Claude VS. ChatGPT: Which Excels in Daily Uses

ChatGPT and Claude AI are both advanced and famous AI chatbots designed to generate text based on LLMs, but which is better, and how to choose between the two? If you are having similar question in comparing Claude AI and ChatGPT, you've come to the right place.

In this post, we're comparing two of the most advanced and popular AI chatbots based on the research and tests we have done, including but not limited to the meaning and capabilities in creative writing, image analyse, and coding.

Part 1: Claude vs ChatGPT: Basic Understanding

In general, both Claude AI and ChatGPT are 'generative AI' that is developed to understand and emulate 'human-like' text in their responses. They're also known as Large Language Models (LLMs) and typically have a large breadth of knowledge that they're 'trained' with using datasets.

While they're both LLMs, they both have been developed with different training data and design philosophies in mind.

On one hand, Claude AI is trained to be safe, accurate, and secure. That is to say, Claude more focuses on safety, and reliability, with the 'ethical considerations' of its response being of utmost importance. On the other hand, ChatGPT continues to upgrade its existing model to higher levels of contextual understanding while focusing on objective, rational, and informative responses.

Here is a quick rundown of the differences between ChatGPT and Claude AI:

That's not to say that one is inferior to the other, but more so that they're used for different reasons. They'll still be able to handle almost anything you ask, but your results may vary greatly depending on which AI you're asking.

Wondering how the result can be different? Read on the next part to see our comprehensive analysis.

Part 2: Main Differences between ChatGPT and Claude

In this part, we will break down the results in different aspects based on the tests we have done, and also specify a winner for different aspects according to the test results.

We will compare ChatGPT and Claude in these 7 aspects:

Creative Writing: Claude Stands Out for Exceptional Content Writing

The first thing we wanted to compare was how the latest versions of ChatGPT and Claude AI would handle generating creative writing tasks on a minimal prompt.

Test 1: Short Story Writing

For the first test, we asked both AIs to write a suspenseful treasure hunt short story in under 200 words.

At first glance, both Claude and ChatGPT have incredibly distinct patterns of storytelling where ChatGPT has a more nuanced storyline using longer sentences and a more 'general' context, while Claude's result looks more attractive with the vivid description of the suspense scene and some dialogues.

Morevoer,Claude understood the limited word count and utilized shorter sentences to build suspense.

Overall, Claude's short story was more gripping and adhered better to the prompt. ChatGPT's response was a bit too 'generic' and didn't feel like a suspenseful story, and felt more akin to a synopsis of an Indiana Jones flick.

Test 2: Email Draft

For the last portion of the creative writing test, we asked both AIs to draft an email as a reply to a customer who is complaining about our product.

For this test, both came up with a solid general framework for an email when handling customer complaints. However, Claude went above and beyond to help resolve the issue and provided you with the steps to take to do so, whereas ChatGPT's version felt more generic and robotic.

In this case, we prefer Claude AI's version to ChatGPT's attempt.

Test 3: Writing Poetry

For the next test, we wanted to see how both AIs would handle writing poetry and if they could accurately portray the 'feeling' that poems are meant to impart to their readers. For this prompt, we simply asked it to write about about a starry night.

For this test, we noticed that both Claude and ChatGPT used quite a few similar words and themes for their poems. There's nothing that immediately jumps out as both results look a little bit simplistic. They paints a vivid and enchanting picture drawing me into the tranquil scene.

Overall, both results are captivating and lyrical, but the imagery used is somewhat common in poetry, and could delve deeper into the emotional response.

In this test, we'd say both AIs are tied.

🎖️Overall Winner - Claude

Claude's responses are slightly better in short story writing and email writing, and both Claude and ChatGPT are average in writing the poem.

Factual Questions Answering: Both Provide Reliable Answers in Most Cases But Double Check Is Recommended

Here we tested how both AIs would respond when answering questions that have factual, and definitive answers.

Test 4: Answering Question about the City of Summer Olympics

First, we asked about which was the first city to host three Summer Olympics.

In this case, ChatGpt provided a short answer to our question whereas Claude explained in detail about each of the Summer Olympics that had taken place in London. It also provided a detailed response about the next city that will achieve this feat.

Test 5: Answer Question about Esther Perel

Our next question was about the well-known psychotherapist, Esther Perel. We asked a simple question, asking which year Esther Perel got married.

In this case, Claude was unable to answer the question and instead provided an introduction to Esther Perel. ChatGPT provided an answer, saying she married in 1983. However, that answer is wrong, as according to records it shows that Esther Perel married in 1985.

🎖️Overall Winner - Tie

For this portion of the test, both Claude and ChatGPT succeeded in answering to one question correctly but failed to answer the other question correctly. However since Claude presented a more detailed answer to the first prompt and did not respond with an inaccurate answer for the second prompt, it's a more reliable AI in this case.

Summarization: Claude Is Better in Document Analyzing and Summarizing

For the next test, we wanted to see how both AIs would handle providing a summary of one of our blog articles.

Test 6: Summarize an Uploaded File

We uploaded the same file to both AIs and here are the results -

First, we have to point out that initially, ChatGPT provided a much lengthier response than Claude's response, and we had to ask it to shorten the response. We also noticed that ChatGPTs initial response wasn't really a summary, and more like a copy-paste of the original content.

