Mindstream
Posts
💻 Should you QUIT and become a prompt engineer!?

💻 Should you QUIT and become a prompt engineer!?

+ AI image or real? 🤔

Matt
August 26, 2023

Saturday slides on in like that uninvited DM.

Now pay attention - you don’t wanna miss it!

We make AI simple and deliver it straight to your inbox, EVERY DAY!

Everything you need to know about AI today in 5 minutes. You can even read it while you sneeze.

TODAY IN MINDSTREAM:

💻 Long read: Should you quit your day job to become a prompt engineer?
🤔 AI image or real?
💡 AI-nspirational quote of the day
📸 AI image of the day

💻 LONG READ: SHOULD YOU QUIT YOUR DAY JOB TO BECOME A PROMPT ENGINEER?

Prompt engineering is a brand new profession, and yet job listings are already offering salaries of over $300,000 per year! Is this the new career of choice for AI masterminds? We dive in with this long article.

A huge thank you to Michael Ryaboy, Alex Velinov, and Karin Pespisa for their work in writing today's feature.

Michael Ryaboy, the Prompt Engineer behind GPT Design Labs in San Francisco chatted with Alex Velinov, CTO of Tag Digital to explore what it takes to enter and succeed at this new and fast-paced career:

What is a Prompt Engineer?
A Prompt Engineer (PE) is someone who interacts with generative AI models by authoring text prompts that drive the creation of images or text, or by developing systems that empower others to create with the help of these models. Most are software engineering roles in disguise.

PEs don’t think of individual prompts as much as we think of frameworks. PEs also work backward - we take high-quality examples written by people and replicate them with AI prompts. When output quality matches the human examples, our work is almost done!

How did you get your start as a PE?
I’ve been a PE for almost three years. When GPT-3 beta was released in June 2021, I emailed Greg Brockman, the president of OpenAI, begging him for access. I was straight out of high school, and somehow made it past his spam filter and received access.

From my first day interacting with OpenAI Playground, I was hooked. I learned by making some fun applications, including a satirical newspaper partly written by AI.

What does your typical day look like?
Every project requires different skills and approaches. However, tasks that I complete daily include:

Research problems the LLM needs to solve.
Design and test prompts.
Debug and improve prompts.
Document and communicate.

What are the biggest challenges PEs face, and how do you overcome them?
For me, the challenges fall into three main buckets:

Data Scarcity or Low Quality
This is overcome by fine-tuning, finding alternative data sources or in some cases just waiting until a model has been improved enough to deliver what is needed!
Prompt Complexity or Uncertainty
This is a difficult challenge to solve. It's about finding the right balance between simplicity versus expressiveness, accuracy versus diversity, and speed versus quality. Once priorities are decided upon, the model can be tuned for the best output depending on application.
Prompt Evaluation
It takes a lot of manual effort to sort the good prompts from the bad. User feedback is key to improving prompts further, once they are in use.

So, Michael, should I quit my day job to be a PE?
AI prompt engineering is a skill much like Microsoft Excel. Everybody needs the basics, but some people can get good enough to make it their career. The two most important qualities are curiosity and a love for making cool things. Whether you’re a Midjourney master or you can get ChatGPT to do incredible things the average user can’t, when you make cool things happen and share them, opportunities will follow.

Should you quit your day job? The answer is contextual. Companies want someone who has interacted with LLMs, but it's more important to have software engineering experience and the ability to learn new technologies. If you're an adaptable software engineer, don't hesitate to apply for PE roles. You might be exactly what the company needs.

Thank you once again to Michael Ryaboy, Alex Velinov, and Karin Pespisa for contributing today's feature, which you can read in full and unabridged by clicking below.