Perplexity: A Hands-On Experience to Streamline Assignments
This transcript is from my Perplexity sharing session—the first one I’ve held in Canada and my first conducted in English.
Published: 2025-01-20
0. Thanks
First, I want to thanks for my Technology Innovation Management program for supporting this sharing session and making it possible.
I also want to thank everyone who attended, especially in these -20°C conditions, and for making the trek all the way to St. Patrick’s Building. I hope spending this hour here proves worthwhile for you.
1. Opening
1.1 Speaker Introduction
- A second-term Master of Engineering student in Technology Innovation Management (TIM) at Carleton University.
- Perplexity Campus Strategist (Volunteer), who facilitated a one-year free Perplexity Pro subscription for the Carleton.
- A dedicated ChatGPT user for over 25 months and a top creator of GPTs.
1.2 Why Sharing
- Perplexity Space made it easy for me to complete my final assignment.
- Even though I shared the guide with a clear explanation, people still didn’t know how to use it.
- Sharing practical, real-life use cases is more effective than presenting flashy ones; it can inspire others more effectively.
- Sharing is learning.
Let’s maximize the benefits of our Pro subscription!
1.3 Disclaimer
Please use these tools responsibly and in compliance with academic integrity guidelines.
This is my personal sharing and does not represent the views of any institution or organization. I am still exploring how to use these tools, and today’s session is merely a way to spark ideas. I hope we can use them effectively and gain practical insights into AI while keeping the experience relaxed.
2. Perplexity Introduction
2.1 What is Perplexity?
Perplexity is a search engine that delivers accurate, real-time answers with sourced links, making it a valuable tool for academic, professional, and personal use.
Perplexity = ChatGPT/Claude/Grok + real-time open web information + citations.
2.2 How can Perplexity help you?
Academic Research
- “Find articles on the history of quantum computing.”
- “Locate studies on climate change’s impact on polar bears.”
Professional Research
Personal Use
- “Explain tuition fees instantly.”
- “Troubleshoot technical issues (e.g., ‘red microphone error on Mac’).”
- “Find upcoming tech events.”
2.3 What is Perplexity Pro?
- Get access the most models like o1, Claude Sonnet 3.5, GPT-4o
- Get more Pro searches, file uploads, and creative tools
(Free access to the most advance o1 mode with higher quotas than ChatGPT Plus.)
*o1 is currently only accessible on the web as of January 20, 2025.
2.4 How to get Perplexity Pro?
Carleton has already unlocked a free 1-year Perplexity Pro subscription for all students. The deadline to claim this offer has been extended to May 1, 2025.
Even without a Carleton email, you can enjoy a free one-month Pro trial as long as you’re in Canada.
2.5 Tips of Perplexity Basic
- Make sure you’re signed in with your Carleton email (avoid auto-login with your personal Gmail).
- Activate the Pro Switch.
- Select Focus mode when needed.
- Pick the Default AI Model
- “Rewrite” or “rewrite with different model” if needed
3. Real-Life Use Cases
3.1 Insurance Query
- Why does Greenshield insurance not cover 80%?
- Why do I not get a student dental discount network? What is the difference between this network and co-pay?
- Is my dental clinic on the list?
(compare with ChatGPT)
(searched on 3 January, 2024)
- I found the SDDN list and share it with my peers as a reminder.
- I realized that if I choose a clinic that joins the SDDN, I will nearly not pay the bill. Perplexity even helped me draft an email to my dentist, and the dentist agreed to join the SDDN network!
3.2 Student & Local Life
U-pass issue
Free museum pass
Sledding place in Ottawa
Rideau Canal Skateway Information
Concert Search
3.3 Quick Information Lookup
History of the O-train
Wipebook inquiry
Short winter days
Promo Code search and decision-making
Quick fact-check
3.4 Technical questions
Word Layout issue
Mac display issues
DIY NFC business cards
WhatsApp sub-group QR Code
3.5 Limitations of Perplexity
Cannot access proprietary data
Perplexity primarily accesses information from the open web. However, it cannot retrieve real-time data from platforms like X (formerly Twitter). For instance, in situations like the following:
In this case, you would need to use Grok:
Similarly, it cannot identify the cheapest place in Ottawa to buy skates at the moment, as it cannot retrieve proprietary data.
Another example is ranking data from app stores in different countries, such as the rapidly growing RedNote app. Neither Perplexity, ChatGPT, nor Gemini can retrieve this data directly:
While Perplexity performs well with open web searches, it cannot handle queries requiring proprietary or inaccessible data. Generally, if open web information can answer the question, Perplexity is likely to succeed. If not, it cannot provide an answer.
Struggle with complex scenarios
Perplexity may also encounter difficulties handling more complex queries, even with the o1 model. For instance:
One possible workaround is to break the query into smaller, more manageable parts:
In some cases, manual searches may be more effective. As the saying goes, “slow is fast” at times.
However, with advancements like o3-mini/o3, Grok 3, and even more powerful models like o4/o5 on the horizon, resolving such challenges may soon become easier.
Small context window
Models like o1, Claude Sonnet 3.5, and GPT-4o available on Perplexity have relatively small context windows. This means they can lose context during long conversations, leading to diminished performance. When this happens, restarting the conversation is often necessary.
