AI Feedback That’s So Good, It Feels Like Cheating (It’s Not)

An AI writing assistant called **Thesa 5.0** is positioning itself as an “academic toolkit” for thesis and research writing—giving structured, field-aware feedback plus a built-in research workflow for finding relevant literature, journals, and conferences. The core value is not just critique; it’s a checklist-style assessment that maps a manuscript’s claims, purpose, evidence, and analysis to what reviewers typically expect, helping writers tighten arguments before submission.

After logging in, users upload a manuscript for a **pre-submission assessment**. The system supports multiple document types—scientific papers, thesis/essay, grant proposals, reports, and bibliographies—and it adjusts feedback based on the selected category and study field (the transcript uses chemistry as an example). Uploading a full thesis is limited by a **10 MB cap**, so the workflow currently favors chapters, drafts, or sections rather than entire dissertations.

Once the upload finishes, the interface presents feedback in a side panel with expandable sections and dropdowns such as **general feedback**, **what works well**, **what can be improved**, and an **overall assessment**. The feedback is framed as detailed and specific enough to resemble a supervisor’s comments—down to pointing out missing elements reviewers look for. For instance, the system evaluates whether the paper clearly states its **purpose**, whether it **summarizes key findings**, and whether it **evaluates advantages and limitations**. In the example provided, the assessment flags that limitations and areas for improvement were not sufficiently addressed, and it notes that alternative interpretations or counterarguments weren’t extensively considered.

The assistant also runs argument-quality checks tied to thesis structure. In the transcript, the “thesis statement” section includes tests such as whether the thesis statement can be challenged (marked green when the statement is supported by presented facts and figures) and whether the essay supports the thesis statement (marked green when alignment is strong). Another feedback area targets **evidence quality**, including cases where evidence is considered missing or where claims are supported but analysis is thin—summarized as issues like weak analysis or insufficient interpretation.

Beyond critique, Thesa 5.0 adds a research layer. In a **PDF view**, it surfaces **resources and collections**, including similar publications and recommended reading. It also provides journal and conference suggestions with metrics like **match factor** and **impact factor** (example values shown include match factor **86%**, impact factor **5.53**, and another listed impact factor **31.8**). A generated section highlights the **research question**, plus **research opportunities** that suggest logical next steps based on gaps detected in the submission. For open-access papers, the tool offers quick scanning via an **abstract digest** with keywords and main claims, and it supports downloading or sharing links for citation workflows.

Overall, the transcript frames Thesa 5.0 as a “researcher-made” system that combines reviewer-style argument checking with literature discovery and publication planning—aimed at improving acceptance odds and reducing the back-and-forth that happens when supervisors or reviewers want clearer purpose, stronger evidence, and more complete analysis before peer review.