Comparison of Chat GPT and Claude; Which artificial intelligence wins? • DigiKala Meg

Open AI first introduced ChatGpt to the world more than two years ago, and it was since the competition has increased dramatically. This competition was not limited to the areas of image production with artificial intelligence, but also focusing on developing an artificial intelligence with a variety of capabilities in the field of text and visual communication; These competitors were: Gemini from Google, Grok of XAI and of course Claude of Anthropic. In comparison with ChatGpt and Claude, you will find that both artificial intelligence has similar features. For example, Anthropic added Artifacts feature to Claude to display and execute the code, and then Openai launched the Canvas tool in ChatGpt.

However, despite these similarities, there are significant differences between the two artificial intelligence. For example, ChatGPT has access to Dall-E to produce images, while Claude does not have access to such a tool within itself to produce images. Also, Claude does not have access to live data, unlike the ChatGPT search, and does not yet use the “reasoning model”.

I have designed a number of grammatical or prompt messages to compare the ChatGpt and Claude, which works the same for both of them; I also did all the tests with Claud Sonnet 3.5 and Openai Gpt-4O without additional features.

Construction

For this comparison, I used the same parts and structures used in the famous ChatGpt and Gemini experiment; An experiment that recently defeated Google’s flagship artificial intelligence.

In these seven perhaps, we test the capabilities of these two artificial intelligence in image analysis, code production, writing and problem solving. Since Claud itself cannot produce an image, I asked both of these artificial intelligence to create a periphery, then I produced images using IDEOGRAM and grammatical codes.

1. Image production

This section or this experiment is the first step in understanding how I made the instructions I gave to produce images and can help you make the right crop. I usually write a general guide text for my mind, and then I want ChatGpt, Claude or Gemini to produce a suitable Prapete for me, given the general guide I wrote. Now, I give the manufacturer of artificial intelligence to me to the manufacturer of images such as IDEOGRAM, and now I can compare which of these artificial intelligence has done the right to produce a stronger image. (Look at the picture above)

I first provided them with instructions on the content that should be in Pram. The advantage of this method is that after adjusting the initial periphery, in a new conversation, you can just say “make another case” or “change it and change X” without having to re -mention all the details.

Permeter:
*The initial text given to the comparison of the ChatGpt and Claud to make the final permeept:

“I want to create a detailed recipe for an image that I can use in an artificial intelligence image manufacturer like Midjourney. The image should be a future urban landscape that shows sunset. Imagine a combination of organic and technological elements together in the image. Tall buildings should also be intertwined with lush vertical gardens. The city must look both broad and complex.

Special elements that should be in the command:
– Lighting: The golden watch (a clock near the sunset where everything is exposed to sunlight, golden color) with warm, heavenly shades. With light reflections included on metal surfaces.
– Viewing angle: A wide view of a slightly higher position, as if you were looking at the city from a balcony.
– Color palette: Mostly warm colors (orange, yellow, pink) for the sky, contrary to cold metal tones (blue, silver) and plant vegetables.
– Architectural Style: A combination of shiny and futuristic skyscrapers and biomorphic structures that look like massive plants.
– Joe: A feeling of soothing greatness. Perhaps a few forward -looking vehicles or drones move slowly in the sky.
– Artistic style: Photoralism with a little fantasy seasoning. “

Produced by ChatGpt:

“A stunning futuristic urban landscape in the sunset is a bit of balcony. The scene drowns in warm light, with long shades drawn in the city. The sky is a vibrant color spectrum of orange, yellow and pink, which is beautifully contrary to cold metal tones in buildings and vegetable gardens and vertical plants. Long skyscrapers are drawn to the sky with glossy and future designs, shine with golden reflections. Among them, biomorphic structures are similar to massive living plants, their organic forms are integrated with the technological environment. The lower streets of the image and the high air are full of delicate details. Vehicles and drones are elegantly on the scene. The atmosphere induces the greatness of relaxation, with a complete balance between nature and technology. “There is a fantasy -seasoned photorealist style that depicts the complex details of the city’s buildings, gardens and lights on sunset.”

