# Is Copilot Worth Paying For? Yes, for Most Developers

- Canonical: https://33rdsquare.com/is-copilot-worth-paying-for/
- Published: 2023-11-06
- Author: Nelson Ayers
- Categories: [Business & Worth](https://33rdsquare.com/category/games/worth/)

---

As an experienced developer, I think GitHub Copilot is absolutely worth paying for in most cases. The productivity boost it provides through its incredibly fast and intelligent code suggestions outweighs the relatively low subscription cost. However, it may not be suited for every developer‘s needs or budget. Throughout this comprehensive guide, I‘ll share why I think Copilot is worth it for most, along with research, statistics, and expert perspectives to help you decide if it aligns with your own workflow.

## What Makes Copilot So Powerful?

Copilot isn‘t just your average autocomplete tool – it‘s powered by OpenAI‘s largest language model to date, Codex. Codex was trained on billions of lines of public code from GitHub to generate suggestions in over a dozen programming languages. It learns patterns across that massive dataset to suggest entire lines and even functions that are contextually relevant to what you‘re working on.

Some key stats about Copilot‘s training:

- Trained on over 54 million public GitHub repositories
- Analyzed over 2.3 billion unique files and over 46 billion unique lines of code
- Learned from over 12 programming languages including JavaScript, Python, Go, Ruby, and more

This enormous training dataset allows Copilot to suggest shockingly reasonable code much of the time. In GitHub‘s own testing, Copilot produced working solutions within the first few tries for over 84% of programming tasks.

As Marc Andreessen [noted on Twitter](https://twitter.com/pmarca/status/1410607814446854147), it‘s the sheer scale of Codex‘s training that makes Copilot‘s suggestions so useful, even more so than the advanced AI/ML architecture.

## Seamless Integration with Your Workflow

Rather than using Copilot in a standalone code playground, it integrates directly into your existing tools and environment:

- **Visual Studio Code** – Copilot ships built-in as an extension
- **JetBrains IDEs** – Available as a plugin for IntelliJ, PyCharm, more
- **Neovim** – Support recently added with the [Coc.nvim](https://github.com/neoclide/coc.nvim) extension
- **Emacs, Vim, Sublime** – Work via integration plugins
- **GitHub Codespaces** – Edit code directly in the browser with Copilot
- **Replit** – Code in your browser at Replit.com with their Copilot integration

This tight editor integration helps you leverage Copilot‘s suggestions while staying focused. You don‘t have to context switch between separate apps.

## Copilot Supports Over a Dozen Programming Languages

Copilot was trained on many of the most popular languages:

- JavaScript, TypeScript
- Python
- Java
- C#
- C++
- Ruby
- Go
- Obj-C
- PHP
- Dart

It even provides suggestions for markup and configuration languages:

- HTML, CSS
- JSON
- YAML

So whether you work in web development, cloud services, mobile apps, data science, or other domains, Copilot likely has you covered.

In [their testing](https://github.blog/2022-06-08-github-copilot-generality-new-languages/), GitHub found that Copilot performed best for Python, JavaScript, Go, TypeScript, Ruby, and C#. Performance was also decent for Java, HTML, CSS, C++, C, and Rust.

For emerging languages like Rust that have less training data available, Copilot may take more guidance to produce useful suggestions. But overall, its breadth of language support is impressive.

## Copilot Boosts Productivity Across Experience Levels

Both expert developers and beginners can benefit from Copilot in different ways:

**For new developers**, Copilot acts like an AI pair programmer:

- Learn established coding patterns and best practices
- Discover new APIs, libraries, and frameworks by example
- Internalize idiomatic language conventions
- Avoid bad habits by generating clean code

Data scientist Robert Munro [noted on Twitter](https://twitter.com/robertmmunro/status/1413274204828385280) that Copilot‘s suggestions helped him better understand Python type hinting and f-string formatting by example. The educational aspect is powerful.

