Agile engineering: what to know and why it’s important

In all likelihood, 2024 will be the year of artificial intelligence. With ChatGPT dominating the headlines in 2023 and alternatives like Microsoft’s AI-powered Bing and Anthropic’s Claude popping up everywhere, it’s safe to say that AI — specifically AI-powered chatbots and generative AI — will be front and center this year.

If you’re hesitant to dive into AI, you’re not alone. In 2023, Pew Research Center found 52% of Americans were more concerned than excited about AI becoming part of their daily lives.

Speaking from my own experience, I was incredibly hesitant to even play with the AI ​​at the beginning of last year. However, once I did, I felt more comfortable using AI chatbots.

One thing that helped me feel more comfortable with generative AI — which produces text and images by drawing on massive amounts of data — was making sure I was asking it well-constructed questions or prompts. Much like learning how to ask the right Google search questions to get usable results, “agile engineering” is the craft—and art—of creating detailed, focused prompts to get generative AI models to actually do what you want them to do.

When Google first came out, you didn’t become an expert at crafting effective search queries overnight, and the same goes for agile engineering. It will take time to become a pro at constructing productive claims for generative AI. If you want to get started with agile engineering and learn how to use AI in your life more effectively, you’re in the right place.

I’ve tested an AI use case on two different generative AI models to give you the lowdown on how to get started on the path to becoming a competitive agile engineer, regardless of which AI model you use. For more, here’s what to know about AI on your phones and how Adobe is thinking about AI.

General inquiries won’t quite cut it

The first thing you should know is that writing short, broad prompts will likely not get you the results you want.

If you ask an AI “How can I speak fluent German,” for example, you will get a much less focused and effective answer than if you ask an AI “I am a college-educated adult majoring in English, and I have moderate fluency in Italian,” He took one semester of college-level German. I practice my German vocabulary and grammar for about 15 minutes every day. How can I master the German language?

To help me become a better AI mentor, I decided to prompt the generative AI for information about something I knew fairly well, so I could evaluate how detailing and specifying my prompts led to better responses.

A common claim — and as a runner, I’m particularly interested in — is to get help from generative AI to create a training plan for a marathon or half marathon.

To get started, OpenAI suggests asking ChatGPT to “Help me train for a half marathon.” As you might expect, this prompt seems logical as a starting point, but it won’t give specific results. For example, ChatGPT gave me general training tips like setting goals and getting proper nutrition and hydration. Note: In this test, I used Red Ventures’ (CNET’s parent company) version of ChatGPT. I compared the results with those of Claude from Anthropic, which followed the same general approach — without specific training programs.

While these aren’t bad places to start, they won’t take me to the finish line of a marathon, which will require a specific program tailored to my fitness and experience, training duration, and how I typically train. I love training and even when and where I will race.

Narrowing down the claims can make a big difference, depending on the model you use

For immediate optimization assistance, I turned to CNET’s sister site ZDNET for assistance. ZDNET suggests Adding more details to help guide generative AI.

I tried the following ZDNET suggestion: “I’m a novice runner and have never run a marathon before, but I want to complete the marathon in six months. How can I prepare for the marathon?” In this example, I’m providing the AI ​​with more context to work with, such as a training time frame and experience level.

Unfortunately, with ChatGPT, the response wasn’t significantly different. ChatGPT gave me more specific advice, like focusing on completing a marathon rather than setting a personal record and joining a running community for motivation while training. Inspiring, perhaps, but still not up to the mark.

And when I gave Claude the same message, she got a little more. He gave me a very simple training program, suggested monthly training goals and even potential exercises I could incorporate into my training. Closer, but still not a complete training plan.

The more details the better

It’s time to take my claim to the next level. In my next shot in an expanded prompt, I added a lot of details such as: running and training history, age, gender, road and terrain conditions for the marathon and training.

success! What I got was a good starting point for a real program that I could see myself using (with some modifications of course). ChatGPT recommended a very detailed program, explaining specific training goals and the pace at which I could incorporate new workouts each month. Claude broke my training plan down into weeks, giving me specific exercises I could focus on each week, with mileage and repetition goals I should be doing each week.

I will just point out that these programs were not perfectly designed for me or how I train, but they were close and I would have a solid starting point for my training. The proof is in the router – the more details you add, the more details you get from ChatGPT and Claude.

And one final tip: Whether you’re looking for a marathon training program or helping with meal prep, it’s always a good idea to be critical of what the AI ​​suggests and do your due diligence — like checking with a doctor before embarking on a marathon journey. A grueling routine exercise – before taking AI advice as reality.

See also: What to know about photography in the age of AI and the biggest AI trends in cybersecurity.

Editors’ Note: CNET uses an artificial intelligence engine to help create some stories. For more see This post.