6 Practical Tips for Using Anthropic's Claude Chatbot - Latest Global News

6 Practical Tips for Using Anthropic’s Claude Chatbot

Joel Lewenstein, head of product design at Anthropic, was recently crawling under his new house to adjust the irrigation system when he encountered a puzzle: The device’s buttons didn’t make sense. Instead of searching the internet for a product manual, he opened Anthropic’s Claude chatbot app on his phone and took a photo. Its algorithms analyzed the image and provided more context for each button’s possible function.

When I tested OpenAI’s imaging capabilities for ChatGPT last year, I found them similarly useful – at least for low-stakes tasks. I would recommend you turn to AI image analysis to identify those random wires in your house, but not to guess the identity of a loose prescription pill.

Anthropic released the iOS app that Lewenstein made available for everyone to download earlier this month. I decided to give the Claude app a try, in line with my goal of experimenting with a wider variety of chatbots this year. And I video chatted with Lewenstein to learn his advice for getting started with Claude and how to ask questions in a way that gets the most useful answers.

Get conversational

Through decades of Google search dominance on the web, we’ve learned to type clear and concise search queries when we want something. To get the most out of chatbots like Claude, you need to move away from this approach. “It’s not Google search,” says Lewenstein. “So you’re not typing in three keywords – you’re actually having a conversation about it.” He encourages users to avoid an overly utilitarian communication style and be a little more detailed with their prompts. Instead of a short phrase, try writing prompts that are a few sentences or even a few paragraphs long.

Share photos

AI image analysis is still fairly new to Anthropic’s chatbot – it was released in March – but can provide a powerful way to quickly ask questions of the chatbot. Lewenstein recommends using images as a starting point for conversations with Claude, like he did underneath his house. While the feature may not always be accurate, it is useful – and fun – if you keep the limitations in mind and look for opportunities where an image can answer your query.

Be direct

Still not getting the results you want? A solid troubleshooting technique is to over-prescribe in your prompts. “Just talking to Claude like a human actually leads you a little astray,” says Lewenstein. Instead, try to give Claude an almost clumsy amount of context about how the answer should be formatted – say, for example, that it should be in bullet points or short paragraphs, and give her clear direction on the tone she should use . Do you want lyrical answers or something that sounds more technical? Also consider telling Claude who the target audience is and what level of knowledge they might have on the topic.

Try it again

If your first query to Claude doesn’t produce a good result, remember that your first query is just the starting point. Follow-up prompts and clarifying questions are crucial to guiding a chatbot in the right direction.

When I interact with a chatbot, I quickly start a new conversation thread if the output goes wrong so I can try a different opening prompt. That’s not the best approach, says Lewenstein.

He suggests staying in the same chat window and giving the bot direct feedback about what you want to do differently, from tone to structure. “I literally just type, ‘No, too complicated.’ I don’t understand what these words mean. Can you try again, but simplify it one more level,” says Lewenstein, pointing to a time when Claude’s summary of a document was confusing.

Upload large documents

Speaking of documents, Claude’s ability to analyze uploaded data is one of his strengths. The possible uses for this are more in workplace use cases, where the chatbot can help with Excel spreadsheets and overflowing email inboxes, but it can also be a useful function outside of the office. When you upload batches of text, Claude can spot trends you might not have noticed otherwise. Ask the chatbot to look for patterns in language usage or the topics covered. Do you have a PDF that you need to read but it’s so long it’s making your eyes glaze over? Claude can help you focus your attention on the most important aspect of the document first.

I uploaded the text transcript of my conversation with Lewenstein to Claude and asked which quotes it highlighted as important. The chatbot captured the key topics of the conversation flawlessly and highlighted many of the quotes that I ultimately selected for this newsletter. (Anthropic’s policies mean that your input data will likely not be used to train Anthropic’s AI models unless you agree.)

Text like you’re friends

Yes, you should experiment with writing longer and more specific prompts to Claude, but it’s also smart to think of conversations with chatbots as a barrage of messages going back and forth. “I actually think the mobile app is a natural form factor for this because you’re constantly chatting with people on your phone,” says Lewenstein.

When I uploaded a photo of a robot mural I saw in a cool bar in San Francisco to the Claude app, the chatbot provided a poetic description of the artwork. There was no way to guess which city the bar was in, a nearly impossible task, but the rhythm of the conversation felt like sending a message to an eager friend. Claude thanked me when I finally revealed the bar’s location: “My suspicions were beautifully turned on their head.”

I’ll need to use it more to really understand Claude, but I already feel like the chatbot’s output has a friendly touch. Although ChatGPT is still my chatbot of choice, I could see adding Claude into the mix if I want to send messages using an AI tool that prefers engaging, human-sounding results over a drier, more efficient communication style. It’s important to stay open to using AI tools you’ve never tried before. Chatbots are constantly improving and changing rapidly, so it’s far too early to commit to a single tool.

Sharing Is Caring:

Leave a Comment