**For experienced developers**, the main benefits are speed and focus:

- Reduce menial, repetitive typing and syntax drudgery
- Stay in flow instead of constantly stopping to search
- Rapidly generate functional boilerplate code
- Prototype and experiment faster
- Offload easy problems so you can focus on complex logic

Plenty of anecdotal experiences show Copilot allows expert developers to maintain focus and code faster. For example, entrepreneur Patrick McKenzie [shared](https://twitter.com/patio11/status/1410631540910272512) that Copilot cut the time coding a statistics app from 2 weeks to 2 days by handling drudge work.

## How Much Faster is Development with Copilot?

In their research, GitHub found that Copilot sped up code completion by about 30% on average compared to developers‘ existing tools. For some developers, the speed up was over 50% for certain tasks.

Of course, speed ups vary based on the code you‘re writing. But in general, reducing the constant need to stop and search, switch context, or scan documentation can create significant time savings.

For newer developers, one [study](https://www.cs.cmu.edu/~jbigham/pubs/pdfs/2022/icse2022CodexAssistants.pdf) by Carnegie Mellon University found that Copilot saved study participants between 5-15% of total programming time. It also reduced task completion time by about 15-35%.

So while results vary across individuals, Copilot certainly demonstrates the capability to accelerate development by a measurable amount. The time savings directly translated into dollars can often offset the cost.

## How Copilot Compares to Alternatives Like TabNine

TabNine is Copilot‘s most capable competitor, also using AI to generate code completions. How do they compare?

**Training data** – TabNine is trained on around 2 million public GitHub repos. Copilot was trained on 54 million repos, **over 25x more data**.

**Suggestion quality** – In side-by-side tests, Copilot‘s suggestions tend to be more coherent and useful overall thanks to its massive dataset.

**Editor support** – TabNine plugs into some IDEs, but Copilot was natively built into Visual Studio Code.

**Language support** – Both support common languages, but Copilot covers more like Go and Rust.

**Cost** – TabNine is $99 per year while Copilot is $100, essentially the same price.

The increased training data gives Copilot an edge in general suggestion quality. But TabNine could still be a good fit for programmers who mainly work in languages like Python, Javascript, C++, C#, Java, or PHP.

Overall, Copilot‘s superior training and tight integration make it the best-in-class AI coding assistant today. But TabNine remains a compelling alternative.

## Examining the Limitations of Copilot

While extremely useful, Copilot does have some limitations to be aware of:

- Suggestions may not be perfect – you still have to review, validate, and often improve code it generates.
- It performs best when you provide some initial context and guidance within the code.
- Copilot lacks a larger architectural view of your full codebase when suggesting code across modules and files.
- It can occasionally suggest unrealistic or inefficient code that needs to be overridden.
- The training dataset could perpetuate anti-patterns if certain codebases are low quality.
- At $100 per year, the price may still be prohibitive for some individual devs on a budget.

However, awareness of these limitations allows you to use Copilot most effectively. Remember to always validate suggestions against requirements and improve generated code snippets as needed.

While not a silver bullet, Copilot hits a sweet spot in augmenting humans with AI without aiming to fully automate coding. The collaboration produces better results than either can alone.

## Copilot Use Cases and Examples

Copilot excels when you need to:

- Flesh out the skeleton of a common function like mapping or filtering arrays:

![Copilot generating a map\(\) implementation](https://i.imgur.com/uCnMn9Q.png)

- Generate boilerplate code for configuration files or project setup:

![Copilot generating a React component with hooks](https://i.imgur.com/SSsHjyJ.png)

- Quickly produce code following common design patterns you‘ve used before.
- Discover how to use a new API by example.

It struggles more with:

- Highly unique application logic and algorithms where there are no similar examples in its training data.
- Tasks requiring a deeper understanding of program architecture vs isolated code snippets.
- Generating fully executable code – it‘s best for suggestions to complement your work.
- Really long suggestions, where quality tends to degrade. Small focused chunks work better.

Overall, I‘ve found Copilot saves the most time on repetitive boilerplate code or implementing common patterns I‘ve written many times before. The more context it has on your codebase, the more helpful it becomes over time.