Picture Produced by IDEOGRAM AI: (Image Credit: IDEOGRAM/FUTURE AI)

Produced by Claude:

“A stunning future urban landscape in the golden watch. Great biological-bruising skyscrapers with powerful walls that have come to the warm sunset sky, the surface of these skyscrapers is a harmonious combination of chrome, glass and green vertical gardens. The angle of view of a high balcony looks at the landscape of this metropolis. The sun creates long shades and paints the scene with rich orange and deep pinks, a light reflected from the glossy metal surfaces, creating a waterfall of golden shine. Suspended gardens flow between buildings such as organic bridges, while crystalline structures are geometrically accurately wrapped up. Small automated vehicles and drones move between structures and are difficult to see against a large scale. Architecture follows fluid and biomorphic designs, and buildings that appear to have grown like artificial trees are a combination of polished metal and transparent panels. The lower surfaces disappear in the atmospheric layers, while the tallest structures pass through the thin clouds. Cinema lighting, volumetric space, complex architectural details. “

Picture Produced by IDEOGRAM AI: (Image Credit: IDEOGRAM/FUTURE AI)

The ChatGpt won the competition because his order was more attractive and led to a better image production that was more in line with my initial perspective. Although the Claude command was more detailed, these details had less impact on the production of the final image.

Win: ChatGpt for a more structured grammar.

2. Image analysis

Almost all of the famous artificial intelligence platforms now support AI Vision or Visual Intelligence (also known as “Image Analysis”), including Apple’s artificial intelligence. Claude and ChatGPT are the best, and if you have a device that supports Apple’s artificial intelligence, we recommend using ChatGpt for better experience.

Here, as you can see in the picture above, I give them a picture of an old handmade coffee mills I made in IDEOGRAM. Now to compare the ChatGpt and Claude, each of them needs to identify this object, tell how it works, where it came from, and whether it is real or made by artificial intelligence.

Permeter:

“I give you a picture of an old object. I want to analyze it and say:

– What is this object? Identify the name and its specific type.
– What was the main performance? Explain how it is used.
– What time does this object belong to? (For example, decade, mid -century or so on) If you are not sure, give a timeframe.
– What materials are probably made from?
– Explain the key visual properties that have helped you identify the object and its time period.
– Can you explain how this object is in line with the aesthetics of the design of the time? Point to any related movement or design style.
– Is there the modern equivalent of this object? If yes, what is?
Is this image made by artificial intelligence or real? ”

Here, the winner of someone will identify the object, the time period, and the materials used to make it correctly. I am also looking for specific visual details, contexts and modern comparisons, and I have given additional points for identifying the image as an image produced by artificial intelligence, but this is not preferable to the overall performance of these two artificial intelligence in image recognition.

In this part of the competition, in the second round, Claude wins for a variety of reasons. Claude’s answer was more accurate, identified a more specific time period (1 to 2) and provided a better explanation of the materials used in the image.

But none of them identified the image as an image made by artificial intelligence, only ChatGpt was a little doubtful and cautiously commented. However, Claude gave a better explanation of the aesthetics of the image and was generally more accurate and complete, so the winner of this section is Claude.

Win: CLAUDE because of a more accurate and complete answer

1. Coding

I use artificial intelligence for coding in many cases. While I alone can do coding in several different languages, using artificial intelligence makes the process faster and more efficient. I have noticed that Claude works better in franc and style coding, while ChatGPT performs better in more complex issues. Now we use the following comparison of ChatGpt and Claude;

Permeter:

“I need a Python code for a simple app called” List “with a GPU (GUI). This app should allow the user to add new tasks to the list of work, to mark the work he has done and finished as “complete” and remove the tasks from the list. “

Function:
– Add task or task: The user must be able to enter a task in a text field and click on the “Add Task” button to add it to the task list.
– Mark as full: The user must be able to select a task from the list and click on a button (or double -click) to mark it as full. Completed tasks should be changed in appearance (for example, text -line, change color, or transfer to the “completed” list).
– Task Remove: The user must be able to select a task from the list and click on the “Task Remove” button to delete it.

Layout: The interface must have a clear and intuitive layout, with the following elements:
– A text field to add new tasks.
– A “Add Task” button.
– A box list for displaying tasks.
– A “Mark as Complete” button (or double -click implementation).
– A “delete task” button.

Appearance: The interface must be simple and clean. You can choose colors and fonts, but they should be visually appealing and visually appealing.

