So Long “Prompt Engineering,” We Hardly Knew Ya
Thinking that prompt engineering is the future is to not fully understand what machine learning does.
Prompt engineering.
We see the term everywhere.
It’s the hot topic, the new darling of the AI world.
The World Economic Forum, Open AI’s Sam Altman, and the Twitterverse can’t stop talking about it.
I get at least two dozen ads in my feeds trying to sell me courses in the “next big thing” and make $500,000 a year with ease. No experience necessary.
Yeah, that’s how life works.
But the uncomfortable truth is: Prompt Engineering is facing a cruel sunset.
Why?
How could this just-discovered, highly lucrative gig be going away so soon?
Three big reasons.
One:
AI is getting smarter.
Fast. Really fast.
The machines are beginning to grasp our words, our phrases, just like you and I do. It is like a child that is learning to talk, and we don’t have to spell out what we want as clearly.
The need for finely tuned prompts is decreasing. The machines will develop their own prompts by just asking a question.
See, they’re learning all the time.
All. The. Time.
Two:
AI is already crafting its own prompts.
I already use small nudge prompts that AI takes over and creates detailed and contextually correct prompts based on my small amount of input.
GPT4 is totally capable of this already.
GPT 5 will have it baked in.
So, while prompt engineering has been gaining popularity for marketers and the techbros, it has a shelf life that is close to expiring.
Three:
Prompts aren’t one-size-fits-all.
They’re custom-made for specific AI models, specific versions.
Their versatility is really limited.
And the AI itself can work around those limitations faster than humans.
As I said, the ability to do something faster, with less input, and less friction is exactly what machine learning has at the core of its DNA.
AI isn’t going to be limited to the frailty of humans and their poorly worded prompts. It is already learning those tools and improving on them.
So what human AI skill is going to withstand the test of time? What’s our enduring power tool in the AI age?
Problem formulation.
It’s all about how we humans identify, analyze, and define problems. When we can illustrate the problem, AI will provide the most efficient way to counter it.
AI cannot identify an unquantifiable problem that is not part of a current system, that’s the purview of humanity.
For now.
A quick comparison.
Prompt engineering is about the words, the sentence structure, the punctuation.
Problem formulation is about defining the problem.
It’s the bigger picture.
The broader strokes.
The difference?
Without a well-defined problem, the best-crafted prompt is just a beautiful set of words.
Sadly, we’ve underestimated problem formulation. It’s been in the shadow of problem-solving for too long.
Perhaps because it is not easy to do, isn’t taught in universities, or written about in books by futurists.
But it is real.
We’ve heard this over and over again: “Don’t bring me problems, bring me solutions.”
And that’s part of the problem. Most of the problem.
It’s a glaring symptom of the lack of problem-diagnosis skills.
It's no surprise that 85% of executives feel their organizations struggle with diagnosing problems.
How can we stay ahead?
Better problem formulation.
And there are four ways to examine and define problems.
- diagnosis
- decomposition
- reframing
- constraint design.
Diagnosis is discovering the problem that AI can solve. This is the human part of knowing that a problem exists. Learning to ask the right questions, look at the different ways that the problem can be seen.
Decomposition is about splitting the big problems into bite-sized ones. Take the problem apart, examine it, and let AI help you determine your findings since it handles data so well. Instead of tackling the biggest problem, take it apart and work on the smaller parts to achieve small successes.
Reframing is about shifting your perspective and seeking new interpretations. Extrapolating and recombining the parts of the problem in order to identify the meta components. Perhaps a new way of looking at the problem may find a solution hidden in plain sight.
Constraint design is about setting boundaries for the solution. Knowing what to accomplish, and when to know it is done. Setting the length, style, and description of the audience can help AI understand its mission. But we have to know that first in order to instruct.
The world of AI is changing fast.
In order to take advantage of AI, learn to identify the problem with as much clarity as possible.
Once the problem is clearly identified and explained, AI will be able to prompt itself to find potential solutions.
Full disclosure: I am of two minds about AI. While I do like the ability of GPT and LLMs to educate and help in many ways, I am also wary of the people who are pushing it into our lives so quickly that we have had no input, no time to discuss this radically new world with our peers, our countrymen, or even our brothers and sisters across the globe. I actually trust AI more than the techbro billionaires who are manipulating it. And us.
I loathe MidJourney and fake art. Clear enough for ya? :-)
Hi, I’m Don Giannatti, a photographer and mentor for up-and-coming photographers. You can find me on my own site, Don Giannatti, and at my Substack site, where I also publish for creative people. All free subscribers to my Substack have access to a free long-form workshop on the business of commercial and professional photography as well as my latest book.