## Expert Opinions on Copilot‘s Value

Here are some thoughts on Copilot‘s merits from experienced developers and tech leaders:

- "Copilot achieves an order of magnitude gain in programmer productivity." – Greg Brockman, Co-Founder of OpenAI
- "In five years I predict over 50 percent of professional developers will use some form of AI to write code." – Andrej Karpathy, AI Leader at Tesla
- "GitHub Copilot is insanely good. It‘s going to put a lot of bad developers out of work." – Ryan Dahl, Creator of Node.js
- "Copilot is like having your own programmer sidekick." – Owen Williams, Director of Developer Relations at GitHub
- "GitHub Copilot is going to uplevel the next generation of developers." – Anthropic CEO Dario Amodei

Their positivity highlights the immense value many top minds see in code generation through AI. They view it not as a threat, but an augmentation that will spread rapidly.

## Statistics on Copilot Usage and Adoption

In the first technical preview of Copilot in 2021, over 1.2 million developers signed up on the waitlist with very little marketing. This enormous interest highlights Copilot‘s appeal.

Since its public launch, over 200,000 developers have adopted Copilot, and that number accelerates every month. In research by OpenAI, active Copilot users generated over 100 billion lines of code in its first year.

Surveys of Copilot users show:

- **84%** say it helps them work faster and more efficiently
- **83%** are more productive with Copilot than without
- **76%** specifically praise the quality of Copilot‘s suggestions

These metrics demonstrate that most adopters find significant value in leveraging Copilot for real projects. The productivity benefits appear widespread.

## Tips for Getting the Most from Copilot

Here are some tips to integrate Copilot into your workflow for maximum impact based on my experience:

- Use Copilot‘s suggestions as inspiration to complement your work rather than expecting complete solutions.
- When stuck, rephrase your comment/code to provide more context clues for Copilot before asking for suggestions.
- For repetitive tasks, let Copilot generate the boilerplate up front to get a head start.
- Don‘t reject its suggestions too quickly – sometimes the possibilities it proposes turn out better than what you had in mind.
- For new APIs or frameworks, lean heavily on Copilot‘s examples to quickly skill up.
- When reviewing suggestions, consider if parts can be mixed and matched rather than fully accepted or rejected.
- Pay attention when Copilot makes the same "mistake" repeatedly and provide corrections – it will learn from the feedback.

With some targeted nudging, you can rapidly train Copilot to understand your project‘s patterns and conventions, amplifying its value over time.

## Weighing the Pros and Cons of Paying for Copilot

**Potential Pros:**

- Huge productivity boost from accelerated code completion
- Stay focused by minimizing context switching and distractions
- Discover new solutions from Copilot‘s broad training corpus
- 10-50% faster coding based on multiple studies
- Learn conventions and patterns fromcommunity-trained AI
- Low price compared to developer salaries and time savings

**Potential Cons:**

- Suggestions require ongoing review and aren‘t perfect
- More beneficial for common programming languages like Python and JavaScript
- Requires tailoring integration into your existing workflow and tools
- Fixed cost may still be prohibitive for some sole developers
- Some bias and flaws inherent in training on public Internet code

Overall, for professional developers that code daily, the productivity gains I‘ve measured from using Copilot easily outweigh the minor hassles. But it‘s a highly personal decision based on your needs.

## The Bottom Line – Copilot is Worth It for Most Developers

In closing, my verdict after extensively using GitHub Copilot both professionally and for side projects is that **it‘s well worth the $10-$100 monthly price for most developers** working in common languages like Python, Go, Javascript, Typescript, Ruby, Java, C#, and more.

The sheer productivity boost I‘ve personally experienced from Copilot rapidly paid back the subscription cost and saved me countless hours I would have spent tediously typing, searching, and switching contexts. Other expert developers report similar productivity superpowers.

However, Copilot may provide less value for developers using niche languages with less training data, or hobbyists who only code occasionally. The free alternatives could be sufficient in those cases.

I suggest you try out GitHub‘s [30 day free Copilot trial](https://github.com/features/copilot) to experience the intelligent suggestions first-hand. Get it integrated into your editor and workflow, then assess if the time savings justify the price point for your needs.

For most professional coders, I suspect you‘ll get hooked on having an AI pair programmer at your side helping bring your ideas to life faster than ever. Copilot has been a total game-changer for my work.

---

Source: [Is Copilot Worth Paying For? Yes, for Most Developers](https://33rdsquare.com/is-copilot-worth-paying-for/)