Claude 2 vs ChatGPT: The Battle of the AI chatbots

Large language models (LLMs) are powerful AI tools that can process and generate natural language. They use deep learning to learn from huge amounts of text data and perform various tasks such as coding, math, and reasoning. One of the latest LLMs is Claude 2, introduced by Anthropic on Tuesday. It is similar to ChatGPT, but it has some new features and improvements.

At present it in available as a beta website that anyone can try. It also offers a commercial API for developers. Anthropic says that Claude is like a helpful colleague or personal assistant that can converse, explain, and remember. The new version also incorporates feedback from users of the previous model.

According to Anthropic, Claude 2 AI has advanced in three key areas: coding, math, and reasoning. For example, it scored 76.5% on the Bar exam multiple choice section, up from 73.0% with Claude 1.3. It also scored above the 90th percentile on the GRE reading and writing exams, and similarly to the median applicant on quantitative reasoning.

The advantages of Claude 2 over ChatGPT

A sketch of a human being interacting with AI
Claude 2 is a rival to ChatGPT

The biggest enhancement of Claude 2 is its longer input and output length. It can handle up to 100,000 tokens (fragments of words) in a single prompt. This allows it to analyze long documents such as technical guides or entire books. It can also create longer documents as outputs. That is far greater than what ChatGPT is capable of.

Claude 2 AI also showed increased proficiency in coding capabilities. It improved its score on the Codex HumanEval, a Python programming test, from 56 percent to 71.2 percent. Similarly, it improved from 85.2 to 88 percent on GSM8k, a test comprising grade-school math problems.

One of the main goals for Anthropic has been to reduce the likelihood of “harmful” or “offensive” outputs from Claude 2. This is a challenging and subjective task, but they claim that their internal evaluation showed that “Claude 2 was 2x better at giving harmless responses compared to Claude 1.3.”

However, LLMs like Claude 2 are not perfect. They sometimes make things up or produce inaccurate results. Therefore, they should not be used as factual references but as tools to process data that you provide and verify. Anthropic writes: “AI assistants are most useful in everyday situations, like serving to summarize or organize information, and should not be used where physical or mental health and well-being are involved.”

Claude 2 is now available for general use in the US and UK for individual users and businesses via its API.


On Key