Claude

Updated for version: Sonnet 3.5

Accessible via: https://claude.ai/

Requires login via: Google, E-mail

Ratings

Accuracy / Quality ★★★☆☆

Flexibility / Features ★★★☆☆ 

Data security / Privacy ★★★★☆

Pros/cons

Pros  

  • Safety centered AI tool - Does not train on your data unless explicitly allowed.
  • Superior coding ability.
  • Ability to remove chat history from back-end.
  • Ability to analyze long documents for free.

Cons  

  • No internet access.
  • Absence of chatbots / chat customization.
  • No image generation.
  • Limited access during high server usage.

Description

Claude Sonnet is a Large Language Model made by Anthropic AI. Claude stands out in terms of privacy and AI safety. The model is trained on a large volume of current data sources, and cannot access internet sources. By training the model on a set of foundational principles aligned with human values Anthropic aims to offer a “helpful, harmless and honest” AI model. Moreover, user data files and chat data are not used to further train or train the model. In the free version, you can use the most advanced version of Claude (Claude Sonnet 3.5) with a limited number of prompts per day. 

Features and examples

Artifacts

Apart from the chat function, Claude has the ability to create and manage content in “Artifact” form. Artifacts are AI-generated pieces of content like code snippets, documents, webpages, diagrams and many more. The artifacts open in a separate window within Claude, which allows users to better distinguish the documents, code and other types of artifacts from the chat. Moreover, it is made easy to view, copy and edit the contents of existing Artifacts and documents. In the example below, we ask Claude to generate a webpage to play the Sudoku game described in the uploaded report. 

This question is  answered by Claude by the creation of a Artifact and a detailed message explaining the code in the Artifact. On the top right, we can open the chat controls, where there is an overview of all documents and Artifacts used in the current chat.

We can now open the Artifact through the Chat controls, or by clicking on the "simple Sudoku Game" button in the chat.

The code generated for this Artifact is now opened. By ticking the switch in the top right corner, we can get a preview of the game.

As can be seen above, a relatively simple Sudoku game is made. We can improve this game by asking Claude to improve the game using the Chat function, or selecting a specific piece in the code and asking for an improvement. In this case we ask for difficulty levels, a timer, a hint system and the ability to generate new puzzles using the chat box.

After this, Claude produces an improved webpage Artifact (without deleting the first one) and once again explains the code it generated. Lets say we now finished our code and want to share it. We can easily publish this artifact by going to the chat control in the top right corner. Then we select the artifact we want to publish, followed by a click on the Publish button on the bottom left corner.

The published artifact can then be opened by copying and pasting the link provided by Claude. If you want, you can remix this Artifact using the "Remix Artifact" button on the bottom right corner.