Mobile Menu

Is artificial intelligence replacing us?

In a recent Editorial, published in Drying Technology, Atreyi Kankanhalli describes her views on if and how artificial intelligence (AI) could replace researchers.

Artificial intelligence

The rapid advancements in AI have shown promise in transforming healthcare. Experts have developed several AI algorithms that can help in the day-to-day activities of healthcare professionals, including in diagnosis. Media outlets have helped fuel predictions that AI will eventually replace people and take over. For example, Kai-Fu Lee, AI expert, predicted that within 15 years AI will automate and potentially eliminate 40% of jobs. He added that AI will surely replace ‘repetitive’ jobs, e.g. jobs in factories and potentially many ‘white-collar’ tasks. Nonetheless, the exact mechanism of how this will happen is unclear.

A few studies have attempted to explore this. This includes studies developing AI automation scores for work activities of all major occupations. A lot of these studies have suggested that highly creative and knowledge-intensive tasks cannot be automated. However, researchers have reported cases where art critics could not distinguish between creative art pieces completed by AI algorithms and human-drawn paintings.

A critical issue is whether AI can and will replace researchers. The issue of how AI will impact researchers is not only relevant for the research community itself, but also the broader community e.g. research institutions and universities. In this Editorial Atreyi shares her views on three questions, summarised below:

To what degree and for which research tasks may, or may not, artificial intelligence substitute for, or at least complement, researchers?

Researchers perform a series of activities including problem formulation, literature review, theoretical modelling, empirical study design, data collection and analysis, discussion of results, theoretical and practical contributions and writing quality publications.

Atreyi highlights that current literature review tools still need the input of a detailed problem description statements and are also unable to perform exhaustive searches. She argues that problem formulation and theoretical modelling are the least likely tasks to be replaced by AI. While data mining tools can generate hypotheses from datasets, it still remains the judgement of researchers to make sense of these suggestions and determine whether a particular direction is right to pursue. Atreyi believes that AI can help speed up the process of finding new relationships among variables but cannot decide which set of hypotheses to focus on.

Some researchers believe that AI can help remove human bias in terms of the design of empirical studies and data collection and analysis. However, Atreyi believes that this again will require humans to compare the literature and determine the significance of findings. For publication writing, there are a few programmes that use AI to generate the draft of a science paper using researchers’ data. However, such software typically generate a first draft that scientists have to revise. Authors need to also provide the project, experiment and task descriptions for the AI tool to create the draft.

Atreyi believes that AI currently cannot replace these research tasks. While AI could assist in for example, data analysis, human interpretation is still essential. As AI lacks semantic understanding it is unlikely to replace these research activities in the near future.

How would research in various fields, and compared to other white-collar professions, be enabled or challenged by AI?

AI can help in business analytics and in some industry prediction tasks, e.g. sales predictions. This however does not translate for high-end tasks, like strategic decision making. AI tools are also being used in knowledge-intensive industries, such as legal and financial, where recommendation systems are used. However, again experts still make the final decision.

Healthcare professionals are using clinical decision support systems, for example, to detect at-risk patients. These machines however are still in their infancy. The lack of transparency and explainability of AI algorithms hinders research progress.

Atreyi is particularly sceptical about AI executing creative and novel research tasks at this time. She believes that there still are vital tasks for researchers to conduct, such as developing new ideas and original thinking. Humans still have to question the ‘why’ of research.

How could researchers prepare themselves for the impacts of AI on our profession?

Atreyi recommends that researchers should prepare themselves for whatever challenges and opportunities AI may bring to their profession. As we are moving towards a more data-driven paradigm in most professions, Atreyi encourages researchers to embarrass and leverage AI in their work.

To summarise, Atreyi encourages a switch from considering AI as a substitute, to seeing AI as a complement to research. She refers to AI as a “teammate” or “collaborator”, in support of human work. This paradigm in AI is known as collaborative intelligence, where both humans an AI join forces to solve problems. We must think about how we can work synergistically with AI, rather than considering it a threat to our existence.

Image credit: Ales_Utovko –

More on these topics

AI / Algorithm / Healthcare

Share this article