidence rather than fluent output.
What the framework focuses on
Instead of complex frameworks or advanced tooling, my minimal AI approach focuses on slowing reasoning down: clarifying what is known, what is uncertain, what depends on assumptions, and what follows logically.
What this supports in practice
This minimal AI approach is introduced as a basic fact-checking aid, not an authority or a shortcut. Across academic, clinical, and leadership settings, this approach reduces cognitive load while increasing judgment quality, which supports decisions that are clear, defensible, and aligned with real-world constraints.
Rigorous AI use helps students and researchers verify evidence, surface assumptions, and align their work with supervisor expectations without weakening scientific standards.
Ri Xu, PhD provides evidence-based scientific writing and AI-assisted workflow support for science-led organizations and technical teams. Her work strengthens scientific rigor, reasoning accuracy, and clarity in fast-moving environments where small errors or poorly used AI tools can undermine good science.
Based in Montreal and working internationally, Ri partners with scientists, lab leaders, and technical professionals to translate complex work into clear, defensible communication and efficient, rigor-first workflows. Backed by an engineering PhD and over four years of experience, she helps teams use AI critically without compromising scientific standards.