The hope and the fear of AI

There is both panic and excitement around the AI in the tech world. Some of them, well founded. Others, exaggerated. In one side, we see celebrations how months-long efforts became an hour-long task. How a complex undertaking that would take a proper training on the topic (for example, write a modular, distributed algorithm in Erlang programming language) now can be done in the matter of minutes in ChatGPT or a similar LLM. Then, the other side, panic that AI is going to wipe off the junior software engineer roles. It is probably widening the gap. In one side, engineers paid in millions to write the tools. In the side, the fresh graduates reducing themselves into "prompt engineers." Do we even need a 4-year bachelors on computer science if all we do is becoming a prompt engineer?

When we grade the take-home assignments and open-Internet exams, it is very clear that most of the answers are ChatGPT-inspired. We do not ban the use of LLMs in assignments. We just ask the students to disclose their source and tool, and almost all of them honestly report the ChatGPT use and it is ok. We are a StackOverflow generation. When we got stuck with an error message, we would ask StackOverflow and someone from 10 years ago had the same problem and there is the solution! Then, we would copy-paste. When we did not know the answer to a question, we would ask Google and Google gave us the answers. So, Internet copy-pasta is not new. Perhaps, ChatGPT has made it easier. It still needed some serious effort and "expertise" to find the answer from Google and StackOverflow. Maybe ChatGPT has made the entire process easier by giving you the answer, entire answer, rather than bits and pieces of it. StackOverflow was a crowdsourcing human effort. ChatGPT is using that human effort second-hand. Proponents of AI would argue even the use of tools like ChatGPT need some level of expertise, what to ask. There is some truth in it. What differentiates us is the human element. Everyone has the tools. But tools are not going to give you the answer we need.