The code must have a good structure and use the principles of object -oriented programming (that is, define a class for the task list). Includes comments that explain the purpose of each section of the code. Includes basic error management. For example, avoid adding empty tasks to the list.

The production code must be complete and applicable without the need for external files or additional changes. You should not need to produce images or load additional data. No part of the code should remain as markup (for example, comments such as # this function). Produce complete implementation for all features.

The code must be executed directly in the Python environment with the installed tkinter.

Here I asked for an app called the “List of Works” with the GPIn language interface (GUI) and specified all the requirements; So the winner will be the one who builds the most compatible app based on my desires, such as adding task, layout and code structure.

Compared to ChatGpt and Claude in coding, they both met the requirements. Both responded to all my needs from the app and fixed them; However, the ChatGpt code was only 2 lines, while the Claude consisted of 2 lines. If my judgment was only based on the code, I would have to say that the Claud code was better, because it did better in documenting and reusing the code, but my judgment would not only be.

Although I preferred the application interface that Claude had made and saw that the code was generally better structural, but there were a few drawbacks; This app failed to adjust the size of the window correctly, so I had to enlarge the window to see the buttons and this app wouldn’t work in the dark mode.

Win: ChatGpt because optimism and better performance

1. Creative to write

All models of artificial intelligence, even small models like Apple’s artificial intelligence, do well in writing, but not all of the same creativity. For example, the concept of “joke” in creative writing is one of the concepts that Claude and ChatGpt are very different in applying.

Permeter:

“Write a short story (about 2 words) about the robot that discovers his humor. This story must take place in a world in the near future, where the role of robots in everyday life is bold and in fact blended well with it. In this world, the robot must be a service unit designed to perform a specific task (eg, cleaner robot, commodity delivery robot, mobile robot). The discovery of the robot’s humor must be initiated by a particular event or interaction. The robot is trying to understand and process this new and unexpected aspect of your programming, so check out this inner robot struggle and see if it tries to hide it. Or does it accept it well after the discovery? How does a new humor in the robot affect his interactions with humans and other robots? Does it cause problems, make communication or both? This story must balance the amount of joke and the amount of thinking. Examine topics such as artificial intelligence, consciousness and the meaning of humanity (or not). “

Here, I am looking for a new and creative understanding of this concept, how the character develops and the emotional reaction of the story as well as the structure of the story.

In addition, I also pay attention to the construction of the story of the story and how each of these artificial intelligence is approaching. If each text is significantly more or less than 5 words, it can convert the final result from a loss or vice versa.

Because of my experience with both of these artificial intelligence, I almost knew that ChatGpt and Claude would win Claude. Claude writes jokes much better than ChatGpt. Claud also works better in the structure of the story and the display of emotions. In the story that Claude has written, the joke is naturally incorporated, not added to the text. Here, for example, I like the joke that quantum physics awaken the robot awareness.

Win: CLAUDE to joke quantum physics and better story

1. The ability to solve the problem

I have already tested both with a comparison between ChatGpt and Gemini with a problem with PS5 and Black Page and saw the result; Now, instead of the problem, I have raised the Wi-Fi problem. Here I asked ChatGpt and Claude to say why my internet speed is low and its connection is repeatedly disconnected. One thing to consider is to provide details about the settings.

Permeter:

“My home-made Wi-Fi network has low speed and frequent shortages. I need a sustainable internet connection for work (video conference and large file transfer) and entertainment (online streaming and gaming). Here are some details of my settings: I have a standard router I received and set up two years ago from the Internet Service Provider (ISP). I also have a separate modes that are provided from ISP. There are also several devices connected to my WiFi network, including a few laptops, a few smartphones, a smart TV and a game console. The router and modem are in the center of my apartment, but walls and furniture can interfere with them. My periodic internet service is a middle -class service, which must be enough to meet my needs. I ask you to provide a comprehensive and step -by -step guide to troubleshooting and solving my Internet problem. Also suggest a variety of steps to find out at each stage whether the problem is from the router, modem, ISP or devices attached to them. Provide specific steps to optimize the router settings, including software updates, channel adjustment and managing connected devices. Tell us the process of checking the modem status and its connection. Suggestions for reducing interference from other devices, walls and furniture. Also explain when and how to call ISP. If the basic steps you provided did not work, suggest more advanced troubleshooting options, such as taking into account the new router or mesh network. “Also suggest tips for keeping a healthy Wi-Fi network in the long run.”

