Quantifying AI-Infused Science

The aim of scAIence is to quantify whether, how, and with what effects generative AI is changing how researchers write, communicate, perceive, and disseminate science. Through large-scale data, quantitative analyses, and controlled experiments, the project explores the opportunities, dangers, and implications of scientists augmenting their scientific writing with AI.

Since ChatGPT's public release in November 2022, Artificial Intelligence (AI) tools, and especially Large Language Models, have been widely adopted across various domains. Recent studies show that AI-generated content has already entered the world of science: Papers co-written by AI are published, companies release models to assist with producing scientific content, and many researchers started using AI-based tools to help with scientific tasks. Given the fast and increasingly widespread adoption of AI tools in science, it is crucial to understand the implications of this technological development.

Central research questions

scAIence builds on a key hypothesis:

Although AI language models have the capability to generate scientific content that may appear as if humans created it, they lack the ability to make connections and associations between knowledge entities in the same manner as humans. They cannot replicate the social information present in scientific publications created by humans.

We test this hypothesis, specifically addressing the following research questions:

1. How can AI generate scientific writing that is systematically similar to that created by humans?
2. How is knowledge connected in an AI-infused science compared to before?
3. What are the implications of having papers (co-)written by an AI on our systems of scientific metrics and models, which heavily rely on publications records? And what are the biases of an AI-infused science?
4. How do humans relate to AI-generated science?

Funded by:

European Research Council – Consolidator Grant

Project: Quantifying AI-Infused Science

Contact

Principal Investigator: Roberta Sinatra, robertasinatra@sodas.ku.dk

For questions regarding the survey on AI and science, please contact us at: ai-survey@sodas.ku.dk

Researchers