Cardiothoracic research through artificial intelligence without hallucinations

A new plugin for ChatGPT, ScholarAI, enables searching academic literature about various topics without hallucinations (mentioning fictitious papers).

In the realm of cardiothoracic surgery research, a transformation is on the horizon, fueled by the dynamic synergy of artificial intelligence (AI) and natural language processing (NLP). Spearheading this paradigm shift is ChatGPT, a new tool that has taken center stage. In the face of existing obstacles and constraints, the potential gains tied to the integration of ChatGPT within cardiothoracic surgery research are promising.

By using a strong language model that has been trained on a variety of data sources, such as academic literature and medical textbooks, ChatGPT harnesses the power of machine learning. Its strength is in its ability to examine and comprehend vast amounts of data, providing significant insights that could revolutionize cardiothoracic research and practice. While ChatGPT has mostly been utilized for text production and other NLP tasks, such as language translation, interest in its potential uses in health care is growing.

Nevertheless, hallucinations (mentioning fictitious papers) are common in ChatGPT responses. Because of that, we have tried also the plugin ScholarAI which provides us with better results than ChatGPT itself. The plugin was developed by a team around Duke University. This solution provides real papers from respected journals.

ScholarAI Premium can also extract tables and figures from papers and will include them in any requests for paper content. This solution is still in beta, results may vary from paper to paper. And the possibility of Paper Question Answering has the ability to search a PDF to find the best parts of the paper to answer a given question.

The imperative now rests with dedicated researchers and adept clinicians to chart new courses, ingeniously employing this technological marvel to amplify patient outcomes. By orchestrating collaborative endeavors and harnessing the intrinsic capabilities of AI and NLP, the landscape of cardiothoracic surgery stands poised for remarkable advancement, promising a brighter horizon for patients across the globe.

The paper is published in the journal Interdisciplinary CardioVascular and Thoracic Surgery.