Artificial Intelligence and PhDing

A survey was performed and the following five questions were asked:

Are you familiar with Artificial Intelligence (AI) applications and their potential uses in your field of study? Have you ever used AI applications in your academic work, such as Chat GPT or other language models? Do you think that having access to AI technology like Chat GPT in the past would have been useful for your research or academic work? Are you planning to use AI technology like Chat GPT in the future, and if so, how do you envision using it? Do you have any concerns about the potential displacement of jobs or ethical implications of using AI technology in academic work?

After sending the poll, a total of 25 participants took part in the survey. About the sample. The participants reside in the following countries 37.5 % in Austria, 4.2 % in Colombia, 8.3 % in Denmark, 4.2 % in Indonesia, 33.3 % in Ireland, and 12.5 % in the Netherlands. Regarding if the surveyed is a Ph.D. student, another type of academic, or simply not doing a Ph.D.: 8 % is in their first year, 20 % on their second year, 32 % in their 3^rd year, 28 % in their 4^th year, while 8 % declared as other and 4 % as not doing a Ph.D.

The poll gave the following results: 60 % are familiar with AI, 36 % have utilized AI language models such as Chat GPT for their academic work, 68 % agreed that AI technology would have been beneficial for their research and academic work, 56 % are planning to utilize this tool in the future and 68 % have concerns about the potential displacement of jobs or ethical implications of using AI technology in academic work.

Three open questions were asked. The first one was which AI applications did the person know. ChatGPT was a common answer. Other mentioned applications were: Grammarly, Deepl, Ernie Bot, Bard, Consensus, and Bing AI.

Other elaborate answers:

·         Neural Networks (NNS) for recreating potential energy surfaces in molecular dynamics, NNs for fitting reaction rates in chemical systems, and NNs for identifying/fitting flow fields from CFD images.

·         There is a wide range of AI applications, language processing models, Computer vision, Machine learning, and deep learning.

·         I work in organic synthesis. I am aware that AI can be used for planning synthetic routes but have quite a low level of understanding of it and no experience with it.

• Code optimization, and rewriting text.

The second open question was in case the person was willing to utilize AI technology in the future, how did he/she/them envision it?

The following answers were given:

·         AI can pose a new way of interacting with machines and can boost other types of creativity. In that sense, it can be a driver for innovation. That doesn’t mean it is ‘the future’ however.

·         it is supposedly very good at writing code.

·         1. For writing: to improve my writing by getting suggestions about alternative phrasings, synonyms or antonyms, and pointing out common grammatical errors (Just like Grammarly); 2. For Summarization and archiving; 3. For finding the most related and cited research topics and studies by a website like “consensus.app”

·         To get faster answers.

·         It is an excellent way to explore preconceived ideas. Also, it can be used as a proofreader, meaning that a previously written text could be inserted into ChatGPT to see if the ideas in the text are clear.

·         Use it to gather ideas about possible technologies or approaches for several tasks. The information obtained must be checked and developed by myself though.

·         I don’t like it, you have too little control over the quality of information you’re receiving and I think it kills creativity as well

·         It’s a good tool to help in parallel with your work it can reduce time-consuming computational research as concept investigation. But it is always important to have a critical reading on the AI outcome and be able to compare them with what you are looking for

·         For quick overviews of unfamiliar research fields

·         Integrating into the research workflow

·         Automation of tasks. Explaining concepts

·         Not as a resource for references, but I use it as a way to assist in writing, for example: ‘can you shorten the following sentence …’

·         Programming: Letting ChatGPT write python scripts

·         It is not very clear yet but helping to find gaps in my current studies, or easily learn other skills for my professional career like programming with python.

·         Assistant role/ as a search engine.

From the second question, the survey results showed that participants had a variety of potential uses for AI technology in their future work. Some saw AI as a driver for innovation, while others believed it was useful for writing code and improving writing skills. Participants also envisioned using AI for summarization, finding related research topics, and getting faster answers. Some participants saw AI as a proofreader, while others envisioned it as a tool for gathering ideas or quickly getting overviews of unfamiliar research fields. Additionally, participants mentioned using AI to automate tasks, assist with programming, find gaps in their studies, and serve as a search engine or assistant. However, some participants expressed concerns about the quality of information obtained from AI and its potential to stifle creativity. Overall, the survey results showed that participants saw AI as a potential tool to augment their work, but not as a replacement for their own critical thinking and creativity.

The third opened question was if the person had concerns about the potential displacement of jobs or the ethical implications of using AI technology in academic work.

For this question more elaborative answers were given:

·         The intellectual work done by a machine might indicate that the person behind it has not made any effort to create the text.

