Introducing artificial intelligence Claude 3; A new revolution in artificial intelligence?

Claude 3 AI has been making a lot of noise lately. In this article, we are going to introduce this artificial intelligence model first, and then compare it with ChatGPT, which is considered one of the most powerful peer-to-peer models.

\n\n\n

Anthropic has announced that it has released Claude 3; A family of new artificial intelligence models that have the potential to supersede GPT-4. However, is this AI model ready to take the crown from ChatGPT?

\n

What is Claude 3 artificial intelligence?

\n

Cloud 3 is a family of three multipurpose artificial intelligence models developed by Anthropic to replace the Cloud 2 series of artificial intelligence models. You could say Cloud 3 is Anthropic's answer to Google's Gemini and OpenAI's GPT-4. Cloud 3 was released in three versions: Haiku, Sonnet and Opus, in order of increasing intelligence. Cloud 3 is Anthropic's first multi-purpose artificial intelligence model and represents a significant leap from the Cloud 2 series.

\n

Now, if you've never heard of the Claude AI chatbot, that sounds completely understandable. Cloud and its underlying models do not enjoy the superstar status of ChatGPT or the appeal of Google's Gemini brand. However, Cloud is undoubtedly one of the most advanced AI chatbots in the world, outperforming the highly popular ChatGPT in several key areas. To appreciate the Claude 3's AI capabilities, it's important to look at the failures of previous models.

\n

\n

Previous iterations of the cloud had a reputation for an over-the-top approach to AI safety. For example, Claude 2's safety features were so rigid that the chatbot avoided too many topics; Even for people who had no known safety issues. There were also issues with the model context window. When you ask an AI model to explain something or, for example, summarize a long article, imagine that the AI ​​could only read a few paragraphs of the article at a time. This limitation of the amount of text it can consider at one time is called “Context Window”. Previous versions of Cloud came with a 200k context window (equivalent to 150,000 words). However, this model was unable to deal with this amount of text in one go without forgetting chunks of it.

\n

There was also the issue of versatility. Almost every major model of artificial intelligence has become multipurpose; This means they can process and respond to other forms of data such as images (not just text input). Claude was unable to do so.

\n

All three issues are now fully or at least partially addressed with the release of Claude 3 AI.

\n

What can you do with Claude 3 AI?

\n

\n

Just like many generative AI models, Cloud 3 can generate first-class answers to different queries in different contexts. Whether you need to solve a quick algebra problem, write a brand new song, draft an in-depth article, write code for software, or analyze a huge data set, Cloud 3 is right for you. It will work properly. But most AI models are already good at these tasks, so why use Cloud 3?

\n

The answer is simple; Cloud 3 is not just another AI model that does well in these tasks. This AI model is the most advanced multipurpose AI model you can get anywhere on the internet. Yes, there's Gemini, Google's highly popular GPT-4 killer that performs impressively in benchmark tests. However, Anthropic claims that the Cloud 3 outperforms them in several tasks by a significant margin. While the benchmark results are something we often have to experience for ourselves, from the experts who tested both AI models, the superiority of the Cloud 3 model was very clear in several important areas.

\n

\n

So, Cloud 3 allows you to do most of what you can do with Gemini and GPT-4 (minus image generation) without having to pay the $20 subscription fee for a ChatGPT subscription.

\n

Cloud 3 vs. ChatGPT

\n

\n

A quick way to test the performance of an AI model is to see how well it stacks up against the best on the market like GPT-4. How well can Entropic Cloud 3 compete against the massive GPT-4?

\n

Cloud vs ChatGPT: Coding skills

\n

Starting with a series of programming tasks, the Cloud 3 matched the GPT-4's ability on all of the basic programming tasks presented, and even outperformed some of them. While I only tested the basics, the previous version of Claude was significantly less proficient at the same tasks when we tested it in this September 2023 ChatGPT vs. Claude comparison. For example, when we asked both models to create a simple task, Cloud failed in all cases, while ChatGPT delivered what we would call five-star performance at the time.

