Go back

Arma 2024: Use of AI in funding must be open and collaborative

 Image: CoreDesignKEY, via Getty Images

Technology—plus transparency and integrity—can transform grant applications, say Anna Aston and Stuart Grey

Many jobs are in the process of moving from data-driven workflows to those driven by artificial intelligence. In a data-driven approach, humans make informed decisions by extracting relevant insights from information. In contrast, AI-driven workflows shift the focus to processing and acting on insights autonomously. 

Systems incorporating AI tools, ranging from classification to autonomous agents, will become increasingly common, further enhancing workflow efficiency. When it works well, this does not eliminate human decision-making but significantly enhances its quality, reducing biases and enhancing efficiency. 

Daily tasks

Research managers and administrators can probably think of tasks they perform every day that are candidates for automation. UK association with Horizon Europe, for example, is hugely valuable, but it has also created more administrative work for already stretched research offices. Automating the data-intensive aspects of such work would free up the expertise of researchers and administrators to provide oversight and strategic direction, particularly for decisions requiring contextual understanding, ethical considerations and building collaborations.

Beyond general large language models such as GPT-4, there are a range of widely understood, nonproprietary, AI and machine-learning techniques that could be applied to the management of research funding applications. 

For example, classification techniques based in natural language processing can help to identify and categorise funding opportunities and match applications to specific elements of funding calls. Techniques such as graph algorithms can analyse collaboration networks to identify potential research partners, while sentiment analysis can gauge the tone of research proposals, providing insights into their potential reception by reviewers.

Tools such as Elsevier’s SciVal and Digital Science’s Dimensions already analyse research outputs and identify funding opportunities. Last month, OpenAI launched ChatGPT Edu, aimed at schools and universities. Companies including IBM and Microsoft are also developing AI solutions for large enterprises, many of which can be adapted for research environments.

At the proposal-writing stage—arguably the most critical step in the application process—AI tools can save time and improve quality. This includes generating drafts, summarising key points and enhancing clarity. 

Again, the human element remains vital. The European Research Council, for example, while recognising AI’s potential to enhance the funding process, has warned researchers to maintain academic integrity and authorship responsibility, ensuring that automation does not compromise authenticity. 

Once a proposal is written, it will undergo internal review before submission. AI can enhance this step by providing objective assessments and identifying potential weaknesses. AI-driven review systems can simulate the evaluation criteria of funding bodies, offering researchers feedback to refine their proposals.

Finally, AI can streamline the administrative tasks involved in institutional approval and submission, ensuring that documents are complete and formatted correctly. AI tools can also monitor submission portals for updates and deadlines.

As well as the potential benefits, integrating AI in research funding applications presents challenges such as data privacy, ethical considerations, and the need for transparency in algorithms. The rise of these tools brings concerns about data privacy and being locked in to one vendor. Many AI tools are proprietary, potentially limiting access and interoperability. Visibility and trust are also important, because ethical, transparent data sourcing is essential for training reliable AI models. To build trustworthy AI tools, institutions need clear guidelines on using data.

Open and collaborative

A sustainable ecosystem for using AI in research funding should be open and collaborative. This involves public-private partnerships, developing and building open-source AI tools, and ensuring equal access to technology for all institutions. Continuous education and training for researchers and administrators are also crucial for effective AI integration.

The transition from data-driven to AI-driven workflows represents a significant evolution in the research funding application process. As funders continue to explore and regulate the use of AI, the research community should embrace this technology while maintaining a commitment to integrity and transparency.

The future of research funding applications is intertwined with advancements in AI, promising a more streamlined, innovative and equitable process.

Research Professional News is media partner for the 2024 conference of the Association of Research Managers and Administrators, held from 18 to 19 June in Brighton

Anna Aston is a section manager in the faculty of medicine at Imperial College London. Stuart Grey is a senior lecturer in engineering systems design at the University of Glasgow and the founder of StudentVoice.ai. They will be speaking at at the 2024 conference of the Association of Research Managers and Administrators in Brighton on Wednesday 19 June

This article also appeared in Research Fortnight