This topic will be explored in more detail in the next section.
3.6 When to use Perplexity?
Use it anytime you need real-time, cited info from the internet. If open web data can answer the question, Perplexity likely can, too.
4. Perplexity Space
Perplexity Spaces is a tool that helps you organize, search, and create content from your course materials, notes, and web resources. It allows you to:
- Upload text-based PDFs (up to 50) and input URLs.
- Set a custom space prompt.
- Search exclusively within your knowledge base or across the web, with citations provided for all references. (Although NotebookLM provides a similar feature with larger context window and inline citations, based on my personal experience, Perplexity Space’s performance is better than NotebookLM thanks for o1 and Claude 3.5 Sonnet).
This makes it an ideal tool for assignments that require responses grounded in course materials or academic papers. Additionally, users can customize their queries based on preference, switching between the knowledge base and web sources, or combining both.
While Perplexity assures user that uploaded files are not accessed or shared for other purpose, it is important to ensure that the materials you upload, such as papers or course materials, comply with ethical and privacy standards. Always double-check institutional policies to avoid potential issues.
4.1 Perplexity Space for Assignment
Common AI Challenges for Assignments
- Content that’s too general or unrelated to specific course materials.
- Professors require answers tied directly to class content.
- Need for proper citations.
Example with TIMG - 5001 Final Assignment
In the TIMG 5001 class final assignment, students need to write 3000-4000 word innovation portfolio including Innovation Philosophy, key learning, case studies, Personal Projects and so on, based on what was learned from the course and should include at least 10 relevant, recent peer-reviewed references not older than 5 years.
I don’t know the other program’s assignment and AI policy. Fortunately, my TIM program encourages to use AI tools, but we need to disclose and attach links to show how we interacted with AI.
So, I explored and leveraged the capability of Perplexity Space to help me effortlessly finish this assignment in 2 days.
Timeline
Day 1: Initial Draft
- Began by drafting thoughts without focusing on grammar or structure.
- Prioritized creating a complete draft to establish a foundation for further refinement with AI.
Day 2:
- Uploading Materials and Creating a Workspace
- Uploaded the course materials and set up the TIMG 5001 Space.
- Started refining the draft section by section:
- Iterative Refinement Process
- Reviewed and improved responses by:
- Repeatedly fed improved drafts back into the space for continuous enhancement.
- Verification and Accuracy Check
- Examined AI-generated responses and citations to ensure alignment with the assignment’s requirements.
- Verified the accuracy and relevance of citations to avoid errors or hallucinations.
- Final Optimization and Cross-Platform Collaboration
- Transferred the draft to other AI tools, such as ChatGPT with advanced models, for additional polishing.
- Used this hybrid approach to achieve a unique and refined final version.
- Citation Management
- Copied citation information and generated APA-style references using AI tools.
- Transparency and Disclosure
- Disclosed the use of AI tools in the creation process.
- Included links to conversations in Perplexity and ChatGPT for reference and transparency.
Step-by-Step Guide to Creating a Space
- Navigate to the Perplexity and locate the “Spaces” option on the sidebar.
- Click the “Create a space”.
- Provide a title and description for your new space. For the AI model, I recommend “Claude Sonnet 3.5.”
- After that, use the “Files” section to upload up to 50 documents, ensuring they are text-recognizable PDFs. (Update in 14 December 2024: Perplexity Space supports input URL.)
- Begin your conversation in the space. To search only within your uploaded files, adjust the sources by unselecting “Web”. This ensures the system only references uploaded materials unless you manually switch source
Outcome
- Finished this assignment with high quality and met the requirement in 2 days.
- Professor Nusa really enjoyed my Perplexity conversations and how I approached structuring my portfolio.
(P.S. Since NotebookLM is free and uses a different technical approach than Perplexity Space, you can also create a NotebookLM to compare their performance and choose the best option.)
4.2 Perplexity Space for Paper Search
Search from the ABDC list
With more advanced models like o3/Grok3 on the way, I’m optimistic that significant improvements will emerge this year.
Search from specific websites
4.3 Other potential use cases for Perplexity Space
- Job search and interview preparation: Input all your information and potential employers’ information for job search and interviews.
- Literature review: Conduct thorough reviews of academic papers for research or coursework.
- Consultant assistant: I have transformed guidance documents and FAQs into GPTs via my past company’s internal platform, making it easy for colleagues to consult my assistant first with relevant questions, especially freeing up my non-working hours.
- Performance Review Assistant: Help colleagues optimize their performance review drafts to save time on text processing. Achieved #1 on our internal GPTs platform in one weekend 1 year ago! Even I left my company for a half year, it is still the #1 internal usage.
5. Takeaways
- Start with Perplexity for internet-related queries.
- Experiment with Models: Use o1 for complex searches; try Claude 3.5, GPT-4o, or Sonar for specific needs.
- Manage Context Limits: Refresh and break tasks into steps when using Perplexity.
- Domain-Specific Searches: Use Perplexity Space, prepare drafts, and work iteratively.
- Human-in-the-loop. Always verify AI-generated outputs and maintain a critical mindset.