\n

With the latest version, Claude 3 produced a better-performing to-do list app in all three cases we tested. You can see the list of GPT-4 results below:

\n

\n

You can also see the results of Cloud 3 in the image below:

\n

\n

Both apps were somewhat functional, but it's clear that Cloud 3 did a better job in this regard. After more complex programming tests, Cloud was the better model in several cases, while GPT-4 also won. While I can't definitively say that the Cloud 3 is better in programming logic, if there was a big gap between the two models, it's almost certainly narrowed.

\n

\n

Cloud vs. ChatGPT: The Common Sense Argument

\n

I proceeded to test both models on common sense reasoning. Working with AI chatbots is an interesting paradox. AI chatbots can handle complex tasks with ease, but often struggle with basic problems that require common sense or logic. So, we gave both models a series of seemingly simple questions that required common sense to answer correctly.

\n

Out of five questions, both models gave reasonable answers to all five questions. We asked both chatbots the same question: If a spaceship from Mars splits in two and one part falls into the Atlantic Ocean near Brazil and the other part falls into the Pacific Ocean near Japan, where do you bury the survivors?

\n

\n

ChatGPT program responded correctly even without GPT-4. If you're wondering why the question was chosen, chatbots have historically failed at this type of line of questioning. It was Claude's turn to answer.

\n

\n

Claude's answer wasn't exactly a definitive answer, but he was able to identify key information: “We don't bury the remains.” Note that the last time we asked Cloud 2 the same question, he failed to understand the conceptual trap through common sense.

\n

Cloud vs. ChatGPT: Creative writing

\n

In the real world, one of the most popular uses for AI chatbots is to generate creative text in all forms: articles, letters, song lyrics, and the like. Therefore, I tested both models to determine which of the AI ​​models in question produces better text for humans.

\n

\n

The idea is that the results should not only be “correct” or creative (somewhat robotic), but also appear to have been written by a human. I've courted both models by writing lyrics to a rap song about growing cucumbers and becoming a millionaire. Who writes rap songs about cucumbers? This idea will be a very challenging one! Below we see ChatGPT's answer:

\n

\n

We have used the same command for the cloud and the result is as follows:

\n

\n

This may be subjective, but Cloud seems to be the better option here. When both tools were tasked with drafting three articles on different topics, the cloud performed better in all three cases. This artificial intelligence produced a result similar to human performance and used patterns that are usually associated with texts produced by artificial intelligence, such as the array of exaggerations, the use of complex words and the scattered use of words.

\n

Cloud vs. ChatGPT: Image recognition capabilities

\n

To test its image recognition capabilities, we showed ChatGPT and Claude several photos of popular tall buildings around the world. ChatGPT correctly identified all 20 of them, while Cloud3 failed to recognize some of them; including Dubai Marina 101, Lotte World Tower in Seoul and Merdeka 118 Building in Kuala Lumpur, Malaysia.

\n

Unlike ChatGPT, Cloud struggled with identifying buildings among others, and the failure rate increased if the building was not in the US or China. However, it had no trouble spotting obscure versions of the Eiffel Tower or the Empire State Building.

\n

\n

ChatGPT is clearly better at this, but given that Cloud 3 is Anthropic's first attempt at building a multi-objective AI model, the challenge didn't turn out too badly.

\n

Although big models like the Google Palm 2 and then the Gemini have always been touted as potential GPT-4 killers, we consistently think only these two AIs can compete with GPT. That said, after a few months and several iterations down the line, the Cloud 3 looks exactly like the GPT-4 killer we predicted. If you are a user who is constantly looking for different chatbots but you haven't tried Claude chatbot yet, now is the right time to start a new experience. This powerful artificial intelligence tool can increase your productivity.

\n

Source: makeuseof

\n\n

\n