I have been playing with LLM tools since the release of ChatGPT 3. There is no doubt that it amazed me with its endless capabilities. Since then, the LLM algorithms and training datasets have improved enormously. Even though it has been less than 3 years since the first release of ChatGPT 3, LLMs have reached the point where they are not only capable of understanding the data on the internet but also interacting with it. Considering this fascinating speed of development, I started to think: Do they even need my initial input to perform complex tasks? In this blog, I will try to answer this question from the perspective of academic research. But for whom wonders the short answer: No, they don’t need us at all.

An academic research, especially in social sciences, involves some fundamental steps such as observation, identifying research question, literature search, hypothesis formulation, experimental design, data collection, analysis, and documentation. Addition to those processes, an academic research should also involves some quality assessment processes such as proofreading and ethical approval. Let’s take a closer look those through the perspective of AI.

Although promising improvements, we are still way behind to an AI that can observe and create complex connections between its knowledge and what it observes. I mean yes, it does observe, identify and make basic pragmatic connections between the objects in the environment. But yet it is not able to build sophisticated interconceptual connections between situations such as an apple fall and the gravity. Since research question formulation rely on identified relations of variables in the observation phase, the systematic progress of research adventure fails. But, let’s assume that the detailed information about the observation and research question were given to the AI tools by a human. At that point, things are getting a bit scarier. Because the current LLMs are already way behind better in reviewing the literature than an above avarage researcher. It can find, read, and understand the academic articles in specific question in an enormous scale in seconds.