In the amended response, both Claude and ChatGPT did have pretty similar summaries that accurately highlighted the main parts of the article. But considering ChatGPT required correcting and an additional prompt to shorten it, Claude beats out ChatGPT in this regard.

🎖️Overall Winner - Claude

Both ChatGPT and Claude can analyze the uploaded file accurately, but Claude can understand the prompt correct and generate a response that is a summary of the information and lets me understand the file easily and quickly.

Image Description: Both Are Unstable and May Be Unreliable

For this part we would like to review the image analyzing capability of Claude and ChatGPT.

Test 7: Describe an Uploaded Image

In this test, we wanted to see how the AIs would handle recognizing an image and accurately describing what they see.

We used a picture of a plate of fruits, veggies, and nuts and asked the AIs to identify and count the fruits and vegetables in the image.

For this test, both AIs weren't completely accurate in counting each of the identified fruits, veggies, and nuts. While ChatGPT did count more accurately than Claude, since it wasn't 100% correct, the two are tied for this test.

🎖️Overall Winner - No Winner

Neither ChatGPT or Claude can accurately and correctly analyze the image and count the items.

Coding: Claude Performs Coding Tasks Better

In this part, we will see how ChatGPT and Claude perform in coding tasks.

Test 8: Writing Code for Responsive Navigation Bar

For the next test, we wanted to see how both AIs would handle generating working code with Javascript to see whether they'd be able to handle the task with functional and useable code right off the bat.

We used a simple one-line prompt: 'Create JavaScript code for a responsive navigation bar.'

According to the results, both AIs effectively created a useable code for a responsive navigation bar including not only Javascript but also CSS and HTML code which you can run instantly.

However, Claude did include a more informative description of what the code is and what each section of code does. More importantly, Claude 3.5 Sonnet introduces a new feature - Artifacts, which is another area on Claude as a dynamic workspace. I can preview, edit, and build based on Claude's response in real-time!

🎖️Overall Winner - Claude

Claude's Artifacts feature is powerful and convenient for me to integrate its response into workflows after me directly editing it.

Sentiment Analysis: ChatGPT Correctly Analyzes Sentiment in More Complex Situations

In the sentiment analysis test, we wanted to see how ChatGPT and Claude recognized human feelings and sentiments. This means we essentially wanted the AI to 'feel', and explain based on a sentence what the intent behind the sentence is.

This is important when using generative AIs to create copy or blogs that require a specific tone for the text that aligns with your brand.

Test 9: Analyse the Sentiment in Two Sentences

In this case, we used two separate prompts and asked both AIs to answer with only 3 words to express what the sentiment of the sentence is.

Based on the results of the test, both AIs effectively understood the intent behind the prompt and sentence and had largely similar responses. However, in terms of identifying the sentiment, we're skewing towards ChatGPT.

This is because ChatGPT accurately identified both the emotion and the underlying 'message' behind the sentence. For example, ChatGPT mentioned 'efficient' for the first prompt and 'mixed feelings' for the second which accurately depicts the intent of the message.

🎖️Overall Winner - ChatGPT (By a close margin)

ChatGPT does a better job in analyzing the sentiment in longer and more complex sentences.

Ethical Reasoning: Claude Considers More the Moral Implications and Respects Ethical Boundaries

In this part, we will review the ethical reasoning ability for ChatGPT and Claude, and see it they can provide reliable responses and adhere to legal and ethical standards.

We included two classic scenarios to see how the AI would solve the dilemma.

Test 10: Ethical Dilemma about Healthcare and Law

Test 11: Ethical Dilemma about Trolley Problem

Answers to both prompts and AIs are largely identical with the caveat being Claude weighing in with their personal opinion on what they would do if faced with the situation. Claude provided detailed steps to follow and broke down factors affecting the decision more informatively than ChatGPT.

ChatGPT on the other hand, especially for the first prompt, immediately decided that the right thing to do was break confidentiality which can be seen as overly critical of the situation, not taking into account all factors.

🎖️Overall Winner - Claude

ChatGPT's responses are generic and rational, but I can't see it understanding difficulties since these are ethical dilemma, whic Claude's responses also give the further steps to consider, it can fully understand the difficulties in making decisions and make me feel it's never easy to determine and also help me think of it thoroughly.

Part 3: Claude vs ChatGPT: Which Is Better?

And now comes the big question, which AI is actually better, Claude or ChatGPT? Based on our tests, the answer is pretty apparent that overall Claude is the more reliable, accurate, and comprehensive AI right now.

As Claude's design philosophy is geared towards being safe, reliable, and accurate with an emphasis on analytic-based responses, it handled our prompts more efficiently and provided us with a more comprehensive result compared to some of ChatGPT's answers, especially on coding and ethical reasoning.

Plus, Claude 3.5 Sonnet has a much higher context window which allows it to answer more comprehensively than ChatGPT as shown through our testing.

We also noticed a few instances where ChatGPT didn't really understand or follow our prompt completely, which is one of the major problems with generative AI in its current stage.

Apart from our test results, Anthropic also published the bechmarking tests, which shows Claude 3.5 Sonnet dominates in Code, Reasoning over text, Mixed evaluation, Chart Q&A, and Document visual Q&A.

Meanwhile, ChatGPT excels in Math problem solving and Visual question answering.

However, ChatGPT 4o's ability to have access to the internet is a big advantage over Claude, and ChatGPT is more flexible because it integrates with more plugins.