This AI chatbot can sum up any PDF and answer any question you have about it – ZDNet
Most Popular
Whether you are a student or a working professional, you likely interact with PDFs frequently through the form of presentations, academic research, business reports, and more.
Regardless of whether it is a 90-page slide deck or a lengthy research paper, PDFs in the classroom or workplace are often tedious to wade through. ChatPDF is here to help.
Also: How to use ChatPDF
As the name implies, ChatPDF allows you to chat with your PDF.
ChatPDF runs on OpenAI’s GPT 3.5 large language model and can answer any question you have about the PDF you upload. The chatbot can even give you a full summary of the PDF without you having to read it.
Also: The 5 biggest risks of generative AI, according to an expert
Free plan users are limited to three PDF uploads of 120 pages or less a day. However, if you need more access, you can upgrade to a plus plan for $5 per month.
I put the chatbot to the test and was thoroughly impressed with the results.
I recently wrote an article on AI based on a 58-page paper by the National Bureau of Economic Research. I chose to use this paper as my test subject for ChatPDF as I knew its contents well.
After uploading the paper to ChatPDF, it automatically generated a very brief summary for the paper and suggested questions I could ask.
ChatPDF’s response: Welcome to the informative PDF file on Generative Al at Work! This paper, authored by Erik Brynjolfsson, Danielle Li, and Lindsey R. Raymond, explores the impact of generative Al on the workforce.
Here are three example questions you may have about this paper:
> What is generative Al and how does it differ from other forms of Al?
> What are the potential benefits and drawbacks of implementing generative
Al in the workplace?
> How can companies prepare their employees for the integration of generative Al technology?
Instead of using one of the suggestions, I asked the chatbot what study was conducted in the paper. I was met with accurate results and even a page number that told me where the chatbot got the information from in the paper, which was useful in case I wanted to refer back to it myself.
Also: How to use ChatGPT to write an essay
I then asked what the results of the study were and was met, yet again, with accurate results and a page reference.
My prompt: What study is conducted in the paper?
ChatPDF: The paper studies the staggered introduction of a generative Al-based conversational assistant using data from 5,179 customer support agents. The authors analyze the impact of access to the tool on productivity, as measured by issues resolved per hour, and examine how the Al model disseminates potentially tacit knowledge of more able workers and helps newer workers move down the experience curve. This study provides evidence that generative Al can increase productivity in the workplace. (Page 2)
My prompt: What are the results of the study?
ChatPDF: The study found that access to the generative Al-based conversational assistant increased productivity, as measured by issues resolved per hour, by 14 percent on average. The greatest impact was observed on novice and low-skilled workers, while experienced and highly skilled workers experienced minimal impact. The authors also provide suggestive evidence that the Al model disseminates the potentially tacit knowledge of more able workers and helps newer workers move down the experience curve. (Page 2-3)
Finally, I wanted to see if the chatbot could sum up the entire article accurately and concisely. When prompted, the chatbot did exactly that — it summarized the study and the findings in one, five-sentence paragraph.
My prompt: Can you sum up the entire paper for me?
ChatPDF: Certainly! The paper titled “Generative Al at Work” explores the impact of generative Al on the workforce. The authors study the staggered introduction of a generative Al-based conversational assistant using data from 5,179 customer support agents. They find that access to the tool increases productivity, as measured by issues resolved per hour, by 14 percent on average, with the greatest impact on novice and low-skilled workers. The authors also provide suggestive evidence that the Al model disseminates the potentially tacit knowledge of more able workers and helps newer workers move down the experience curve. Overall, this study provides evidence that generative Al can increase productivity in the workplace. (Page 2-3)
After the positive results I got from using ChatPDF to summarize an academic paper, I decided to try a presentation. I went all the way back in my Google Drive archives to find a final presentation I worked on in college, which made the perfect candidate for the test.
After uploading the PDF, the chatbot output a brief, one-sentence summary of my own presentation as well as suggested questions. When I asked for a more thorough summary, the chatbot delivered again, as seen below. It even included an anecdote about two videos in the presentation that I forgot were in there.
ChatPDF: Welcome to this informative PDF file about the rise of misogyny in the digital age! In this file, you will learn about the impact of media on female celebrities and their bodies. As you read through the content, you will gain a deeper understanding of the ethical issues surrounding the topic and how we can change the narrative. Enjoy!
Example questions:
>What are some examples of female celebrities who have been subiected to media scrutiny about their bodies?
> How does the media’s portrayal of women’s bodies contribute to the rise of misogyny in the digital age?
> What steps can we take to shift the focus away from women’s bodies and towards their accomplishments and talents?
My prompt: Can you sum up this PDF?
ChatPDF: This PDF discusses the rise of misogyny in the digital age and its impact on female celebrities. It explores how the media’s portrayal of women’s bodies contributes to this issue and offers suggestions for changing the narrative. The PDF includes a brief description of two videos that discuss weight bias and fat shaming.
To truly test its ability, I asked it a more specific question about one slide. The results, once again, were accurate.
My prompt: What media ethics lessons are mentioned?
ChatPDF: The PDF mentions six media ethics lessons on page 8. These include keeping the audience’s reaction in mind, checking boundaries, determining if the content is story-worthy, consulting your conscience, checking wording, and being sensitive when handling sensitive topics.
To use ChatGPT to sum up articles, you are supposed to type in “TDLR”, and then follow that with the link of the article.
Also: How to use ChatGPT to summarize a book, article, or research paper
I used that method and the results were incorrect. ChatGPT gave me a summary for an article that wasn’t even related to the topic at hand.
ChatGPT: The paper titled “The Effects of School Spending on Educational and Economic Outcomes: Evidence from School Finance Reforms” examines the impact of school finance reforms on educational and economic outcomes in the United States. The authors use a quasi-experimental design to study the effects of court-ordered changes in school spending on student outcomes, such as test scores and high school graduation rates, as well as later-life outcomes, such as earnings and employment. They find that increased school spending has a positive effect on educational outcomes, with particularly large impacts for low-income students. However, they find little evidence of a causal link between school spending and later-life outcomes
The incorrect results make sense, since ChatGPT doesn’t have access to the internet or to information after 2021. However, this fault makes ChatGPT wildly unreliable for our use case.
If you need to summarize a PDF, I would recommend utilizing ChatPDF instead of ChatGPT as the results are accurate, quick, and free.
