Getty Images/Westend61
In my previous career, I held various managerial positions, including product marketing executive and publisher. I had the responsibility of overseeing editors, salespeople, programmers, manufacturing teams, and other executives. However, one of the best aspects of my current career as an advisor, pundit, and columnist is that I no longer have direct reports to manage.
Contrary to popular belief, being a manager is not just about bossing people around and delegating work. Managers spend a significant amount of time trying to ensure that their employees execute their job duties as instructed. This challenge often arises due to misinterpretation of instructions, passive-aggressive behavior, or the need for negotiation.
This is why I find programming so appealing. When you write code, the computer will do exactly what you tell it to do. Although there may be bugs initially, the code itself is consistent and predictable. However, working with AI is a different story. Instructing an AI, such as ChatGPT, is more akin to managing a programmer than actual programming. Everything is subject to interpretation and negotiation. While you can achieve results, it often requires multiple attempts and adjustments to get it right.
For instance, giving the same prompt to an AI twice can yield different results. On the other hand, running the same code twice will produce identical outcomes, barring any randomization or serious bugs. This raises the question of whether AI will replace programming jobs. I pondered this question while working on an article about advanced prompt writing. During my experimentation, it took numerous attempts and several hours to get ChatGPT to consistently solve a simple problem. The prompt was to find a word similar to “devolve” that begins with a “B.” Despite my corrections, ChatGPT repeatedly provided answers starting with a “D.” It felt like negotiating with a stubborn employee who couldn’t grasp my instructions.
This reminded me of a time when I managed salespeople who were tasked with making calls to potential clients. I provided them with a precise script on how to pitch our services. However, one salesperson refused to stick to the script, resulting in misconceptions about our offerings among the prospects she contacted. She preferred her own description because it made scheduling appointments easier, even though our goal was to make sales. This situation parallels the experience with ChatGPT. After spending hours trying to convince it that “devolve” starts with a “D,” I felt like I was negotiating rather than coding. I found myself haggling with a robot and questioning whether this is truly progress.
As someone who has long been fascinated by AI and has witnessed its evolution, I expected it to be closer to my dreams. However, after struggling with ChatGPT, I felt frustrated and reminiscent of managing difficult employees in the past. Eventually, I did find a reliable prompt and detailed its effectiveness in the article. Yet, it became evident to me that while AI may be capable of taking on low-level programming jobs, its similarity to human employees might provide some protection for human workers.
In summary, there are tasks where coding is easier and others where using AI is more convenient. The combination of both can be particularly intriguing. However, it is important to recognize that AI does not eliminate the need for human skills and expertise. The following table illustrates this:
| Task | Coding | AI |
|————-|—————|—————|
| Getting data| Requires | Simplifies |
| | finding large | data gathering|
| | datasets | |
In conclusion, AI and programming have their respective strengths and weaknesses. While AI shows promise in certain areas, it does not diminish the importance of human involvement and expertise. As we continue to explore the possibilities of AI, it is crucial to strike a balance between human and machine capabilities.
