In this article, we sat down with two CTOs to discuss this revolution, the changes that we can expect to see emerge, how to prepare for them, and how to stay ahead as a software engineer.

What we'll cover in this article

• Generative AIs are definitely impressive, but they are far from being completely reliable. It’s very likely that they’ll always require human supervision.
• Far from replacing software engineers, AI will tremendously improve our productivity and help us write more resilient and secure code.
• As for previous technology advents, we can expect to see new jobs and opportunities emerge and engineers will need to adapt and build new skills.
• Finally, we’ll go over advice and resources so you can prepare yourself for the changes ahead.

• Paul Vidal, former software engineer at Datadog and now CTO of Collective, a platform that helps freelancers team up to bring even more value to their clients.
• Sébastien Lefort, CTO of Primaa, a company that develops AI-based software for histological diagnoses to improve the detection of cancer.

Let's get into the heart of the matter!

1. AI vs. engineers: is AI really after your job?

It’s hard not to wonder after seeing the launch demo video of ChatGPT-4 by Greg Brockman, OpenAI’s President & Co-Founder. In that video he demonstrates how the generative AI can create a simple website from a drawing he made on a piece of paper.

One can understand how such technology can frighten graduating engineers who might think that, by the time they enter the market, they will be less proficient than those tools.

For Paul Vidal, this feels like déjà-vu: "Each generation has its technological revolution. Before AI it was the Internet and Google, before that it was television, and the telephone and electricity before it. We believe it's new but it's really not." Each of those revolutions came with its changes and required humans to adapt. Google and other search engines changed how we search for information and how we learn, moving from libraries to the Web. Generative AIs are the next technological step.

Sebastien Lefort definitely shares this view. "There’s no doubt those tools will have an impact but not in the magnitude that is advertised today" he says.

Indeed, both agree that although the technology is impressive, it’s not going to replace engineers, mainly for two reasons:

Reason #1: AI needs human supervision

Generative AIs are notably unreliable. OpenAI, the creator of ChatGPT, is well aware of those potential errors, as they mention in the limitations of ChatGPT:

“ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers.”

This lack of reliability has actually led StackOverflow, the famous developers' forum, to ban the use of ChatGPT generated text from its platform “because the average rate of getting correct answers from ChatGPT is too low”.

“The posting of answers created by ChatGPT is substantially harmful to the site and to users who are asking and looking for correct answers.”
Stack Overflow official statement

And as OpenAI states on their site those limitations have no simple fix. One of the challenges is the lack of commonly accepted sources of truth for the machine to learn from. Which we may never have.

As a result, it’s very likely that generative AIs will always need some level of human supervision to differentiate a wrong answer from a correct one.

“As the technology improves, we can expect some of the rough edges to be sanded down, but fundamentally, there will always be some level of inaccuracy. These tools just don't have any mechanism to objectively validate their responses.”
- Josh Comeau, full stack engineer and educator, ex-Gatsby and DigitalOcean, in The End of Front-End Development

‍Reason #2: AI lacks subjectivity

One point that’s raising questions is the idea that the AI could learn from itself, auto-correct and iteratively become autonomous and perform any job as a human would. This is forgetting a very important thing: machines are not, and will never be, humans.

What makes a product great, and successful on the market, is how well it solves problems that we, humans, are facing in our lives. To produce that value you need to be able to experience life, what it means to be human, what makes us happy, angry and everything in between. Actually, all of this requires subjectivity, whereas AI tends to be as objective as possible.

“Experience” does not come from reading a piece of text to infer answers from it. It comes from experiencing life, living in a society with its own culture, inherited from thousands of years of human evolution, having friends, a family, a job, growing old. No AI can grow such an experience.

“AI is not about the total automation of content production from start to finish: it is about augmentation to give professionals and creatives the tools to work faster, freeing them up to spend more time on what humans do best.”
- Charlie Beckett, London School of Economics, in the Guardian

Both our CTOs share that point of view. What’s expected from engineers is less their proficiency to write code than their ability to be creative in face of the unknown, which they both consider to be a human-only quality.

“The job of an engineer remains unchanged: creativity, pragmatism, and the ability to solve problems.”
- Sébastien Lefort

2. Enter the new era of AI-driven development

Far from replacing software engineers, AI will take developer experience to the next level. And on that front, Paul Vidal is excited about the productivity gains that those tools can offer.

“For sure, developers will now code with AI. ChatGPT helps me solve issues in a matter of minutes that could otherwise have taken me hours. It’s obvious that this will impact developers' performance. There will be the developers who know how to use AI and those who don’t.”
- Paul Vidal

GitHub, the most widely used repository system and development platform, announced GitHub Copilot X, the new version of GitHub Copilot which was already claiming to make developers code 55% faster. Autocompleting code during development was just a first step, GitHub Copilot X also includes:

• a ChatGPT-like experience integrated into VS Code, Microsoft’s IDE, which can provide in-depth code analysis, write unit tests, and propose bug fixes;
• AI assistance for pull requests;
• AI-generated answers on documentation for languages and libraries a project relies on.

And this is just the beginning. Companies have entered a race to AI. More and more AI-powered tools come out every week and especially to make developers' life easier. Let’s take a look at the main changes we can expect to see in our daily lives as programmers.

