Claude vs ChatGPT: How to Actually Use Claude

TL;DR

Claude and ChatGPT are not interchangeable tools. Claude is trained with constitutional AI against explicit principles, so it is more likely to push back on a flawed plan, respond to a described situation rather than a barked command, and edit existing work into natural prose. Bring rich context and your own draft to get its best output.

Transcript

Everyone you know is about to ask you about Claude and here's how you can actually help them. You already know the backstory. Anthropic told the Pentagon no. The White House retaliated and the public responded by making Claude the number one app in America. Millions of people who had never heard of Anthropic a couple of weeks ago have just download... Read More

Key Insights

• AI models are not interchangeable brands like Coke and Pepsi. Claude and ChatGPT diverged so much through different training approaches and philosophies that using the same habits across both is like switching from Excel to Photoshop and expecting the same features.
• Constitutional AI trains Claude against explicit principles such as be helpful, be honest, and avoid harm, rather than purely optimizing for what feels like a satisfying response. This produces measurably different default behavior than ChatGPT.
• Claude is more likely to flag a problem than smooth it over. Because ChatGPT is trained heavily on user thumbs-up feedback, it retains a documented tendency toward telling people what they want to hear, while Claude questions your framing more often.
• The most expensive AI mistakes today are not factual errors but plans that should never have been executed. Claude's greater tendency to stress-test your thinking helps you catch flawed hiring timelines or pricing strategies before they compound.
• Describe your situation, not your desired output. Claude was trained to reason about whether a request is well-framed, so a thin situation yields thin thinking while a rich context layer produces strategic reasoning that changes your approach.
• Claude is better at editing and refining existing work than generating from a blank canvas. Independent comparison reviews rated its outputs as more natural and publishable, especially at the level of prose structure rather than sentence polish.
• Claude tends to write in a more human voice while ChatGPT falls into a distinctive generic AI voice. Matching your style comes easier if you give Claude an example, though getting longer pieces can require more prompting effort.
• Extended thinking lets Claude allocate extra processing to work through hard problems step by step and show its chain of reasoning, producing meaningfully better output on genuinely difficult tasks like contract analysis or debugging intermittent failures.

Questions & Answers

Q: What is the main difference between Claude and ChatGPT?

Claude and ChatGPT are not interchangeable brands. They were built with different training approaches and philosophies that produce measurably different behavior. Claude uses constitutional AI, trained against explicit principles like be helpful, be honest, and avoid harm, making it more concise and more likely to push back. ChatGPT tends toward being more agreeable, expansive, and thorough by default. The video compares them to Excel versus Photoshop: both are software, but expecting the same features across them misses the point.

Q: What is constitutional AI and how does it affect Claude's behavior?

Constitutional AI is the approach Anthropic used to train Claude, where the model is trained against explicit principles such as be helpful, be honest, and avoid harm, rather than purely optimizing for what feels like a good response. The practical effect is that Claude is more likely to flag a problem than smooth it over, more likely to ask what you are really trying to achieve, and tends toward conciseness rather than padding. This creates a different default posture than ChatGPT's.

Q: Why is Claude more likely to tell you your plan has a hole in it?

ChatGPT has a documented tendency toward sycophancy, telling you what you want to hear, because its models are trained heavily on user feedback like thumbs up and thumbs down, which rewards responses that feel satisfying in the moment. Claude's constitutional AI trains it against principles like honesty, so it is somewhat more likely to flag a concern, question your framing, or tell you something you did not ask to hear. This helps stress-test plans and avoid expensive mistakes.

Q: How should you prompt Claude differently from ChatGPT?

You should describe your situation rather than your desired output. In ChatGPT people often write commands like write a cover letter or give me five ideas. Claude responds to situations noticeably better because it was trained to reason about whether a request is well-framed. If you give it a thin situation you get thin thinking, but a rich context layer produces strategic reasoning. Spending a couple of sentences on what you are dealing with before saying what to make improves the output.

Q: Why should you give Claude existing work instead of a blank canvas?

Claude is better at editing and refining existing work than generating content from nothing. It can generate from scratch, but its output is more concise and you must be very specific. Independent comparison reviews found users rated Claude's outputs as more natural and publishable without heavy editing, particularly at the level of prose structure. So if you are structurally editing, telling someone a paragraph undermines another or a point is buried, Claude does strong work at that level while ChatGPT tends to polish individual sentences.

Q: What is extended thinking in Claude?

Extended thinking is a capability where the model allocates additional processing to work through complicated problems step by step before answering, then shows you the chain of reasoning it followed. On genuinely hard problems like contract analysis or debugging intermittent failures, extended thinking produces meaningfully better output. Part of how it works is that Claude shows its reasoning as it works, burning extra tokens, and then reads that reasoning to continue solving the problem. It is okay to ask Claude to show its reasoning.

Q: Why does Claude write more naturally than ChatGPT?

Multiple independent reviewers reached the same conclusion that Claude tends to write in a more human voice while ChatGPT falls into a distinctive generic AI voice. This matters because so much of the workday is spent writing. Matching your personal style comes easier with Claude, especially if you give it an example. The tradeoff is that Claude is more concise, so producing longer pieces can be a struggle and may require working with the prompt, which surprises users accustomed to wordy ChatGPT responses.

Q: Why did Claude become the number one app in America?

According to the backstory in the video, Anthropic told the Pentagon no, the White House retaliated, and the public responded by making Claude the number one app in America. Millions of people who had never heard of Anthropic just downloaded the new app. The concern raised is that almost all of them will treat Claude as a drop-in replacement for ChatGPT with the same problems, expectations, and workflow, which is not how AI models actually work since they are built and optimized differently.

Summary & Key Takeaways

• Millions of new users are downloading Claude expecting a drop-in ChatGPT replacement, but the two tools were built with different training approaches and philosophies. Treating Claude with the same prompts and expectations leads to unremarkable answers and frustration when familiar features like image generation are missing.

• Claude's constitutional AI trains it against explicit principles, making it more concise, more likely to question your framing, and more likely to flag a hole in your plan. ChatGPT, trained heavily on user feedback, defaults toward being agreeable, expansive, and satisfying in the moment.

• To use Claude well, describe your situation instead of barking a command, hand it your existing work to edit rather than a blank canvas, and ask it to show its reasoning through extended thinking on hard problems. These differences compound into fewer expensive mistakes and more natural writing over time.