Is the NEW JetBrains AI Assistant BETTER than Copilot? | July 2024 Update

Marco Lenzo
10 Jul 202404:29

TLDRJetBrains' AI Assistant has seen improvements in code completion for Java and Kotlin, with reduced latency and locally powered single-line suggestions. However, it still lags behind Copilot and Codeium in responsiveness and contextual understanding. The assistant's suggestions are often smaller snippets and can be less relevant, requiring more context to provide useful code. Despite updates, issues like including imports within the class body persist. The reviewer has canceled their JetBrains subscription, favoring ChatGPT Plus for its superior code suggestions and speed, and using Codeium for its free individual use and quality similar to Copilot.

Takeaways

  • 🚀 JetBrains announced improved AI Assistant code completion in July 2024.
  • 🔍 To test the new AI Assistant, you need the 2024.2 early access program version 7 of IntelliJ.
  • ⏱️ JetBrains claims reduced latency and higher quality code suggestions with their own models.
  • 🔄 The AI Assistant provides single-line and multi-line code completion for Java and Kotlin.
  • 📉 The assistant's latency has improved but is still less responsive than Copilot and Codeium.
  • 📝 Single-line suggestions are prioritized to reduce perceived latency, with multi-line suggestions following if available.
  • 📊 The AI Assistant's suggestions are often smaller code snippets compared to Copilot or Codeium.
  • 🔍 The assistant requires more context to provide suggestions, unlike Copilot and Codeium which capture project context better.
  • 🛠️ The prompting system is still slow and sometimes includes imports within the class body incorrectly.
  • 🔄 The reviewer canceled their JetBrains AI Assistant subscription due to it not being competitive with other products.
  • 🏆 The combination of Microsoft, GitHub, and OpenAI makes GitHub Copilot a tough competitor to beat.
  • 💡 The reviewer prefers ChatGPT Plus for its capabilities and speed, despite initially favoring inline completion.
  • 👥 For enterprise use, the reviewer recommends Copilot for its GitHub integration, while Codeium is used for personal projects due to its free individual use and similar quality to Copilot.

Q & A

  • What did JetBrains announce on the first of July regarding their AI Assistant?

    -JetBrains announced improved code completion in their AI Assistant, which is powered by their own models, aiming to provide higher quality and reduced latency.

  • What version of IntelliJ is required to try out the improved AI Assistant?

    -To try out the improved AI Assistant, you need the 2024.2 early access program version 7 of IntelliJ.

  • How does the new AI Assistant perform in terms of latency compared to previous versions?

    -The new AI Assistant shows a definite improvement in latency, with inline completion being timely and usable, although still less responsive than Copilot and Codeium.

  • What are the two different ways IntelliJ performs code completion?

    -IntelliJ performs code completion in two different ways: single-line and multi-line. Single-line suggestions are shown first and are powered by a model deployed on the device.

  • How does the quality of the suggested code from JetBrains AI Assistant compare to Copilot or Codeium?

    -The JetBrains AI Assistant often provides very small snippets of code compared to Copilot or Codeium, which may not necessarily lead to more relevant suggestions and requires more context to give suggestions effectively.

  • What issue with the AI Assistant's code generation was the reviewer hoping to see resolved in the new version?

    -The reviewer was hoping that the issue of the assistant including imports within the class body would be addressed in the new version, but it was not resolved.

  • What is the reviewer's conclusion regarding the JetBrains AI Assistant in comparison to the competition?

    -The reviewer concluded that despite some improvements, the JetBrains AI Assistant is still not on par with the competition, such as GitHub Copilot, and they canceled their subscription.

  • What is the reviewer's opinion on the combination of Microsoft, GitHub, and OpenAI in the AI assistant market?

    -The reviewer believes that the combination of Microsoft, GitHub, and OpenAI will be very difficult to beat in the AI assistant market due to their mature products.

  • Why did the reviewer subscribe to ChatGPT Plus and what is their experience with it?

    -The reviewer subscribed to ChatGPT Plus for $20 a month because they were impressed by the capabilities of the GPT-4o model, which produces the best code suggestions they have seen and is incredibly fast.

  • What is the reviewer's personal preference for an AI assistant integrated into their IDE and why?

    -The reviewer prefers using Codeium as the AI assistant integrated into their personal IDE because it's free for individual use, fast, and provides suggestion quality similar to Copilot.

  • What does the reviewer suggest for viewers to do if they find the video useful?

    -The reviewer suggests that if viewers find the video useful, they should like and share it so that everyone can learn something new.

Outlines

00:00

🚀 JetBrains AI Assistant Update Review

The script begins with the announcement of improved code completion in JetBrains' AI Assistant, prompting the author's subscription for a test. The author aims to compare it with GitHub Copilot and evaluates it on latency, responsiveness, and code suggestion quality. The AI Assistant is noted to have faster inline completion than before but still lags behind Copilot and Codeium in responsiveness. The script discusses the two types of code completion offered by IntelliJ: single-line, powered by a local model for reduced latency, and multi-line, which follows after accepting a single-line suggestion. The quality of the AI's suggestions is criticized for being too small and context-dependent, contrasting with the broader and more project-aware suggestions from Copilot and Codeium. The author also expresses frustration with the AI's prompting system, particularly its inclusion of imports within the class body, which was hoped to be fixed in the new version but was not. The conclusion sees the author canceling the JetBrains AI Assistant subscription due to its inability to compete with more mature products like GitHub Copilot, which benefits from the combined strength of Microsoft, GitHub, and OpenAI. The author's preference now leans towards ChatGPT Plus for its impressive GPT-4o model, which provides fast and high-quality code suggestions. The author's personal setup favors Codeium for its free individual use, speed, and similar suggestion quality to Copilot, while Copilot is preferred in an enterprise context due to its GitHub integration. The script ends with an invitation for viewers to share their setups and to like and share the video for educational purposes.