·         The problem with AI interfaces is the dispossession of knowledge about our behaviors. AI can be used through an open source. But the processes by which AI decides often remain unclear. And because AI depends on large quantities of (expensive) human-labeled data, the technology mostly leads to an extractive political-economic framework intended to steer human behaviors in the future for commercial gains. This is thus a threat to freedom and self-development.

·         Work can be less original because it uses what it knows from the database. Overall, didn’t really think about it.

·         I think my main concern stems from a broad misunderstanding of models like ChatGPT with a lot of people that are not very familiar with the topic. I do not think AI language models in their current state have the ability to really replace a lot of jobs. There is too much oversight needed in editing the results of such models. If you write anything that needs to be based on facts and represent facts you cannot rely on ChatGPT. However, it might increase the efficiency with which certain simple/repetitive tasks can be performed. Then again efficiency and productivity have always been increasing since humans started building stone tools thousands of years ago and we still all have to work… so I would say nothing will change, as usual.

·         privacy concerns, and using AI to deceive or manipulate

·         I believe there will be both job displacements and ethical problems will happen with emergence and application of this technology. Also, we will have improved work quality at the cost of less originality. In general, it is a big change in our world that must be embraced but used under certain defined regulatory rules.

·         Most of them are related to dissemination.

– Papers are written partially by natural language processing algorithms such as ChatGPT.

– Papers review made by ChatGPT.

– Displacement of jobs concerning to news presenters (China case), lawyers (USA case), accountants, and much more.

  If checks and regulations do not happen false and/or simplified information could be a massive problem.

·         Bias – for example, AI CV-readers have been found to train themselves to be biased against women and people of color

·         Especially ethically, if AI-written texts are not labeled as such

·         I don’t think it will replace jobs in academia but I think the technology is developing so fast and that it could potentially be disruptive to many areas in life

·         How to control the intelligence of a method/tool for which human beings barely can grasp their understanding of their own intelligence?

·         Decrease the quality of academic work or decrease the effort of the students to learn and find Information. Do not value the work of specialized technicians in certain areas and remove some positions anymore necessary.

·         Avoidance of learning and bad citation. I’m concerned about what it will do to the concept of truth and reality. I’m concerned that it will be used as a means to do more, which leads to employers demanding ever more from everyone, instead of enjoying the benefits for having more leisure. I’m concerned about data and data harvesting; I’m concerned about a view of the world as a set of data rather than a living web that is not data.

·         If people and especially scientists become too dependent on AI, they would no longer be able to approach research problems in a scientific way and thus, they would not be able to conduct proper research, which takes lots of critical thinking. AI can assist scientists (even more in the near future) and help them be more efficient; however, a scientist should have the basic knowledge and be able to do his job even without any assistance, if needed. There can be ethical implications, too, and I also have the feeling that there is lack of regulations regarding AI applications.

Regarding the 3^rd open question. Some individuals expressed concerns about the intellectual work done by a machine, indicating that the person behind it has not made any effort to create the text. They suggested that AI interfaces can lead to the dispossession of knowledge about our behaviors, which can be used through an open source, but the processes by which AI decides often remain unclear. Others shared their thoughts on how the use of AI technology mostly leads to an extractive political-economic framework intended to steer human behaviors in the future for commercial gains, which is a threat to freedom and self-development.

There were also concerns about the potential job displacements and ethical problems that could arise with the emergence and application of this technology. Some suggested that it could lead to less original work because it uses what it knows from the database, while others expressed concerns about privacy, deception, manipulation, and false information.

There were also concerns about the potential bias in AI algorithms, particularly in the case of AI CV-readers, which have been found to train themselves to be biased against women and people of color. Some were concerned about the ethical implications of AI-written texts not being labeled as such, while others expressed their worries about the potential disruption of many areas of life due to the rapid development of this technology. There were also concerns about the potential decrease in the quality of academic work, the decrease in the effort of students to learn and find information, and the devaluing of the work of specialized technicians in certain areas, which may no longer be necessary. Some individuals were concerned about the impact of AI on the concept of truth and reality, the demands it may place on employers, and the view of the world as a set of data rather than a living web that is not data.

Lastly, there were concerns about the potential dependence on AI and the implications it could have on scientific research. While AI can assist scientists and help them be more efficient, individuals expressed concerns that if people and especially scientists become too dependent on AI, they may no longer be able to approach research problems in a scientific way and may not be able to conduct proper research, which takes lots of critical thinking. They suggested that scientists should have basic knowledge and be able to do their job even without any assistance if needed. Some individuals also felt that there was a lack of regulations regarding AI applications.