Attention points and tools to check research grant applications

• A research application is a scientific product and must meet the same standards as a scientific publication.

• Researchers should be aware of the risks of using AI, since the output of generative AI (GenAI) can be biased, invalid, inaccurate or completely wrong (“hallucinated”).

• Researchers must be able to vouch for their own contribution to the research application and request the coordinator to ensure transparency in all contributions.

Relevant issues to consider for proposal check, especially in applications with several and non-familiar partners

1. Apply DTU procedures for proposal preparation

DTU’s research applications to both public and private funds have to be submitted in accordance with DTU’s financial regulations and in accordance with DTU’s principles for good scientific practice.

2. Request traceability mechanism from coordinator

Before joining a proposal, request that the coordinator uses an internal traceability mechanism for contributions (e.g., OneDrive shared folder with history, accessible only to logged-in users). Remind the coordinator of the new guidelines on Generative AI stated in Horizon Europe proposal templates.

For further reading, see EU’s Living Guidelines on the responsible use of generative AI in research.

3. Ensure transparency

It is your responsibility to ensure that all AI-generated content used in your contribution to the research application is:

- relevant, valid and correct (e.g., correct references),
- does not include proprietary or protected information introduced by the AI,
- does not plagiarize and respects authorship by citing where appropriate.

See more on research integrity and AI-assisted research.

4. Other aspects

• Check that the researchers listed in the team exist

• Check that the list of publications (and/or software), projects and infrastructure is correct

• Check that the budget is correct

In case of suspected irregularities:

• Inform DTU through the formal process (see "Guidelines for handling suspicions regarding research misconduct and questionable research practice at DTU").

• Clearly state that the suspected misconduct refers to another organization, not DTU.

• Inform your immediate manager of the situation.

Available tools

Reference validation

Use Google Scholar (or other academic tools like Web of Science or Scopus), to verify the validity of each reference by pasting the citation.

DTU subscribes to scite.ai. The tool can check if indexed papers are retracted or whether the citation for the paper is a supporting, mentioning or contrasting citation. – Note that currently, scite.ai only indicates if a reference is retracted, and does not include non-existent or controversial references in its report.

Save the results of the reference validation in your records.

Checking for text similarities and AI-generated or paraphrased text

DTU researchers can get access to the text screening tool, iThenticate. If you have severe concerns that text in an application has been plagiarized, or that generative AI has been used, iThenticate can be used as a tool. Note that iThenticate only flags texts if more than 20% of a text is AI generated or AI paraphrased, and that the AI writing assessment may not always be accurate. Also be aware that iThenticate not finding AI generated or paraphrased text does not mean it isn’t there, and that false links etc. will most likely not be flagged. 

Read more here.

If you make use of the tool, remember to save the text screening report in your records.

Image check

Use Google Reverse Image Search for each figure in the proposal to check for potential plagiarism and/or copyright infringement. If a match is found, ensure that all rights to the figure are cleared and that there is proper referencing in the figure caption.

Document the outcome of the image search.

Workflow for Declaration for research grant application

You find an electronic workflow for Declaration for research grant application on Research Application Declaration - Power Apps.