3 areas to watch for changes

1️⃣Developer tools

Tooling will evolve, for the biggest pleasure of developers. Erica Brescia, former COO of GitHub, stated during an online Sourcegraph conference earlier in March that especially areas like testing and code reviews will become mostly automated and AI-generated.

For Paul, AI is the next stage of information search and tooling. “Every engineer today code within an IDE in which there is some level of assistance. Whether it’s formatting with Prettier, linting with ESLint or typing with TypeScript. Tools like GitHub Copilot are just the next level of IDE.”

For Sébastien Lefort, the DevOps field can also benefit greatly from AI for tasks like bootstrapping an infrastructure on AWS with a very specific configuration. Saving us time and money in the process.

2️⃣ Search

The way we search for information and the tools we use for search are bound to evolve. So much so that Google declared a Code Red after ChatGPT’s release end of last year. Microsoft seized the opportunity to make a multibillion-dollar multiyear investment in OpenAI and has announced new versions of both its search engine Bing and web browser Edge powered by OpenAI.

“AI will fundamentally change every software category, starting with the largest category of all – search.”
- Satya Nadella, Chairman and CEO, Microsoft.

Developers who have been heavily relying on Google and StackOverflow until now, have already added ChatGPT to their toolbox.

3️⃣ Hiring

Another area where we can expect to see some changes is in how companies hire engineers and the skills they might expect for certain roles.

Companies looking to heavily rely on AI to power their features might start to look for experience using such APIs in engineers’ resumes. And when it comes to evaluating programming skills, there might be room for companies to rely less on at-home exercises and more on live-coding.

“Small at-home exercises will probably no longer be the right way to assess engineers' skills, if they ever were”
- Paul Vidal

3. New opportunities for software engineers

The US Bureau Of Labor Statistics projects a massive 25% increase in software engineering jobs for the decade of 2021-2031. For reference, the average growth rate (reported by the same source) for all occupations is 5%. The demand is so high that dozens of bootcamps and high-paced training programs have seen the light of day in the past decade to train more and more engineers to meet companies’ needs.

New jobs will emerge

Two thirds of the job we know today didn’t exist in 1940. And engineering jobs make no exception. We’ve had to adapt, learn new technologies and change how we work.

The increasing amount of data along with the rise of artificial intelligence led to new careers as data analysts, data engineers and data scientists. And with the rising amount of developers came the advent of the developer tools industry that led to more new jobs like developer advocate, developer experience and developer relations engineers.

Even if today the mention of Prompt Engineering is mostly making engineers smile, it makes no doubt that new careers and new opportunities will arise in this new AI-driven era.

“There’s never been a technological revolution that’s destroyed more jobs than it has created.” - Tsedal Neeley, author and professor at the Harvard Business School on CNBC

Existing industries will continue to grow

Along with new career opportunities, existing fields, which already grow in demand year after year, will still be looking for engineers.

Starting with artificial intelligence of course which has already been a “trending software engineer job” for years. And the new wave of generative AI is only going to make it more in-demand as all AI companies will be racing to provide the most reliable “AI as a service”.

The cloud industry should continue to grow as well to cover the need for infrastructure which will necessarily increase. Which means more cloud engineers, DevOps engineers, infrastructure engineers, database engineers, and so on.

4. How to prepare to be a future-proof engineer

With all those AI-powered tools to help developers deliver software more efficiently, we’re entering a new era of AI-driven development. And, according to Paul, it’s a bandwagon developers can’t afford to miss.

🤔 Be curious

For engineers, staying up-to-date is in the job description. For Paul: “Technology evolves at a crazy pace. It’s inherent to the field. What makes a good engineer is their ability to stay up to date. You have to keep yourself informed, trained and be proactive. One of the major qualities of an engineer is curiosity. Curiosity leads you to look into new technologies when they come out and evaluate their benefits. Without curiosity you risk falling behind.”

“The two fundamental qualities of the engineer of tomorrow will stay the same as those of the engineer of today: curiosity and pragmatism.”
- Paul Vidal

🛠 Start using AI tools

Just like you’re used to seeking answers on Google, Stack Overflow and documentation, you should start getting acquainted with using AI tools.

There’s ChatGPT and Github Copilot of course but also Tabnine and Codesnippets which bring code completion and more into your IDE, Mintlify which helps document your code with AI-generated comments, or AutoRegex so you never have to write a regular expression yourself ever again.

📝 Learn to prompt like a pro

A good place to start is by learning how to get the best value from those AI tools by learning how to prompt them effectively. OpenAI and DeepLearning.AI released a free course on prompt engineering for developers. In this short one-hour course you’ll learn prompt engineering best practices for application development, the new ways to use LLMs and practice writing and iterating on prompts using the OpenAI API.

👩‍💻 Learn the fundamentals of generative AI APIs

Using generative AI APIs comes with several challenges. Supabase has implemented OpenAI’s API in its docs and published their methodology in a blog post and video which are perfect to discover how OpenAI APIs are used in a real-world use case.

From there, you can move on to learn how to use those APIs yourself. Here are three courses to get you started:

• freeCodeCamp’s 5-hour free course where you’ll learn to use text completion models, create images with the DALL-E model, and more.
• A complete course on OpenAI Python APIs by Colt Steele, one of the most popular programming educators on Udemy.
• Scrimba’s free course to build AI apps with ChatGPT, Dall-E and GPT-4.

Finally, this article gathers a large number of resources from books to courses to social accounts to follow on this topic.