top of page
Writer's pictureStewart Severino

Stop focusing on prompt engineering

As artificial intelligence continues to develop, the limelight is often on the buzzworthy skill of 'prompt engineering' - the art of perfecting the textual input to optimize communication with AI systems. You have seen them on Linkedin. Professionals bombard your feed with content like “Best Prompts to use in GPT-4”, or “100 awesome prompts to help with AI productivity”.

Hailed as a 'job of the future', it currently enjoys significant acclaim. However, with AI becoming more self-reliant and intuitive in processing natural language, the need for such human-engineered prompts may decrease over time. More critically, the effectiveness of these prompts is often limited to specific AI models, restricting their wider applicability. Considering this, we need to shift our focus towards a more enduring and applicable skill set - problem formulation.

Problem formulation steps into the spotlight as an essential tool to effectively engage with AI. Instead of meticulously curating the perfect linguistic input, problem formulation encourages us to understand and outline the problem, its scope, and limitations. This involves a deep dive into the domain of the problem and the ability to extract and identify real-world issues. Although it may not be as exciting as prompt engineering initially, a well-defined problem statement trumps a finely crafted prompt when a clear solution is required. Think of it as detailed directions or instructions that give the AI the context it needs to answer effectively.

Despite its importance, many professionals struggle to grasp the essence of problem formulation. The focus is often skewed towards problem-solving, overshadowing the critical process of problem formulation. The common saying, "don't bring me problems, bring me solutions," embodies this unbalanced emphasis.

Evidence of this trend can be seen in a recent survey, which revealed that a staggering 85% of top-tier executives believed their organizations were not adept at identifying problems. This calls for an immediate shift in focus towards refining our problem formulation skills. To master problem formulation, we need to understand its four key components: problem diagnosis, decomposition, reframing, and constraint design.

Similar to Root Cause Analysis, problem diagnosis focuses on pinpointing the core issue, while decomposition is about breaking down a complex issue into smaller, more manageable parts. Reframing enables us to look at the problem from different angles, and constraint design sets the boundaries for our problem.

Many experts, despite their extensive knowledge in their respective fields, find it challenging to effectively diagnose, decompose, reframe, and set the constraints of a problem. This limitation restricts their ability to devise effective solutions, emphasizing the importance of mastering problem formulation.

Looking ahead, problem formulation could prove to be as crucial as the learning of programming languages during the nascent days of computing (remember writing SQL statements?). It ensures that AI outcomes are well-aligned with task objectives and cultivates effective collaboration with AI systems.

While the glamour of prompt engineering may fade and most likely will, the art of problem formulation promises to remain an enduring skill. It might just be the guiding beacon we need in the rapidly changing AI landscape. It is time we turn our attention towards nurturing this critical skill.

7 views0 comments

Recent Posts

See All

Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page