Artificial Intelligence
I see the boom and I poke at AI. I've got a pretty good understanding of Artificial Intelligence (AI), Machine Learning (ML), and Data Science. I will NEVER say I'm an expert, but I can hold my own when necessary.
My fear has always been misuse.
I've heard stories ...
- People losing work.
- Databases getting deleted.
- Students using AI generated products and feeling the understand the content (code or otherwise).
Is there a place for AI in the coding world?
I firmly believe there is and that we're just beginning to scratch the surface of "good implementations."
I also firmly believe that we've got more "poor implementations" than good.
The Pixel-Perfect Design
A friend of mine, someone that is a good developer, asked me to step in and help him learn CSS.
Learning CSS
This is a huge task, monumental.
I started researching and built an outline of content.
This could be ...
- A new course for Code Squid.
- A series of articles.
- A series of videos.
- A series of conference presentations.
Oh, my ...
Back to the Design
Yesterday, he showed me the Adobe XD design. It wasn't complicated. He told me that his company truly wanted a pixel-perfect design.
Pixel-perfect
Pixel-perfect design is the meticulous creation of digital designs where every element is rendered exactly as intended, with precise attention to detail, alignment, spacing, and visual consistency across all devices and screens. It involves ensuring that the final product, such as a website or app, appears the same to users as it does in the original design file, paying close attention to every pixel. This approach aims for visual fidelity by meticulously controlling aspects like typography, color, and layout, using tools and systems to achieve exact specifications.
What Is Actually Pixel Perfect Design and How to Create It?
Then, I asked him to show me how he was working. What I saw was painful to watch.
He showed me the code related to the design he was working on. We started with adjusting a breadcrumb element under the header.
Then, I saw a strong, mid-level developer go into the DOM and select the element, right-click, and copy the element. Within his AI-tool of choice, he pasted the element an prompted it to "change the margin to 24px". This prompting had to be adjusted a few times to get the result "pixel-perfect."
I actually watched him do this a few times.
I asked about the title on the breadcrumb since it wasn't right.
He started to write a prompt to change the text on the breadcrumb using the same copy, paste, prompt method.
At this point, I stopped him ...
Confrontation
I stopped him and asked why he was using AI for such simple tasks.
In this case, doing a global search for the text to change yielded 3 results. The third value was the one that needed changed.
This whole process really relied on basic problem solving:
- Define the problem (incorrect text).
- Generate potential solutions (the text probably exists within the project, find it and change it).
- Choose the best solution (step through the three versions and see if the content on-screen changes).
- Implement the solution (save).
- Evaluate the results (verified when changed above).
The whole process took just a few seconds and relied on problem solving skills. It didn't have to be offloaded to a tool
I confronted him.
You have the skill to solve these problems. Try to do it yourself and learn rather than offloading the problem onto a tool that might be able to solve the problem. Use the AI-tool when you get stuck and learn from the answers provided.
Pixel-Perfect
I confronted him ... and presented things in a different light.
A pixel-perfect design is difficult to achieve.
I suggested that this is as important as any online code challenge. The skills he can gain as a frontend developer will take him to a completely different level.
Pixel-perfect designs were the most challenging for me, and the most fun. I am extremely good at what I do as a frontend developer and a lot of that comes to doing this kind of development work early in my career.
Notes from a Friend
This is a different friend.
People need to be a "good or GREAT" prompt engineers to use AI since AI's age level is (currently) around 5-8 years old brain. You have to explain in detail, and I mean every detail, then it will execute it correctly. It's all about weights in training. Most people don't know how AI is trained. They just want to use it and think it is on par with talking to a mid level developer or mid level expert -- which is all hyped up by social media.
There's nothing "intelligent" in todays AI.
It's an extremely good search engine that is providing results efficiently and in a human-like manner.
That doesn't mean it's bad. It's not a bad set of tools; neither is it a good set of tools. It's just simply that, a tool.
If the tool is used correctly, you get good results.
If the tool us used incorrectly, you get bad results.
Conclusion
I understand that AI is everywhere and my fear has always been misuse.
This is one more story.
My friend is looking to do the best job he can and is doubting his abilities (maybe a bit of imposter syndrome). The results that AI can generate for him are alluring.
The reality is, that at the end of the day he's got the skill and can do the job, he just needs to see it in himself.
Is there a place for AI in the coding world?
I firmly believe there is and that we're just beginning to scratch the surface of "good implementations."
I also firmly believe that we've got more "poor implementations" than good.
Top comments (0)