We’ve seen how generative AI and the Large Language Models (LLMs) attached to them can build websites, create art, and write text, but aside from the pros and cons of using these tools to churn out new content, there’s another use for them that doesn’t get quite as much attention: analyzing and summarizing large blocks of text.
It’s a job that’s tedious and time-consuming for human beings, and which AI tends to be rather good at (though we’d always caution against relying entirely on artificial intelligence when accuracy and clarity is important).
Of course there are times to do this and times not to do this: You won’t get the richness and emotion of Bleak House out of an AI summary of it. But if you need an almost-instant précis of a long and detailed document, AI can help.
Adobe is the latest company to roll out this kind of AI capability, adding a beta AI Assistant to its Reader and Acrobat products. The idea is you can literally chat to your PDFs to get insights and information from them—though you need to be signed up for one of the paid plans (from $22.99 a month).
If you want to test it out, there is a free trial and a 14-day money back offer available. Here are the features to know about to help you decide if it’s worth it for you.
Chatting to PDFs in Acrobat
Load up a PDF in Adobe Acrobat (or Reader) on the desktop or the web, with a paid subscription active, and you should see an AI Assistant button up in the top right corner. Right now, the tool can work with PDFs up to 120 pages in length, and we tested it with the first chapter of the aforementioned Charles Dickens novel, Bleak House.
Click AI Assistant to open the side panel, and you’ll get some suggested questions you might want to ask—you can either click on one of these suggestions, or type in your own query. We asked for summaries of what was happening in the chapter, and asked about the key characters mentioned, the setting, and the weather. On the whole, the answers were well-judged and correct.
The AI Assistant can produce accurate summaries.
Credit: Lifehacker
The overall summary was really good, too: “The document is about the bleak and chaotic atmosphere of London during November, particularly in the Court of Chancery, and the complicated and never-ending legal case of Jarndyce and Jarndyce.” Top marks for Adobe and its AI.
However, as the program itself says, “AI-generated responses may be inaccurate or misleading”—so it’s not something to rely on entirely for your most important work. While the AI identified the characters in the chapter pretty well, “Michaelmas term” was included too, which is obviously wrong.
You’re able to ask any questions you like about the PDF.
Credit: Lifehacker
One really helpful touch is that a lot (though not all) of the answers come with links to the source—so you can see where in the document the AI assistant got its information from. There are quick links for copying individual answers or the entire conversation to the clipboard, so you’re easily able to get the AI’s responses out somewhere else. Even in beta form, the AI Assistant seems like a really useful tool for anyone who spends a lot of time dealing with PDFs.
You can ask follow-up questions, and ask questions about topics that aren’t covered in the document—we wanted to know who Charles Dickens was, and the AI answered well enough. However, when it’s going off tangent and outside of the document, the assistant doesn’t give you source links, so you need to be even more wary about inaccuracies.
Chatting to PDFs in other tools
Adobe and its products are always going to be heavily associated with PDFs and PDF management, but there are other options out there too. We’ve previously written about the aptly named ChatPDF, which is simple to use and free as well (paid plans are available for larger PDFs): It’s based on the ChatGPT engine developed by OpenAI, and you just need to drop in a PDF to start asking questions about it.
ChatGPT itself can handle PDFs as well, provided you’re paying for the $20-per-month Plus package. On the web interface, click the paperclip icon to the left of the text input box, then choose your PDF: You’re then able to ask as many questions as you want about its contents, getting answers or summaries, or whatever you need. You can tell ChatGPT to stick to what’s in the document, or bring in other knowledge from elsewhere too.
ChatGPT Plus can query PDFs for you too.
Credit: Lifehacker
Then there’s Perplexity, which will scan, summarize, and answer questions about PDFs for you for free (the paid Pro plan gives you access to a smarter model and more features, however). Just point the AI to a PDF of your choice, and it’ll come up with a list of suggested questions you might want to ask—or you can provide your own.
Another option is to open your PDF in Microsoft Edge: If you then click the Copilot button (top right), you’ll be able to use the AI to query the document in whatever way you like, whether you need overall summaries or have to pick out specific information. The only major AI missing this functionality right now is Google Gemini—though it probably won’t be long until it gets added.
