iDEC Policy on the Use of Artificial Intelligence

This policy sets out how AI may be used in iDEC projects, reports, presentations and judging activities. Its purpose is not to restrict useful tools, but to protect scientific integrity, transparency and accountability.

AI may assist a project, but the team remains responsible for the accuracy, validity and originality of everything submitted under its name. The use of AI does not transfer responsibility to the tool or its provider.

Teams may use AI wherever it provides legitimate assistance to research, engineering or communication. Examples include:

• brainstorming research questions, experimental approaches and design options;
• literature searching, summarising and language support;
• writing, editing, proofreading and translation;
• data analysis, modelling, prediction and computational design;
• generation or debugging of computer code;
• protein, nucleic-acid, pathway or other bioengineering design;
• preparation of posters, slides, websites, diagrams and graphical material; and
• use of AI or machine learning as an integral part of the research itself.

Using AI will not, by itself, disadvantage a team during judging.

AI must not be used to fabricate, invent or falsify experimental results. AI-generated material must never be presented as genuine experimental evidence.

This includes fabricated or misleading microscopy images, gels, spectra, sequencing results, assay results, photographs, numerical datasets, graphs or other records of experiments that did not take place.

AI-generated illustrations, simulations and predictions are allowed, but they must be clearly presented as such when there is any risk that they could be mistaken for experimental observations.

AI may assist with scientific writing, coding and visual communication. Teams must, however, check the output before using it.

Scientific claims, references, numerical values, sequences, protocols and quotations must be verified. Fabricated or non-existent references must not appear in an iDEC submission.

AI-generated or AI-assisted code is permitted, but teams should understand its role and make reasonable efforts to test it. AI-generated diagrams, artwork and other non-data graphics are also permitted, provided that illustrative material is not represented as scientific evidence.

Substantial use of generative AI or AI-assisted tools must be disclosed in the Final Report. The purpose of disclosure is transparency, not penalty.

A short AI Use Statement should identify the tool or type of tool used, what it was used for, and how the team reviewed or verified the output. Routine spelling correction or basic formatting tools do not require detailed disclosure.

iDEC welcomes projects in which AI or machine learning is part of the research itself, including AI-assisted sequence design, protein engineering, evolutionary optimisation, phenotype prediction, experimental design, high-throughput analysis and design-build-test-learn workflows.

Such projects will be judged on the same principles as other iDEC projects: scientific quality, originality, evidence, understanding and potential impact. Computational predictions should be distinguished clearly from experimental validation.

Teams must be able to explain and defend the work they submit. Judges may ask how AI was used, how outputs were checked, how AI-assisted designs were developed, or how AI-generated code works.

Unless specifically permitted for an iDEC activity, participants should not use real-time generative AI to produce answers during a judged presentation or question-and-answer session.

The following are inconsistent with iDEC scientific integrity:

• fabricating or falsifying data or experimental evidence with AI;
• altering scientific evidence in a materially misleading way;
• presenting AI-generated results as experiments that were actually performed;
• knowingly using fabricated references or false scientific claims;
• concealing substantial AI use where disclosure is required; or
• using AI to misrepresent work performed or understood by the team.

Serious breaches may result in affected material being disregarded, changes to assessment, disqualification from an award, or referral to the iDEC Judging Committee or iDEC HQ.

iDEC treats AI literacy as part of modern scientific education. Participants should learn not only how to use AI, but also how to question its output, recognise its limitations and verify its results.

AI should support scientific reasoning, evidence and creativity—not replace them.