Mindmap

Keywords

💡JetBrains AI Assistant

JetBrains AI Assistant refers to a software tool developed by JetBrains that uses artificial intelligence to assist in coding tasks. It is designed to improve code completion and enhance the coding experience. In the video, the assistant is compared with other AI coding tools, such as GitHub Copilot, to evaluate its performance and usefulness. The script mentions that JetBrains claims improvements in single and multi-line code completion for Java and Kotlin.

💡Code Completion

Code completion is a feature in integrated development environments (IDEs) that suggests possible code completions as a programmer types. It helps to speed up the coding process and reduce errors. The video discusses the improvements in code completion latency and quality in the new JetBrains AI Assistant, comparing it with other tools like Copilot and Codeium.

💡Latency

Latency in the context of software refers to the delay before a system's response to a request. In the video, the author notes a definite improvement in the latency of the JetBrains AI Assistant's code completion feature, making it more usable compared to previous versions where it was slow and almost unusable.

💡Single-line and Multi-line Suggestion

These terms refer to the two modes of code completion provided by IntelliJ, a popular IDE from JetBrains. Single-line suggestions are shown first and are powered by a local model, reducing the need for network interaction and perceived latency. Multi-line suggestions are displayed after accepting a single-line suggestion and may involve more context. The video script discusses the differences in these approaches and their impact on the user experience.

💡GitHub Copilot

GitHub Copilot is an AI-powered code completion tool developed by GitHub and integrated with Visual Studio Code. It is mentioned in the video as a competitor to the JetBrains AI Assistant. The script compares the responsiveness and quality of suggestions between Copilot and the JetBrains tool, noting that Copilot is more responsive and provides better context capture.

💡Codeium

Codeium is another AI coding assistant tool that is mentioned in the video script. It is praised for its speed and quality of suggestions, which are said to be similar to those of GitHub Copilot. The author uses Codeium in their personal IDE due to its free individual use and integration with their coding workflow.

💡GPT-4o

GPT-4o refers to a model from the GPT (Generative Pre-trained Transformer) series developed by OpenAI. The video script mentions that the author subscribes to ChatGPT Plus for its capabilities, particularly the GPT-4o model, which is said to produce the best code suggestions and is incredibly fast.

💡ChatGPT Plus

ChatGPT Plus is a subscription service that offers enhanced capabilities over the free version of ChatGPT, including faster response times and priority access to new features. The video script mentions that the author is impressed by the capabilities of the GPT-4o model available through this service.

💡Enterprise Context

The term 'enterprise context' refers to the use of software tools within a business or organizational setting. The video script suggests that while the author would use GitHub Copilot in an enterprise setting due to its deep integration with GitHub, they prefer Codeium for personal use.

💡Inline Code Completion

Inline code completion is a feature that provides suggestions as the programmer types, without the need for additional prompts or commands. The video script discusses the author's preference for inline completion and how their opinion has evolved to appreciate both inline completion and prompted code generation for faster feature development and better code quality.

💡Prompting System

A prompting system in AI coding tools is a method where the tool generates code based on specific user prompts or requests. The video script describes the prompting system of the JetBrains AI Assistant as clunky and slow, with questionable quality, especially when it includes imports within the class body, which is an issue the author hoped would be resolved in the new version.

Highlights

JetBrains announced improved code completion in their AI Assistant on July 1st, 2024.

Subscribers can test the new AI Assistant to compare with Copilot.

To try the AI Assistant, one needs the 2024.2 early access program version 7 of IntelliJ.

JetBrains claims improved single and multi-line code completion for Java and Kotlin.

The AI Assistant is powered by JetBrains' own models for higher quality and reduced latency.

There is a noticeable improvement in latency compared to previous versions.

The AI Assistant is still less responsive than Copilot and Codeium.

IntelliJ performs code completion in two ways: single-line and multi-line.

Single-line suggestions are shown first to reduce perceived latency.

JetBrains AI suggestions are often very small code snippets compared to Copilot or Codeium.

The AI Assistant requires more context to provide suggestions, unlike Copilot and Codeium.

The prompting system in the AI Assistant is still clunky and slow.

The AI Assistant sometimes includes imports within the class body, which is an irritation.

The reviewer has canceled their JetBrains AI assistant subscription due to competition.

The reviewer prefers ChatGPT Plus for its capabilities and speed.

Both inline code completion and prompts can coexist for faster development and better code quality.

Copilot is preferred in an enterprise context due to GitHub integration.

Codeium is used for personal use due to its free individual use, speed, and similar quality to Copilot.

The reviewer invites comments on setups and encourages sharing the video for learning.

Casual Browsing

Is Cursor's Copilot++ BETTER than Github Copilot? FAST AI Coding Master Class

2024-08-20 18:35:00

Is FLUX better than Midjourney?

2024-08-14 04:17:00

JAN: This 100% LOCAL AI ASSISTANT is better than ChatGPT (works w/ RAG, Local Models, Groq & OpenAI)

2024-05-17 10:35:01

New Photoshop ai BETTER Than Generative Fill?

2024-05-17 22:30:01

Cursor: Build Software with The AI-first Code Editor with a CoPilot! Better Than VS Code!

2024-08-20 19:05:00