How to Use Perplexity's FREE Deep Research to Save *Hours* on Research

Key Concepts

Perplexity AI, Deep Research, Academic Sources, Literature Review, Nanomaterials, Transparent Electrodes, Figure of Merit, Stability, Peer-Reviewed Paper, References, Export Formats (PDF, Markdown, Perplexity Page), Free Access, R1 Reasoning, O03 Mini, Pro Search, Web Search, Academic Search, Social Search.

Perplexity AI: Deep Research Capabilities

The video focuses on the deep research capabilities of Perplexity AI, particularly its "Deep Research" feature, and compares it to tools like ChatGPT. The speaker highlights Perplexity's ability to specifically target academic sources, making it a valuable tool for students and researchers.

Access and Cost

• Perplexity offers a "Pro Search" feature with deep research capabilities, including reasoning with R1 and deep seeks R1 reasoning with o03 mini.
• Users get four enhanced queries remaining today, which means five a day.
• A significant advantage is that it offers five free "Pro Searches" per day, unlike ChatGPT, which requires a paid subscription (around $200/month).

Benchmark: Humanity's Last Exam

• Perplexity uses "Humanity's Last Exam" as a benchmark for its deep research tools. This exam includes challenging questions from various fields like eology, chemistry, computer science, and mathematics.
• Perplexity claims its performance on this benchmark is superior to other models, including open AI.

Deep Research Process

1. Accessing Deep Research: The speaker demonstrates how to access the "Deep Research" feature within the Perplexity AI interface.
2. Setting Search Sources: A key feature is the ability to set sources for the search, allowing users to choose between "Web," "Academic," and "Social" sources. The speaker recommends turning off "Web" and focusing on "Academic" sources for research purposes.
3. Prompting: The user provides a specific prompt, such as "I'm writing a peer-reviewed paper about opv devices and nanomaterials create a short 300-word intro for that using references."
4. Research Plan: Perplexity AI creates a research plan, outlining the steps it will take to answer the query. The speaker notes that this process can take a few minutes.
5. Source Retrieval: The tool retrieves a large number of sources (e.g., 58 academic sources in one example).
6. Answer Generation: Perplexity AI generates an answer based on the retrieved sources, structuring it in a logical way (e.g., sections on emerging nanomaterials, novel fabrication methods, figure of Merit analysis, and stability).
7. Source Review: Users can review the sources used and remove any that are not relevant.

Example: Literature Review on Nanomaterials for Transparent Electrodes

• The speaker provides an example of a literature review on nanomaterials for transparent electrodes, focusing on new materials, new fabrication methods, figure of Merit, and stability.
• Perplexity AI generated a comprehensive response with 58 academic sources.
• The response was well-structured, covering the key areas specified in the prompt.
• The speaker praises the quality of the response, giving it an "Andy ston bloody Excellence stamp of approval."

Exporting Results

• Perplexity AI allows users to export the results in various formats:
  • PDF: The PDF export includes clickable links to the original references.
  • Markdown: For users who prefer Markdown format.
  • Perplexity Page: A web page format that provides a clean and easy-to-read interface.
• The "Perplexity Page" format organizes references in order of where they're actually referenced.

Comparison with ChatGPT

• Information Depth: ChatGPT may provide more in-depth information than Perplexity AI.
• Usability: Perplexity AI's export options and focus on academic sources make it more usable for research purposes. ChatGPT requires more manual work to extract and format references.
• Source Quality: Perplexity AI prioritizes academic sources, while ChatGPT may use non-academic sources.
• Control: Perplexity AI offers more control over the search sources (Web, Academic, Social).

Limitations

• Perplexity AI may struggle with short, referenced outputs. In one example, it failed to provide references for a 300-word introduction to a peer-reviewed paper. It seems to work best with either long, referenced outputs or short, unreferenced outputs.

Notable Quotes

• "Andy ston bloody Excellence stamp of approval" - referring to the quality of the literature review generated by Perplexity AI.
• "completely for free Okay" - emphasizing the significant advantage of Perplexity AI's free access.

Conclusion

Perplexity AI's "Deep Research" feature is a powerful tool for academic research, offering free access to a large number of academic sources and the ability to generate well-structured literature reviews. While ChatGPT may provide more in-depth information, Perplexity AI's focus on academic sources, export options, and free access make it a compelling alternative, especially for students. The ability to specify search sources (Academic, Web, Social) provides a level of control not found in other AI research tools. However, it may struggle with generating short, referenced outputs, which is an area for improvement. Overall, the speaker highly recommends Perplexity AI for students and researchers.