China
Africa
Explore chapters and articles related to this topic
Case Studies
Published in Abhijit Pandit, Mathematical Modeling using Fuzzy Logic, 2021
The related terms “data dredging,” “data fishing,” and “data snooping” refer to the use of data mining methods to sample parts of a larger population dataset that are (or may be) too small for reliable statistical inferences to be made well-nigh the validity of any patterns discovered. These methods can, however, be used in creating new hypotheses to test versus the larger data populations.
Artificial intelligence and the role of researchers: Can it replace us?
Published in Drying Technology, 2020
Regarding the design of empirical studies (e.g., experiments), data collection and analysis, some researchers advocate that AI (e.g., using predictive analytics) can help in following an inductive approach that is free of human bias. However, as a critic of empiricism based on AI processing of big data, I would argue that the data itself is oligoptic, i.e., shaped by conditions, such as the technology used to collect and process it, the ontology imposed by humans and the regulatory environment. Indeed, the inductive strategy to research does not occur in vacuum – it is framed by researchers’ previous findings, theories, experience and training. When it comes to results interpretation, it is clear that patterns found within a dataset through AI techniques are not inherently meaningful. Correlations among variables could be random in nature and have no or little causal association. Data dredging may not necessarily produce meaningful results. For identifying the theoretical and practical contributions of a research study, this again requires comparison with the literature and interpretation of the significance of the findings. This is challenging for AI to perform, since it has not solved natural language understanding problems in a general sense. For the last step of writing quality publications, there are a few programs that use AI to generate the draft of a science paper using researchers’ data. However, such software usually generates a first draft that the scientist must revise, add the discussion and other parts. And authors need to provide the project, experiment and task descriptions for the AI tool to create the draft.