AI is changing how software gets built. Explore the skills you need to keep up and stand out.

The GitHub Blog provides updates, announcements, and insights from the world's leading software development platform, covering topics such as new features, community highlights, and industry trends. Developers can learn about GitHub's latest developments, best practices for collaboration, and tips for maximizing productivity on the platform.

GitHub Blog

The developer role is shifting from pure coding to orchestration, strategy, and collaboration as AI tools handle more implementation work. Success requires mastering three key skills: providing better context to AI coding assistants like GitHub Copilot Spaces, leveraging AI-powered code review for team collaboration, and committing to continuous learning in AI fundamentals and frameworks. While technical skills remain important, the ability to direct AI systems, blend machine efficiency with human judgment, and adapt through ongoing education will differentiate developers in an evolving job market.

The developer role is evolving. Here’s how to stay ahead.

Skill #1: Make AI coding more reliable by providing better context

Skill #2: Provide insight, judgment, and strategy

Skill #3: Recognize that learning is never done

Oof.
<blockquote>
In fact, AI is on track to write 95% of code within the next 5 years.
</blockquote>
If we keep at it with the predictions, one of them will eventually be right. Last I checked this was supposed to happen in 6 months 6 months ago.
The rest of the article is just an ad. But kudos for consistently including a leading space in all inline links.

<pre><code class="language-text">The opportunity: Robots are terrible at creativity, collaboration, and big-picture thinking. Which, conveniently, is where the future of development is headed.

</code></pre>
People aren’t always so great either at those. Spending hours trying to solve a problem versus asking a peer. Brute forcing a O(n^2) solution versus a conversation on a more elegant O(log n) solution. If AI spends more time generating code, developers can increasing interact more, and social skills become more valuable.

I’ve recently been thinking about this huge schism between developers and these tools. One side says “It doesn’t help much, just a better autocomplete” and the other side says “I literally shipped something in 2 hours that would take 2 months”. I realized that it comes down how you slot and thread it into your workflow.
“They” say that learning to code with an LLM assistant is sort of a new skill…but then I thought…is it? The skill we’re really talking about is: delegation.
Most developers are absolutely terrible at delegating. That is a skill largely held by technical PMs and senior developers. Now, any dev at any level needs to be able to delegate to properly use these tools. This would explain why it works well for juniors, because they just delegate the vast majority of it. And senior devs love it too, because they’re used to delegating. And then there’s this massive swath of developers who are great devs, but are used to just writing code on their own and not asking others for help or assistance much, nevertheless learning to how to delegate tasks or chunks of a project, so they just lean on the autocomplete/interactive documentation features of these tools. When they do try and use the delegation aspects, its awkward and they don’t often structure the requests in a way that would yield effective results, so they dismiss the output as “slop”, but the reality is that these tools can generate code as good as any developer given the proper context, examples and support…basically the way you would if you delegated the task to someone else and would expect good results.
(Although, LLMs require a lot more handholding because they really are just machine learning algorithms with no opinion or experience to bring to the problem).
If a developer wants to gain the most from these tools, they need to learn the art of effective delegation.