"It's not artificial intelligence that will replace programmers, but other programmers who will use artificial intelligence". Interview with Daniel Costea

Daniel Costea, Microsoft Most Valuable Professional (MVP), has been a senior developer for over two decades. 

speaker at more than 50 conferences in the US and Europe

mentor on >the interactive online Generative AI course aimed at all programmers, regardless of level, who want to make the most of artificial intelligence tools.

The 20 hours of course with Daniel cover several key areas: Generative AI and LLMs, basic & advanced prompt engineering, AI orchestrators and image generation. 

He tells us himself what the IT job market looks like right now, and why a course like this is needed.

1. Will artificial intelligence replace programmers in the near future?

It's not artificial intelligence that will replace programmers, but other programmers who will use artificial intelligence effectively.

Programming in the traditional senses (structured, functional, object-oriented) is a set of conventions, algorithms and paradigms through which we solve problems with the help of machines. 

Traditional programming, or more precisely programming languages, have syntax as their fundamental element, but this is about to change because of artificial intelligence, where semantics plays a central role, more important than syntax.

Here are a few reasons why  replacing programmers with artificial intelligence won't happen,  however, in the near future:

• • it can't think outside the box, although steps are being taken in this direction with Autogen, AI assistants, etc.;

• • it has limited context. Context is the piece of information that serves as input to guide the model (LLM) toward an answer;

• • it has limited computational power. Today's graphics processors (GPUs) need many seconds, or maybe minutes, to prepare a complex answer, and most of the time we need several answers at the same time to make a decision;

• • it has limited energy available. LLM models are very costly compared to traditional applications;

• • the high price of graphics cards with GPUs, or of services offering generative AI. These limitations will probably disappear once we have quantum computers and cheap, virtually unlimited energy;

• • it requires human oversight for validation and guidance toward solving problems (human in the loop).

2. If ChatGPT can generate code, does it still make sense to learn a programming language from scratch?

Let's simplify a bit. GPT (or LLM) models know a lot of languages because they were trained on source code written in those languages. As a beginner, you might not know any language at all. How could ChatGPT help you if you don't know at least one language, or if you don't know how to approach solving a problem in a programming language?

I think what would help you most is learning a multi-paradigm programming language (procedural, object-oriented, functional) such as the C# programming course, the online Python course, Java, online C/C++ courses.

The newest courses at Informal School of IT:

Agentic AI with .NET artificial intelligence course

Power BI for Analysts course: Modeling, DAX & Dashboards

Many companies are now looking for AI knowledge in software developers' profiles

3. What's new about this Generative AI course for programmers, and why is it an opportunity?

Understanding and using AI in software development can open new career opportunities. Many companies are now looking for software developers with AI knowledge. Generative AI can solve new scenarios that used to be difficult or impossible to solve. 

For example, it can be used to generate programming code from natural-language prompts, to complete partially written code with suggestions, or even to translate code from one programming language to another. Using AI in software development can improve efficiency by automating certain tasks, saving time and resources, or by integrating bots capable of improving the user experience.

Unlike other types of AI, such as machine learning or deep learning, with Generative AI you can get a fast start, with fewer mandatory disciplines to know.

If you liked grammar, you have an advantage, because, as I said earlier, semantics is the central element in programming generative AI. As in any human conversation, in interacting with AI it's essential to be coherent, concise, and precise.

4. So graduating from a course like this will matter to employers.

If you have knowledge of generative AI, it will be just as ordinary a question as some people still remember being asked: Do you know how to use a computer? Do you know Office?

It's like the leap accountants made in the past, moving from paper to software

5. How do the AI course sessions run? What teaching methods are used?

The course sessions explain the basic concepts in language accessible to everyone, and they're very practical, combined with many examples from personal experience. Whether you're a programmer or not, you have no choice, generative AI is a skill that will make the difference between you and your colleagues. It's a qualitative contribution you have to bring if you want to keep up with technology. It's like the leap accountants made in the past, moving from paper to software.

6. Daniel, you've been a senior developer for over two decades, a speaker at dozens of conferences in the US and Europe. What message do you have for those toying with the idea of a career change into IT, but who no longer have the courage?

At Informal School of IT you get the chance to have a mentor

• I want to share something. I don't program at all in the languages I learned in school, and I was self-taught in everything I know how to do, but I would have shortened this path enormously if I'd had someone to ask about concepts, principles, best practices, the minute I needed them. At Informal School of IT you get the chance to have a mentor.