Two-thirds of enterprise leaders won’t hire you without AI skills. That’s not clickbait – that’s from a recent ThoughtWorks article. And if you’re still calling AI coding assistants ‘too risky,’ you might already be in the unacceptable column at companies like Zapier. I was skeptical and had mixed feelings about LLMs. But now I believe we passed the tipping point, and AI assistants will fundamentally change software development.
Two-thirds of enterprise leaders won’t hire you without AI skills. That’s not clickbait – that’s from a recent Thought Works article. And if you’re still calling AI coding assistants ‘too risky,’ you might already be in the unacceptable column at companies like Zapier.
I’ve been watching the rise of GenAI with skepticism and mixed feelings. I see both its potential benefits and its issues, and for a while it was unclear which way it will go. I was torn between two equally possible options. Option A: it’s a bubble that will burst. Or Option B: GenAI will completely change the face of knowledge work.
I am now certain that GenAI is here to stay and it will revolutionize how we work. I think we’re beyond the tipping point, and that the next months and years will be about adjusting to this new reality.
Before we discuss further, I need to specify a few things. I don’t buy into the hype that the AI companies are so eager to promote so that they can get more investment money so that they can continue existing. I don’t believe AI will replace people, and I don’t believe AGI (Artificial General Intelligence) is close by. I don’t believe that the way LLMs were trained was ethical – by taking everything they could from the internet without paying creators anything. I don’t believe that the energy required to train the GenAI models we have is necessarily a good investment for the whole of the society. I know LLMs have unsolved issues related to copyright, security, and privacy. I also don’t like that we are giving so much power to a hand full of corporations.
I also believe that none of this will matter in the medium term, no matter what you or I think about it.
So here’s my prediction: GenAI is here to stay. If you work in software development, it’s likely you will have to learn how to work with coding assistants, how to enhance them, and how to use them from code.
Here are some arguments in favor of my prediction.
First, a recent article by Thought Works states “2/3rd of enterprise leaders now say they wouldn’t hire a candidate without AI skills”. On the same line, Wade Foster, the CEO of Zapier talks in a recent X post about how they started measuring AI fluency. On the engineering role, in the unacceptable column is the attitude defined by: “Calls AI Coding assistants too risky”, “Has never tested AI-generated code”, “Relies only on StackOverflow snippets”. The trend is clear: the future belongs to software engineers who use AI assistants.
Second, AI assistants have increased their capabilities and provide real value to large organizations. Morgan Stanley has announced that their AI tool called DevGen.AI has processed nine million lines of legacy code this year, saving the investment bank’s developers an estimated 280000 hours by translating outdated programming languages into plain English specifications that can be rewritten. Our industry has long struggled with legacy code, and AI assistants at least enable the modernization process, if not speed it up considerably. In the meantime, marketing teams have used AI assistants to create internal dashboards with minimal input from developers. Prototypes for product discovery are easier to build than ever due to the same assistants.
Third, coding assistants enable developers to solve problems we never had the time to tackle. I had the need to develop some simple custom WordPress Guttenberg blocks for years, but I never had the time or energy to read the whole manual. It also wasn’t a good investment since I don’t plan to make this a habit. But with Claude, I just had to ask for a starting code that did kind of what I needed. It was average code, as a junior programmer would have written. But once I had it, it was easy for me to simplify, test, refactor, and finally solve the problem.
I am not alone. Every team I know has low priority improvements that they want to make but are never prioritized. How about using the assistants to tackle them? Provide the prompt, run them a number of times, pick the best solution and that’s done. How many backlog items from the world of software development can be solved in this way?
Fourth, organizations are starting to look into the AI transformation. Moderna Tech merged its Tech and HR departments because they realized HR is better placed at figuring out a mixed workplace of AI and people. I wouldn’t call this an industry trend yet, but it’s a signal of what’s to come.
By now you should realize that we are facing an event similar with the industrial revolution, in that it will fundamentally change the way we work. Some of us will have to re-skill. Jobs will be lost, as other roles are created. We are still in the early stages of this transformation, and it’s not uniformly spread. But chances are, once it starts it will be very fast, much faster than anything our industry has seen before.
So here’s my question for you: are you going to be the developer who adapts, or the one who gets left behind? Because ignoring AI assistants won’t make them go away.
The only thing we can do now is to get prepared. Learn about LLMs, work with coding assistants, think about their role in software development.
I’ve already started, and I plan to share with you what I learn. I plan to record videos of using AI assistants, without the hype, without promoting them, but checking how they help and if they help in various scenarios. If you want to go on this journey together, subscribe to this channel and let me know in the comments what you’d like to see in practice.
