I regret not writing any of this down sooner.
When I first encountered generative AI in December 2022, I had no idea how much it would shape the next few years of my life. The trial and error, the shifting opinions, the tools I loved and then watched disappear — none of it was documented. This blog is my attempt to fix that.
The Day I Met AI
My introduction to generative AI came through my father, who explained ChatGPT to me back in 2022. What felt like yesterday is now more than three years ago. In the time since, I've sat in on his lectures at Korea University, taken newly introduced AI courses at my own university, and woven these tools into nearly every project I've worked on. What started as a fascinating novelty quietly became an indispensable part of how I work.
Analyses with a Three-Month Shelf Life
Since then, I've used ChatGPT, Claude, and Gemini consistently — and as of May 2026, I'm subscribed to all three. I've also experimented with Perplexity, Copilot, Character.AI, and a handful of others along the way.
Early on, I spent a lot of time comparing models: response quality, creativity, coding ability, how well they handled long documents. But I kept running into the same problem. The conclusions I'd carefully worked out would be outdated within three months. Sometimes within one. Distinctions I'd taken seriously looked almost quaint just a few months later.
That's exactly why documentation matters. News in AI moves fast, and without a record, it's almost impossible to reconstruct what the landscape actually looked like at any given moment — or what you were thinking when you saw it.
Between Design and CS
I'm double-majoring in Design and Computer Science, which puts me in an interesting position when it comes to AI. These two fields don't experience it the same way.
In art-focused design communities, I've often sensed real unease about AI — a feeling that it threatens creative authorship. In CS and product design circles, the mood tends to be more pragmatic: AI is a tool, and the question is how to use it well. The same technology gets read as a creative threat in one room and a productivity multiplier in the next. Being in both rooms at once has given me a perspective I want to keep exploring here.
Things I Used, and Things That Are Gone
My relationship with Claude started earlier than most. Before it became widely used in professional settings, I came across it through character chatbot communities, where I spent time writing lorebooks, system prompts, and character configurations. Comparing how different models responded to the same setups — that's where I first got hands-on with APIs and first started paying attention to Claude specifically.
Around the same time, I was running Stable Diffusion locally at home. Not just using image generation through a web UI, but actually loading models, adjusting settings, watching how outputs changed. For a while, that community was full of energy — LoRA, prompt engineering, model merging, checkpoints. Then Midjourney matured, DALL·E got folded into ChatGPT, and things shifted. Stable Diffusion didn't disappear, but it stopped being the center of gravity it once was.
Then there's Sora. I used it when image-to-video AI felt genuinely exciting and new. On April 26, 2026, OpenAI shut down the standalone Sora platform, citing a shift toward next-generation AI and enterprise priorities — though the rise of competitors like Kling and Veo was part of the picture too. The Sora model itself may still surface inside ChatGPT eventually, but as its own product, it's done. I still have the storyboards I made with it for a class project. I'm glad I kept those, at least.
(Storyboard interface from Sora, before the service ended)
This is part of what I want to document here: not just what's trending, but what gets built up, what changes shape, and what quietly disappears.
Why This Blog Exists
This space is for recording what I notice — AI news, chatbot updates, shifts in how people actually use these tools, reactions from real communities rather than just press releases.
It's less a formal analysis column and more a running log: what caught my attention, why it mattered at the time, and what it looked like from where I was standing. A CS and Design student who's been watching this space since before it was mainstream, trying to write things down before they're gone.
No grand conclusions. Just consistent documentation, moving at the pace of a field that doesn't slow down for anyone.