ChatGPT in scientific research

AI tool, ChatGPT, was listed as a co-author on a research article (See the recent article in Nature: https://lnkd.in/gyx9cUvJ). With that all hells broke loose and the use of ChatGPT in scientific research has been a topic of discussion and debate in various scientific circles. On one hand, there are researchers who see the potential benefits of using ChatGPT, such as increased efficiency and productivity, improved accuracy, and the ability to handle large amounts of data. On the other hand, some researchers have expressed concerns about the potential limitations of using ChatGPT.

What are LLMs?

Large Language Models (LLMs) are a type of artificial intelligence that are specifically trained to understand and generate human language. They are called “large” because they have been trained on massive amounts of text data, which allows them to generate coherent and convincing responses to a wide range of questions. LLMs work by using a technique called “deep learning” to analyse the patterns and relationships between words, phrases, and sentences. Over time, the model learns to recognise common patterns and generate appropriate responses. In simple terms, LLMs are like really advanced language computers that can understand and generate human speech and text. This makes them incredibly useful for a wide range of applications, including natural language processing, data extraction, predictive modelling, machine translation, text to text generation, question-answering systems, chatbots and virtual assistants.

Several LLMs are currently available for use. The popular tools are OpenAI’s GPT3, Google’s BERT, Microsoft’s Turing NLG, Facebook’s RoBERTa, Alibaba’s ERNIE, Amazon’s SageMaker Ground Truth, Hugging Face’s Transformers and so on. However, currently ChatGPT is the most popular one. ChatGPT is a computer program created by OpenAI that can generate written text in response to prompts provided by users. OpenAI is a research organisation founded in 2015 by Elon Musk, Sam Altman, Greg Brockman, Ilya Sutskever, and Wojciech Zaremba to focus on developing and promoting friendly AI that benefits humanity. It is part of a larger family of AI tools called language models, which are designed to understand and generate human language. In simple terms, ChatGPT works by using a large database of texts and language patterns to generate responses to user prompts. When given a prompt, it analyses the input and generates a text response based on what it has learned from the training data. The result is a text that resembles human writing, and which can answer questions, generate stories, or provide information.

How ChatGPT is useful in research?

For decades, software tools have been used to collect, compile, and analyse research data. With the advent of AI, novel tools like ChatGPT have emerged, and the possibilities are immense. They can be used to compile research, generate content, and do keyword research. By leveraging ChatGPT, researchers can streamline their workflow and focus on their research questions. Additionally, ChatGPT can provide new perspectives and insights by combining and synthesising information from multiple sources, helping researchers make better decisions and advance their research more quickly. It can also be used to identify competitors and their strategies, as well as save researchers time.

They can assist in research by providing natural language generation capabilities, such as the ability to generate human-like text, summaries, and translations which can also be used to create research paper and reduce the time spent on mundane tasks like spelling, grammar, and other aspects of writing. They can get quick answers and insights without having to spend time searching for information, formatting documents, or manually summarising data. ChatGPT’s ability to generate coherent and relevant responses based on a prompt makes it particularly useful in the scientific community, where there is a need for quick and accurate information retrieval and analysis. Moreover, its size and training data allow it to hold a vast amount of knowledge and understand complex scientific concepts, making it a valuable tool for researchers.

The immense possibilities with ChatGPT are summarised below:

How ChatGPT helps a researcher

What can be wrong with tools like ChatGPT?

However, researchers and publishers are wary of the use of AI tools to generate research results because they are not always reliable or authentic, and they can be misused or used for malicious purposes. They undermine the thinking process by reducing the need for critical thinking and problem-solving skills. such as the risk of relying on AI-generated results without proper validation, the potential for biases in the data used to train the model, and the need for robust data management practices to ensure the privacy and security of research data. Writing tools using AI are not foolproof and should be used to supplement the writing process, not replace it. However, there is a general consensus that the use of ChatGPT and other AI tools in scientific research should be approached with caution, and that proper protocols and standards should be established to ensure the quality and reliability of AI-generated results. Last month, both the Science and Nature journals declared their positions on the use of ChatGPT to generate articles. While it can not be considered an author, it is important to acknowledge the use of any such tools or resources in the acknowledgement section of the research article which is an age-old practice.

Conclusion

Despite the recent controversies, I am sure that AI tools like ChatGPT are going to be increasingly used in academics and research in future. They are becoming increasingly popular in a variety of applications, from customer service and virtual assistants to creative writing and scientific research. Additionally, the increasing popularity of chatbots and virtual assistants in various industries has also driven interest in ChatGPT and other similar models. Overall, the combination of its advanced language processing capabilities and practical applications in various domains make ChatGPT a widely discussed topic in the scientific community. I am of the opinion that tools like this are highly useful for for non-English speaking researchers.

OpenAI provides access to ChatGPT through its API, which is not free but requires a subscription. The API can be used to build custom language models or to integrate ChatGPT into other applications.

The URL to access OpenAI’s API is: https://beta.openai.com/docs/guides/how-to-use-openai