8 Case Study: GenAI vs Library databases
The scenario
Elena and Hans are working together on a group research project where they need to each find academic sources for an annotated bibliography. As a part of the project, their instructor wants them to track how much time it takes to craft a search, find sources, and reflect on the efficiency and effectiveness of their searches.
They decide to take different approaches in different tools.
Elena crafts her search strategy for a library database (“cat videos” OR “cute animal videos”) AND “mental health” to get results in Google Scholar. Her results are a list of academic articles with clear citations and links to the full text. She browses through the sources, reads the abstracts and decides what sources to include for the annotated bibliography. The information is reliable and verifiable, but the process is slower than she’d like and requires a certain skill set crafting the search strategy.
Hans prompts a GenAI with the following:
“What does the academic research say about the impact of watching cute cat videos on mental health.”
Academic research overwhelmingly supports the idea that watching cute cat videos (and other cute animal videos) has a positive impact on mental health. Here’s a breakdown of the key findings:
- Stress Reduction and Mood Enhancement: Studies, such as one from the University of Leeds, have shown that watching cute animal videos can significantly reduce physiological signs of stress, including lower heart rates, decreased blood pressure, and a reduction in cortisol (the primary stress hormone) levels.
- Cognitive Benefits and Productivity: Improved Focus and Concentration: Research indicates that looking at cute animal photographs or videos can enhance focusing and concentration.
- Potential Therapeutic Applications: Low-Cost Pet Therapy: Researchers suggest that online cute animal videos could serve as a low-cost, accessible form of “digital pet therapy,” especially for those who cannot have real pets or are allergic.
- Nuances and Considerations: Individual Differences: The emotional impact can vary from person to person, influenced by factors like empathy and perceived connection to the animals.
In conclusion, academic research strongly supports the beneficial effects of watching cute cat videos on mental health, primarily by reducing stress, improving mood, and even enhancing cognitive functions. This “fluffy” pastime appears to be a legitimate and accessible tool for emotional well-being.
Hans gets a quick, summarized answer, but it’s a generated text summarizing the issue, and providing a few links to sources.
When Elena and Hans compare their methods, they find that while Hans’ search may seem more efficient time-wise at this stage in the process, Elena jokes that her search was more energy efficient. Hans was surprised at this, and looked up the differences in web searches and GenAI prompts.
He found that Elena’s database search was relatively energy-efficient. The library database processed her query and retrieved existing information. This process is similar to a standard web search, which has a relatively low energy cost per search.
Hans’ use of GenAI, however, was significantly more energy-intensive. Generating a response from a large language model requires a massive amount of computational power. A single query to a GenAI can consume roughly ten times more energy than a traditional web search. This is because the GenAI isn’t just retrieving information; it’s using a process of prediction to “create” a new, unique response based on its training data.
Capabilities of GenAI
- GenAI quickly summarized this issue for Hans, and gave him a list of sources. It probably saved him some time
- GenAI created a summary of the issues that Hans can use to generate keywords and phrases to use in further research.
Limitations of GenAI
- Han’s results from GenAI may sound confident but could be incorrect or fabricated—or it may be based on outdated or biased information from its training data. He has to invest additional time to cross-reference and verify the sources he found.
- The sources listed aren’t all academic, despite Hans’ prompt. This is a big limitation depending on the research topic and the available data.
- GenAI prompts use a lot of energy. It is estimated that a GenAI prompt takes 10 times more energy than a web search, like we use for a library database. If Hans is concerned about the environment, he should consider how he can more effectively craft his prompts to lessen his environmental impact.
Next steps
GenAI can be a great tool for summarizing a topic, and generating ideas. Library databases like Google Scholar are currently better at effectively, efficiently and reliably finding scholarly sources. Hans and Elena should compare their results and pull together the most relevant sources on their topic.
