- Prompt engineering isn’t a skill you need to have to use AI.
- AI Chatbots are maturing rapidly and gaining a better understanding of natural language.
- Be clear about what you need and rely on the prompts the AI provides to dig deeper into conversations.
AI prompt engineering—it’s the newest buzzword (phrase), and it carries with it the connotation that you have to have some special knowledge to be able to use AI. But I’ve been using AI for a while now, and that whole notion is false.
It’s not at all uncommon to see numerous job postings for AI prompt engineers. And it seems that all you ever hear anyone talking about is “the right prompt” for whatever purpose they have in mind. There are even Chrome browser extensions to help you write prompts.
But I noticed something while attending the Adobe Max conference recently. In all the applications that were being shown off, the prompts used with the built-in AI were just normal requests, like you would make from a designer. So, I asked Parimal Deshpande, Adobe’s Senior Director, Adobe Express & Creative Cloud, about that. “Prompt engineering is evolving, it’s a new language for humanity,” he responded and then explained that if users could put words to the images they have in their head, Adobe’s tools could help them create it.
Trying to convince people that they need to become “prompt engineers” isn’t just wasting their time—it’s robbing them of the chance to discover AI’s true potential and causing real damage
Is AI Changing or Are People Learning?
It seems like AI is shifting, morphing into something a little easier to use, and maybe getting better at translating what’s asked of them. For example, I used ChatGPT to conceptualize and outline a new (fictional) podcast series. Here’s the first prompt I used:
What would be 10 possible broad topics for a podcast created by someone who is a single female, that loves making and selling art in the categories of writing, watercolor, and pottery. Choose broad topics that get high interest levels that could lead to larger audiences.
ChatGPT spit out ten really good topics, and from my perspective, that’s the same kind of question that I would ask any expert on the topic or use when brainstorming with friends about possibly creating a podcast.
But is it? One school of thought is not that AI is changing and getting better at understanding but more that people are getting better at asking questions. It’s like when Google first introduced its version of search (yes, I was around when that happened). You had to use special ‘search operators’ to get the best results—like putting quotation marks around a specific phrase to find exactly that phrase.
The thing is, you no longer have to do that. Those search operators are pretty much gone these days. You just tell Google what you want and it throws results back at you, usually very successfully (it’s almost like Google can read your mind).
Already, AI prompting is heading in that direction.
The Democratization of AI
Let’s go back to my podcast example for just a minute. Among the topics that ChatGPt returned was “Living a Creative Life.” So, my next step was to ask ChatGPT to help me build a podcast on that topic:
Could you suggest 15 possible podcast episodes that would be created and narrated by one person and targeted toward people who are Living a Creative Life: Discuss the joys, struggles, and lessons of living a creative lifestyle, balancing personal expression with practical challenges.
The bold part of that is just a copy and paste of the results that ChatGPT returned from the first query. There’s nothing there that anyone couldn’t come up with. It’s all just having a conversation just in this case, it’s with a software application rather than a real human being.
That said, there are some tips for making AI prompts work best. I usually create prompts using a three-pronged approach: Give the AI a perspective, provide an objective, and ask for a deliverable. Here’s how that broke down in the original prompt:
- Perspective: a single female, that loves making and selling art in the categories of writing, watercolor, and pottery
- Objective: Choose broad topics that get high interest levels that could lead to larger audiences
- Deliverable: What would be 10 possible broad topics for a podcast
Using that approach initially, I was able to carry on the conversation and use prompt suggestions from ChatGPT to dig deeper into the process of creating the podcast. And I did it all—concepting, outlining, and creating a first draft script for 15 episodes—in a couple of hours.
AI is evolving, just like Google evolved. You don’t even think about writing a search query anymore, and soon, the same will be true for writing AI prompts. You won’t even have to think about what to write; you’ll just do it. And the whole concept of having to be a prompt engineer to get anything useful from AI will just be a dull memory of the past that you might one day use in an ‘I walked uphill both ways in the snow’ story with your kids.
“Trying to convince people that they need to become “prompt engineers” isn’t just wasting their time—it’s robbing them of the chance to discover AI’s true potential and causing real damage,” Sudipto Chanda, an engineer and the Chief Mentor at BetterManagersClub wrote on Medium.
Thanks for letting us know!
News Summary:
- Forget Prompt Engineering: Here's How AI Can Work for You
- Check all news and articles from the latest Tech updates.
- Please Subscribe us at Google News.