Comparing Claude’s Deep Learning With Gemini and ChatGPT
Claude , Anthropic’s AI-powered bot, may not make as many headlines as ChatGPT, Gemini , or Copilot, but it’s probably the one I turn to most often these days: it’s smart and thorough, and its responses often feel less robotic and sycophantic than the results you get from many of its competitors.
One of the latest features to be rolled out to Claude, and one you’ll already find inside bots from Google and OpenAI, is deep research, which Claude simply calls Research. It’s like a supercharged web search, where you’ll get a detailed, long-form report rather than a few paragraphs. It’s great for research projects, as the name suggests, or any scenario where you need to dive deep into a topic.
Now, these AI models rely on any information that is available online. So their reports won’t be as comprehensive if there is very little material available online on the topic you are interested in. However, if there are resources to pull information from, these deep research tools usually do a good job of gathering it, and they always come with citations and web links so you can check their work .
At the time of writing, you need to be a paid Claude user (starting at $20 per month) to access the research tool. Just select the Research option in the prompt window before entering what you want the AI to research. To test the capabilities, I compared it to similar features available in ChatGPT and Gemini to see how Claude handled using topics I know something about (so I can more easily check for errors): the Samsung Galaxy S25 Ultra and the works of David Lynch.
Samsung Galaxy S25 Ultra Review
Here’s the prompt I used for the bots: “I need to know as much as I can about the Samsung Galaxy S25 Ultra. What are its key specs and features? What’s changed from the Galaxy S24 Ultra? How is it different from other Galaxy S25 models? What kind of user is it for? What does it tell us about the state of the mobile industry in 2025?”
Claude jumped right into the research and had a report in seven minutes. He can show you his thoughts as he goes along if you want to see them. The final document he submitted was detailed and to the point, with embedded web links to back up everything he wrote. He touched on key points like the AI software and the S Pen downgrade, but the writing tended to be a bit generic and lacking in substance in places. It was the shortest of the reports, at 1,200 words.
Gemini got their report out the door the fastest, just ahead of Claude, and wrote the most, at over 5,500 words. I first got the research outline for review and approval, then got detailed updates on where Gemini was by scouring the web. I found the finished research document to be quite detailed, even including processor timings and the date and location of the phone’s launch. It was also accurate and well referenced, as far as I could tell, though it veered into Samsung marketing language quite regularly.
ChatGPT started by asking a few follow-up questions, and then took twice as long as Claude and Gemini to spit out just over 5,100 words. This bot will show you the most details in terms of activity while it’s running, though they’re hidden by default. In the end, I got another well-researched, well-formatted, and accurate report with a good selection of details, though Gemini offered a bit more context. Like Gemini, ChatGPT tended to just lift sales pitches from Samsung’s own press releases, which isn’t necessarily what you want.
Research by David Lynch
Moving on to the world of David Lynch, this time my brief was: “I need a detailed research report on David Lynch’s film and television work (in this case, I don’t need to know about his music or art). What are his most influential works? What are his most well-known and accessible works? What defines David Lynch’s style? What impact has he had on the industry?”
Claude was again the quickest and most concise in his response, writing just under 1,200 words in just over five minutes. He correctly picked out what is considered Lynch’s best work and explained why, and the report did a good job of identifying the films and shows that were influenced by Lynch. While not the most in-depth in terms of research, Claude provided a great overview.
Gemini took a couple of minutes longer than Claude to write nearly 7,000 words on David Lynch, and again came up with a research outline first. As with the Samsung report, Gemini added far more detail than Claude: a resume, a timeline of his career (complete with chart), and a comprehensive look at what “Lynchian” actually means—which was pretty accurate, complete with quotes from the man himself. It would certainly be a comprehensive starting point if you were researching Lynch.
ChatGPT took a few minutes longer than Gemini to submit its copy, which stretched to nearly 8,000 words. Again, there were follow-up questions at the beginning, which seemed obligatory. There was a detailed chronology, and the report did a good job of selecting Lynch’s works that were either celebrated or open to discussion, as well as picking out some deeper cuts and accurately summarizing what makes Lynch special—though it seemed to rely too heavily on a few sources.
The Future of Research
It’s hard to pick a winner because there are so many variables, and when it comes to something like report length, it really depends on what you’re looking for. All of the chatbots did a good job of doing their research, but if I had to pick, I’d go with Gemini’s reports: they just seemed a little more detailed and thoughtful than the others.
There is no doubt that there is value in what AI search can do, producing detailed and (apparently) accurate reports in minutes that would take real human hours. When bots are in this mode and forced to check their work, there should theoretically be less chance of them fabricating facts. As far as I can tell, they have chosen reputable and authoritative websites as sources.
As this way of working becomes the norm, two questions arise: first, what might happen to our ability to research and compare information using our own brains (see also our ability to summarize and write well)? Second, what is the end result of all of us spending time inside AI bots rather than on the open internet?
If AI takes down most of the sites it scrapes, those tools will be far less useful. For example, bots will try to summarize the works of David Lynch from Reddit and Wikipedia, rather than the work of professional critics and interviews with industry insiders. That’s one of the best uses of bots like Claude, Gemini, and ChatGPT, but only because they can scrape from the vast library of content published online, which they also compromise.
Disclosure: Lifehacker’s parent company, Ziff Davis, filed a lawsuit against OpenAI in April, alleging that it infringed Ziff Davis’ copyrights in the training and operation of its AI systems.