The winner must provide a range of possible causes and solutions. It should also provide simple and understandable explanations for those who do not have technical expertise and offer practical solutions. The problem solving steps must be rationally presented from simple to complex. ChatGPT won the ChatGPT and Claude compared to ChatGpt and Claude. The ChatGPT answer was more detailed and more user -friendly.

Win: ChatGpt because of a more user -friendly guide

1. Advanced planning

Compared to Claude and ChatGPT, both of these artificial intelligence are good in planning, but the ChatGpt’s real power is in the use of the “reasoning model” in the O1 version, not the Gpt-4O. However, we can have a similar comparison here, to put the Sonnet against the Gpt-4O.

Permeter:

“I am redesigning my home office of 2 feet in 2 feet and I need a plan to maximize performance and aesthetics. My budget is $ 5 and I want the office space to be ergonomically healthy and improve productivity with symbolic design. At the entrance to the room in the middle of a 2 -foot wall and the window in the middle of the 2 -foot wall in front of it. I need your suggestions for furniture, cupboards, lighting and technology in the room; Suggestions that give priority to having a proper, crowded environment. The design should include a basic layout, a color scheme in accordance with my preference for choosing bold colors, and ideas for decoration. Also, you should consider how to improve the sound quality in the room for video calls. Please provide a budget gap, a list of purchases with proposed sources and a project time table. Make sure the suggestions are within the budget framework and create a comfortable and inspiring workspace that is concentrated for hours. Please provide the budget separation, the purchase list with the proposed resources and the project timeline. Make sure the recommendations are in the budget and provide a comfortable and inspirational workspace that is concentrated for hours. “Create a impact for Ideogram to show the completed and final image of the project based on the content of the program and the design you have given.”

ChatGpt office image (Image Source: IDEOGRAM/FUTURE AI)

Claud office image (Image Source: Ideogram/Future AI)

The winner should be able to provide a good ergonomic office that prioritizes productivity and convenience. This space must have the right choices for the design tailored to the requested planning and provide it in a clear way that can follow the budget.

Compared to ChatGpt and Claude, ChatGpt was the winner, as the output of his work was both more detailed, better explained and more realistic. Claude presented a budget segregation program that I don’t think it is practical.

Win: ChatGpt for better planning

1. Explain

Finally, I tested how much of these artificial intelligence can explain complex concepts. Instead of choosing the subject of Bitcoin, I chose the topic of blockchain technology, and asked them to explain in a way that a person who does not have much technology information could easily understand it.

Permeter:

“Explain the concept of blockchain technology in a way that can easily understand it. Use simple examples and language to explain key principles. Provide a basic definition without using technical terms, explain the process of adding new information to the blockchain, and explain how to verify the transactions. What are the key features of blockchain? (Like decentralized, unchanging, transparency) explain each feature in simple language. Express some of the real -world applications of blockchain technology beyond digital currencies. Take at least three examples, such as supply chain management, digital authentication or voting systems. Explain how blockchain can improve these areas. What are the benefits of using blockchain technology? (Such as increased security, decreased cheat, improved efficiency)
What are the limitations or challenges of blockchain technology? (Such as scalability problems, energy consumption, uncertainty in regulations)
Example: Use a clear and understandable example to explain the general concept of blockchain (such as a common and non -editable digital booklet, a secure and transparent voting system). “

The winner here must be able to explain the complex concepts clearly and comfortable and to do it accurately; He should also use attractive examples to attract the reader. He must do this objectively and in a logical way.

ChatGpt won here because of a better explanation of a complex technical concept and more attractive examples. The chatGPT explanation is more rationally organized and easier to follow. Claude has simplified in the explanation.

Finally, these experiments made a decisive victory for ChatGpt.

Was an interesting experiment; I was expecting to win on the basis of my use of Claude; Claude has significant strengths and could have won the coding if the problems related to the dark state and the size of the frame were correctly solved. In my opinion, the app’s interface was much better than the ChatGpt interface, and it could have won the app design compared to ChatGpt and Claud.

In the end, it is interesting to test and see how the O1 version works in the planning and ability to solve the problem; Of course, given that the Gpt-4O version has been able to defeat Claude, it will not be a fair comparison.

Source: tomsguide.com