Google has recently invested over $300 million in an AI startup company called Anthropic. Anthropic is an AI-based startup that focuses on creating generative AI technologies, such as AI chatbots and large language models. These models allow users to create various types of content, positioning Anthropic as a potential rival to OpenAI and ChatGPT. What makes this investment even more interesting is that the founders of Anthropic are former employees of OpenAI who left because they disagreed with the direction of the company. In this blog post, we’ll examine Anthropic’s large language models – Claude and Claude+ – and explore their potential as competitors in the AI world.
Anthropic’s AI Models: Claude and Claude+
Anthropic has developed its own unique large language models called Claude and Claude+. These models are trained on their own dataset, allowing them to compete with well-known models like GPT-4 and ChatGPT. The Claude and Claude+ models are currently accessible through the Po.AI chatbot platform.
| Model | Description | Access |
|---|---|---|
| Claude | Anthropic’s basic large language model | Public (through Po.AI) |
| Claude+ | More powerful version of Claude | Limited access (through Po.AI) |
First Impressions of Claude
Upon using Claude for the first time, the output quality was quite impressive. The AI model was able to provide intuitive and in-depth answers to search queries. This initial performance sparked curiosity about how Claude stacks up against GPT-4 and ChatGPT in terms of output quality and capabilities.
Claude’s Performance in Different Tests
To better understand Claude’s capabilities, we conducted several tests, examining its performance in generating code, providing up-to-date information, and creative storytelling.
Generating code for a simple HTML website
When asked to provide the code needed for a simple HTML website, Claude was able to generate specific code that included a heading and two paragraphs. Additionally, the model offered explanations of terms and concepts by clicking on highlighted or bolded text.
<!DOCTYPE html>
<html>
<head>
<title>Simple HTML Website</title>
</head>
<body>
<h1>Welcome to My Website</h1>
<p>This is a simple website made with HTML.</p>
<p>Thanks for visiting!</p>
</body>
</html>
The ability to generate code and assist users in understanding the code’s components demonstrates Claude’s potential as a useful tool for web development, especially for beginners who are just starting to learn HTML and web design. With Claude’s help, users can get a head start in creating and customizing their own websites, which can be a significant advantage in today’s digital era, where having an online presence is crucial for businesses and individuals alike.
Up-to-date information and training data limitations
To explore Claude’s ability to provide up-to-date information, the model was asked about the most important specs of the iPhone 14. However, Claude was unable to provide this information, stating that the iPhone 14 had not been released yet. This limitation is similar to that of GPT-4, which is trained only up to 2021. When asked about the latest iPhone model, Claude correctly identified the iPhone 13 as the most recent release, indicating that its training data extends up to September 2021.
This limitation presents a challenge for users seeking real-time information or updates on current events, as the model’s knowledge is restricted by its training data. However, it’s worth noting that updating the training data is a common challenge faced by most AI models, and not just Claude. As AI research progresses, we can expect improvements in the frequency and speed at which AI models are updated, allowing them to provide more accurate and up-to-date information.
Creative storytelling
To test Claude’s creative capabilities, the model was asked to write a creative story involving an apple and an orange in a bin. Claude generated the following story:
The Apple sighed as he sat at the bottom of the bin. Another day, another bruise. The bin was filled with apples and oranges, and the frequent jostling left his red skin with permanent marks. “Cheer up,” said the orange next to him, “at least we’re not alone.”
This output showcases Claude’s ability to create engaging and imaginative content, which can be an invaluable resource for writers, marketers, and content creators. The ability to generate creative content on demand can streamline the content creation process, allowing users to focus on refining and personalizing the generated stories to suit their needs.
However, when asked to write the same story in the style and tone of comedian Chris Rock, Claude was unable to do so. This limitation contrasts with the capabilities of ChatGPT, which can generate content in the style of specific individuals.
Comparing Claude and Claude+
While Claude has some limitations, its more powerful counterpart, Claude+, offers additional capabilities. When asked to write the same story in the style of Chris Rock, Claude+ was able to generate the following output:
“Man, this bin is like a comedy club,” said the apple, rolling his eyes. “One minute you’re the star, and the next, you’re getting tossed in with the rest of the fruit.” The orange laughed, “Yeah, but at least we’re in this together. We’re like the fruit version of Chris Rock and Kevin Hart!”
This output demonstrates that Claude+ can generate content in the style of specific individuals, making it a more versatile tool for creative content generation. It also highlights the potential of Claude+ to be used in a variety of contexts, such as writing scripts, generating social media content, and crafting engaging marketing materials.
The comparison between Claude and Claude+ illustrates the importance of continuous AI development, as more powerful models can offer increased capabilities and better performance. As Anthropic continues to refine and improve its AI models, we can expect even more advanced features and capabilities to emerge.
Conclusion
Anthropic’s Claude and Claude+ models show great promise as competitors to OpenAI’s ChatGPT. Although Claude has some limitations, its more powerful counterpart, Claude+, offers capabilities similar to those of ChatGPT. With Google’s $300 million investment in Anthropic and a potential integration of Claude into other AI models, the future of AI chatbots and large language models is undoubtedly exciting.
As Anthropic continues to develop and improve its AI models, it will be fascinating to see how they compare to ChatGPT and other major players in the AI world. If you have specific use cases or comparisons you’d like to see with Claude, feel free to share them in the comments below. As AI technologies evolve and become more sophisticated, the possibilities for their applications in various industries will continue to expand, leading to even more exciting developments in the world of AI.
