Very nice and timing question! Indeed, generative AI tools like ChatGPT are used in several industries to provide quick and effortless suggestions (I think of consulting, marketing, even literature and copywriting).

I am not sure if it works for academic purposes, though. When prompted with requests on recent research, ChatGPT tends to provide answers that still do not pass critical investigation. For inquiry purposes on the capabilities of ChatGPT, I have tried several times and got: i) blatant stereotypes, ii) references to non-existing works (even with fake DOIs), iii) subtle inaccuracies that nonetheless took quite some time to debunk.

Reasons for ChatGPT to fail on me may include: i) ineffective prompts from my side, ii) limited ChatGPT database scanning on recent years (especially on the academic literature), iii) poor training dataset from academic literature (from all times; this could be related e.g. to need for subscription for many journals), iv) limited ready-made responses for testing the network, v) the intrinsic statistic-based way of neural networks to work, that privilege likely (read: closer to the "mean") chains of words rather than out-of-the-box answers, which are those needed to advance academic research.

In a nutshell: maybe time will make the AI networks better, and with better responses, but they are currently very limited in scientific research. They are useful to get a bird-eye view on a topic and to summarise some trends, but I'd be very careful in demanding more - and if I'd do, like asking for suggestions on scientific questions, I'll doube and triple check it.

PS: My investigation was just preliminary. A full study on this would be very appreciated, to make a good statistic out of my quick attempts and to scan for effective prompts :)

I believe it's quite evident that AI tools can be used for research purposes. Researchers from previous centuries relied on books to gather knowledge, which they then used to develop new concepts. Nowadays, we use search engines to gather data and information from the internet. This is entirely within legal bounds, as no one is born knowing everything; learning is an ongoing process that occurs every millisecond.

In a similar vein, AI models can be employed to gather, process, and review data, thereby saving time and simplifying work. This technology also aids in reshaping our thoughts, given its vast collection of data in comparison to a single human brain. Using AI for research undoubtedly eases workloads.

I am aware that many might not agree with my perspective. However, it's essential to recognize the continuous evolution of our era. Look, we've moved away from using CD players (I might not even be familiar with all of its uses), and now we prefer alternatives like YouTube and other platforms. Similarly, it's crucial not to confine ourselves to the past, but to remain with the time. Time won't stay a moment for anyone. We must have to run with the flow of the era. We have to keep ourselves updated with time. Otherwise, we will loose. It's reality!

Daniele Proverbio Thank you for your comprehensive feedback on the limitations and potential of AI tools like ChatGPT in academic research. Your observations about inaccuracies and the need for critical evaluation are certainly valid. These limitations do warrant caution, especially when using AI-generated information for scholarly pursuits.

However, as a clinician who is also involved in basic science research, my experience has shown that ChatGPT can be highly useful, particularly in the realm of research methodology. When equipped with effective prompts and relevant background information, the tool can provide insightful preliminary suggestions that have aided my work. From generating potential research questions to helping refine methodologies, ChatGPT has been a valuable resource for me.

I do acknowledge the issue with inaccurate citations and DOI numbers. One way to circumvent this could be by integrating ChatGPT with specialized citation management plugins like ScholarAI, Mendeley, or EndNote. While ChatGPT itself may not be equipped to verify citations, its compatibility with these plugins could offer a well-rounded solution for academic research. By doing so, we can leverage the tool's generative capabilities while ensuring that all references are accurate and up to the academic standard.

So, while generative AI tools are far from perfect, when used responsibly and in conjunction with other specialized software, they can become quite valuable. It's all about understanding their limitations and advantages and tailoring their use accordingly.

Lastly, I fully support your suggestion for a comprehensive study on this topic. Such a study would not only validate anecdotal experiences but also help us understand how to optimize the use of AI tools in academic settings.

Al Mustafiz I wholeheartedly agree with your optimistic outlook on the role of AI in modern research. Your point about the evolutionary trajectory—from traditional books to search engines, and now to AI—hits the nail on the head. It's a natural progression in the way we gather and process information, and it's a shift that I believe researchers should embrace, not resist.

As a clinician who is also engaged in basic science research, I've found AI tools like ChatGPT invaluable for formulating research questions and refining methodologies. The time-saving benefits you mentioned are particularly impactful for me. Instead of combing through dozens of academic papers, a well-crafted prompt can yield a useful summary or even generate hypotheses for further investigation. This frees up time for the hands-on, experimental aspects of my work, which is an enormous benefit in a field as time-sensitive as medical research.

Your observation about the need to "run with the flow of the era" is spot on. It's not just about keeping up with technological advances; it's about harnessing them effectively to accelerate the pace of discovery and innovation. And as these tools become more sophisticated, I anticipate we'll only find more ways to integrate them into our research efforts responsibly.

So, yes, while some may still need convincing, I believe those of us who are already leveraging these tools are trailblazers in what is undoubtedly the future of research.

İlter Alkan I wholeheartedly agree with your comment. As long as limitations are known and care is applied, as any other tool AI can and should be used to improve our doing and the progression of research

AI models like ChatGPT find utility in research. They serve as valuable tools for investigating language comprehension, generation, and interaction. Researchers employ these models to analyze linguistic patterns, validate hypotheses, and innovate algorithms. The models also aid in data analysis, content creation, and information retrieval. However, it's essential to ensure proper utilization and methodological rigor for accurate research outcomes.