Zed and AI will save us millions

Zed and AI will save us millions

... and you might soon lose your job

The software engineering world has changed a lot, but it seems like both workers and companies haven't fully caught up yet. Recently, I've been having a lot of fun using Zed. It made programming enjoyable for me again, just like it was many years ago. Some people think Zed is just another unfinished editor, but that's not right. Zed is an AI tool. If you're not using Zed with GitHub Copilot and OpenAI GPT, you're not using it correctly, and you likely don't need Zed at all.

Working with AI now feels like coding alongside a really cool, productive, and smart partner who's always there. For example, I recently wanted to add Huffman coding to my toy project. The last time I worked with Huffman encoding was about 30 years ago, so without AI, I would've needed to spend some time looking things up on Wikipedia and other places. With Zed and GPT, I added Huffman encoding to my project literally in 5 minutes. You can see the result here on GitHub.

What does this mean? It changes everything. Imagine I'm a manager at a tech company, and I decide to add Huffman encoding to see the effects of compression in my library. Normally, it would take several meetings just to decide if we really need it and if we have the resources for it. Then, someone might be assigned to look into it, which also involves planning resources. This person would likely spend days evaluating different libraries. A week later, I might expect something to be done, maybe not. With AI, I had it working in my project literally in just 5 minutes. This is a huge deal.

But the impact is even bigger. AI is now better than 95% of software engineers in the world at many tasks. Take the Huffman example and try to compete with ChatGPT on implementing algorithms like that. This would be hard for many engineers and take much longer. If you disagree, just ask next candidate to write Huffman coding on coding interview and enjoy her suffering. Sure, GPT's output may require some tweaks, but overall, I'm achieving results 10 to 100 times faster than when working with humans, and often with superior quality.

Current hiring processes are outdated. If you're still making engineers do tasks like reversing linked lists or solve other Leetcode problems in interviews, you're missing the point. GPT-4 can do these tasks better. If you're asking about "system design" in interviews, you're even more behind. GPT-4 will probably give better answers than almost any engineer you could hire. This is a huge shift in the industry, and many people will start realizing this soon.

I do not want to say you should rely on ChatGPT's opinion on design of your system. Moreover you should not, and this is why real engineers are still needed and will be in more demand than before. But the point is that the hiring process is outdated and needs to be changed. You should not ask engineers to solve Leetcode problems or reverse linked lists in interviews. Why would you hire a mediocre coder when GPT-4 can do it in seconds for $50 a month?

I'm really happy with these changes. It's a big boost for the industry, but it's also going to change things a lot. Likely, 95% of software engineers (as most people think of the job) will be out of work soon. They're just not needed anymore, and they're falling short in the tasks they're currently doing. Industry should rethink software engineer role and focus them on what's really important. It's not about writing code anymore.

Think about how AI is changing software engineering like the tractor changed farming. Sure, we still have farmers, but they don't need to do the hard work in the fields themselves anymore. Their job has changed to managing the farm, making important decisions, and planning for what's coming. They don't need to rely on physical strength -- they need to be smart and know how to use the machines.

This is pretty much what's happening with software engineers now. They don't have to spend all day writing code. Their role is shifting towards project management, deciding what needs to be done, and preparing for the future. If your engineers are still spending days on something AI can do in a few minutes, it's like being that farmer who still uses a horse to plow the field. The world has moved on, and so should the role of the software engineer.

If you want to be productive, not only with coding tasks but also with everything related to teamwork, including collaboration and project management, you should try Huly. Huly aims to be the Zed of project management and team collaboration.

We're not quite there yet, and Huly's current process management capabilities are somewhat similar to what Linear, Asana, Jira, and others offer. However, we're passionately working on it.

Stay tuned, and you'll see how Huly will positively change the lives of many teams.