I can get answers with ChatGPT but Deep Research gives a dissertation I’ll never need.

While I love to learn about new things, I also enjoy getting lost in research rabbit holes. But sometimes I need a concise answer to a particular question or a guide to a specific task. If I’m trying figure out whether Pluto was reinstated as a world or how long to roast a chicken, I just want a list of bullets and a simple yes/no.

While ChatGPT’s Deep Research has been a great researcher and is perfect when I want to immerse my self in a particular topic, I haven’t made it the default tool for the AI chatbot. The AI model’s search tool and database can answer any question or issue that I may have. I don’t want a formal report that takes 10 minutes to prepare on how to cook a meal. I find the answers from Deep Research to be very comprehensive, so I decided to compare it with the standard (GPT-4o), ChatGPT model, and give it a few prompts I could imagine submitting at random or without a long-term requirement.

Beef Wellington

I wanted to test both models on a classic and intimidating recipe, Beef Wellington. This is not a dish that you can throw together in a hurry. This is a multi-step, time-consuming process that requires patience. Deep Research was useful for this meal. I asked both models : “Can you give a simple recipe of kosher Beef Wellington?” (

Regular ChatGPT replied almost instantly with a well-structured, straightforward recipe. It was easy to follow, had clear measurements and even offered some helpful tips for avoiding common pitfalls. It was just what I wanted in a recipe. Deep Research took 10 minutes to complete and included a mini-cookbook that was very detailed. I received multiple versions of Beef Wellington that adhered to my specific requests but varied from Jamie Geller’s method to a 19th century traditional preparation with some substitutes. Not to mention the extra suggestions for decorations and an analysis on different types of puff pastry, their butter-to flour ratios and other details. It was a great piece of trivia. If I wanted to make the dish myself, it was too similar to those recipe blogs that require you to scroll through someone’s life story to get to the ingredient list.

TV hour

I wanted to know if Deep Research would help me choose a TV, so I kept the question simple: What should I consider before buying a new television?

ChatGPT provided me with a quick answer. It broke down screen size, display type and smart features into four categories: ports, smart features and resolution. It told me 4K was standard, 8K was overkill, OLED had better contrast, HDMI 2.0 is great for gaming and budget is important. I felt confident that I knew what I was looking for and could have easily walked in a store armed with this information.

Deep Research asked me the usual questions about my priorities, but this time it took only six minutes to deliver a full report. Instead of a simple pros-and-cons list, I received a lot of unnecessary details on things like TV refresh rates and video games and the impact of compression algorithm on streaming quality. This was all very informative, but not necessary for my needs. I won’t be returning to the TV guide as often as I would to Beef Wellington.

Telescope look

I decided to make the final test a bit more academic, given my recent decision to take astronomy as a serious hobby. I asked How does a telescope function?

Regular ChatGPT immediately responded with a simple and digestible answer. Telescopes use either lenses (refracting scopes) or reflective telescopes to gather and magnify the light. It was easy to understand, as it only briefly touched on magnification and resolution, without getting too technical.

Deep Research provided me with a report similar to one I would have written in high-school. After I told them I didn’t need a technical answer, they spent about eight minutes explaining how radio telescopes work, how different types of telescopes were developed, and how optics works. The report included a guide to buying your first telescope, and a discussion of atmospheric distortion when making ground-based observations. It answered questions I had not asked. In this case, the anticipation of further questions was not a big negative. Even so, a few sentences about mirrors could have been enough for the moment.

Deep thoughts

I still think of Deep Research as an impressive AI tool, but now that I use ChatGPT regularly, I am much more aware of the excesses of this tool. The reports are well-organized and written in a surprisingly professional manner. It’s great for a random tour through interesting information, but I need answers more often than a thesis. A shallow dive is sometimes better than a deep one.

I’ll choose the regular ChatGPT method 99 times out 100 if it is accurate and provides in seconds what Deep Research takes several minutes and unnecessary context to provide. Sometimes, less can be more. Deep Research’s advice on shopping would be perfect for a bigger purchase, such as a car or house. Deep Research does too much for everyday purchases. I don’t want a jet motor for an electric scooter but it would be useful for a transcontinental trip.

You might also enjoyI tried Perplexity’s Deep Research, but it didn’t live up to ChatGPT’s research potential.

Perplexity’s Deep Research doesn’t quite match ChatGPT’s research capabilities.

ChatGPT’s Tasks feature replaced my to-do-list and completely changed the way I planned my life.

Eric Hal Schwartz has been a freelance writer at TechRadar for more than 15 years. He has covered the intersection of technology and the world. He was the head writer of Voicebot.ai for five years and was at the forefront of reporting on large language models and generative AI. Since then, he has become an expert in the products of generative AI, including OpenAI’s ChatGPT and Anthropic’s Claude. He also knows Google Gemini and all other synthetic media tools. His experience spans print, digital and broadcast media as well as live events. He’s now continuing to tell stories that people want to hear and need to know about the rapidly changing AI space and the impact it has on their lives. Eric is based out of New York City.