Claude AI is the new kid on the block; is it better than ChatGPT?

Amid all the AI excitement with ChatGPT from OpenAI and Google’s Bard, it’s honestly becoming tougher to get excited about yet another chatbot. That’s exactly how it might feel when you hear about Claude, a fairly new chatbot from Anthropic, based in San Francisco.

Even though Claude might seem like your run-of-the-mill chatbot at first glance, if you delve a bit deeper, you’ll discover that it adheres to a distinctive AI safety-focused philosophy. What does that imply, and could Claude be a game-changer in the competitive world of chatbots? Let’s dive into it.

Claude AI explained

Claude is a chatbot powered by a hefty language model, similar to ChatGPT or Bing Chat. The folks who made it put a strong focus on AI safety. They claim Claude is all about being “helpful, harmless, and honest.” In other words, Claude won’t throw any curveballs in its responses, unlike the unpredictable hallucinations we’ve seen from ChatGPT in the past.

In general, Claude seems and acts pretty much like any other chatbot you’ve likely messed around with in the past year. Some users say it comes off as more chatty compared to its competitors. On top of that, Claude can tackle tricky tasks like summarizing, coding, and spotting patterns in hefty datasets. According to Anthropic, this chatbot can juggle larger chunks of context than ChatGPT, allowing it to remember more details during extended conversations.

Claude stands out from the crowd due to its unique training process. Anthropic calls it “constitutional AI,” which boils down to a bunch of human-written rules that the AI has to follow. As time goes on, the language model learns to generate safer text by itself, guided by its constitution or core principles. This is different from GPT-4 and other language models that depend on human feedback to prevent them from spitting out harmful content.

In a nutshell, Anthropic has crafted an AI that learned to follow the rules of society on its own. This means it relies less on constant human oversight. Claude is essentially the front-end chatbot representation of this autonomous AI. Besides, using Claude won’t cost you a dime. Just like ChatGPT and many other chatbots, all you have to do is sign up for a free account.

Anthropic also has a Pro subscription available, and it’s priced at $20 per month. With this plan, you can chat with Claude more frequently, enjoying up to five times the usage limit compared to the free version. I’ve hit the rate limit a couple of times on the free plan, but if your messages are shorter and to the point, you can make your allowance last longer. Plus, Pro subscribers get early access to new Claude features and priority access during busy times.

Also Read: AI hallucinations could be a huge threat to science, experts warn

Who’s the brain behind Claude AI?

Claude is the brainchild of the San Francisco startup Anthropic, and they still call the shots. Anthropic came into being in 2021, spurred by Microsoft stepping up its game with OpenAI. The founding faces of this startup are none other than Dario and Daniela Amodei, former VPs of Research at OpenAI. They teamed up with a crew of other ex-OpenAI researchers who were part of the GPT family of language models.

Anthropic has raked in quite a sum in investments, crossing the $7 billion mark by late 2023. Google was an early big spender, dropping $300 million through its cloud division to snag a 10% slice of Anthropic. Amazon jumped on the bandwagon too, pledging a hefty $4 billion. It seems Google was so impressed that they kicked in an extra $2 billion in funding for the startup.

Claude AI pretty much covers all the bases that a chatbot like ChatGPT does, but with a focus on being safer and more helpful. It’s got the chops to tackle questions, whip up creative content, handle code writing and explanations, translate languages, and dive into regular chit-chat. So, throw a prompt its way like “Craft a quick tale set in a cyberpunk megacity where robots and humans live together,” and then ask for tweaks and enhancements.

Can Claude AI beat ChatGPT?

Even though they might seem alike at first glance, Claude and ChatGPT have some significant differences in their inner workings. Check out this table for a technical breakdown.

Training Approach
ChatGPT: Utilizes reinforcement learning from human feedback.
Claude: Relies on self-supervised learning within Anthropic’s Constitutional AI framework.

Document Uploads
ChatGPT: Available with a paid subscription.
Claude: Included in the free tier; no extra charge.

Internet Access
ChatGPT: Requires a paid subscription for internet access.
Claude: Even with a paid subscription, internet access is not permitted.

Response Style
ChatGPT: Offers informational responses, functioning like a digital assistant.
Claude: Prioritizes conversational interactions with shorter default responses.

Language Support
ChatGPT: Wide language support due to diverse training data.
Claude: Officially supports English and “common languages.”

Privacy Practices
ChatGPT: User chats are saved for future training by default.
Claude: User chats are not used for training purposes.

Message Limits per Day
ChatGPT: Allows a higher number of messages per day.
Claude: Has a lower limit, which may increase over time.

Also Read: AI can mimic human social skills in real-time, should we be scared?

How to use Claude AI

In the past, you had to use Slack to get to Claude, but now you can use it as a standalone app on the web. Here’s a simple guide on how to use this chatbot:

Here’s a quick rundown on how to get started with Claude:

• Go to claude.ai, pop in your email, and hit “Continue.”
• Check your email for a confirmation code; type it in to move forward.
• Toss in your name, give Anthropic’s terms and privacy policy a read, and then accept them.
• Last step: Enter and verify your phone number.
• Voila! Now you’re all set to chat with Claude for free. The setup looks just like ChatGPT — type in your prompt, hit send, and wait for Claude to finish up before shooting over another message or asking for a response redo.

You’ve got the option to upload your own docs for Claude to dig into. Just click on the little attachment icon in the text box. The app will grab the text from your document and fire it over to the chatbot along with your prompt. Easy peasy!

Say you want Claude to spill the beans on specific details from a hefty document or just give you the lowdown on the entire thing. Keep in mind, though, Claude can’t tackle Excel or XLS files at the moment. No biggie – you can upload structured data in CSV format as a workaround.