Siqi Jiang

@siqijiang
14 Followers
14 Following
44 Posts
Used ChatGPT to generate 30+ ideas fast—plus it flagged risks & ethical issues for each, which was impressive. It even combined top ideas into stronger features (nice touch). But during evaluation, it kinda cut corners. Next time I’ll try DeepSeek… if it stops freezing on uploads .Full log here:https://docs.google.com/document/d/1LmNMNMKTN1qDUlk32nWpvcFux0ej0q4PU5DOUDkcK50/edit?tab=t.0
#aiexperimentlog #3pinterest #medsuccess
Tested IdeaMap to turn my HMW questions into a mind map. It pumped out tons of ideas in under a minute—which was cool—but the quality was meh.Gonna try GPT or DeepSeek next for better, more inspiring ideas.Full log here:https://docs.google.com/document/d/1V_ljn17ZCCbHRqwHtoK-CwySApQAYzgrHA4vVMQJddE/edit?tab=t.0
#aiexperimentlog #3pinterest #lowsuccess
@chengchenzhang I didn’t know that Google AI Studio could generate visual sketches — it’s an AI platform I’ve never used before. Maybe I’ll try it out in one of my upcoming experiments. 
@ai_sha5 It looks like Claude was really trying hard to use a mind map to connect all the concepts… but it just turned into a tangled mess, 
Tried using Deep Thinking mode to generate HMW questions based on my journey map, persona, and interviews—and it crushed it. The questions were grouped by theme, prioritized with solid reasoning, and grounded in actual user pain points.
Full log here: https://docs.google.com/document/d/1LQ1BLZDBBjzf8omMtxjd7EA7N5igbgaAymdD_7doBns/edit?tab=t.0
#aiexperimentlog #3pinterest #highsuccess
Tried GPT’s new image generation feature with my persona—major upgrade overall. It now handles text in images really well and generates visuals that actually make sense. But when it came to user journey maps? Total fail. It couldn’t handle the complexity or large chunks of text at all. Still, cool progress. Going to test next in Chinese 👀
Full log here: https://docs.google.com/document/d/1f6rhB16rEnvmuQWOJ7Qj-uL_pGSmLvr3MgtWAOpzfq0/edit?tab=t.0
#aiexperimentlog #3pinterest
#medsuccess
Tried the Deep Thinking mode this time—and wow, total upgrade. Instead of giving me a shallow response, it asked smart follow-ups to get more context. The final output was super aligned with user pain points and actually useful. Took 8 minutes, but honestly? Worth it. Full log here: https://docs.google.com/document/d/1Ybn0Ay4HqANpHdY2Gh7J80HO4dxbYGnFh2NAFJWXqEM/edit?tab=t.0
#aiexperimentlog #3pinterest #highsuccess
Kicked off the Define → Ideation stage using NNG Research Assistant—shared all my findings and asked it to guide me step-by-step. It gave me a solid framework fast, which definitely helped me move forward. A little too surface-level, though. Next time I’m diving into deep thinking mode. Full log here: https://docs.google.com/document/d/1mAjwK8PRltiwP-VWtMvZmDYglovi63kaV2XFQYwGLS4/edit?tab=t.0#heading=h.6swsupg9gig5
#aiexperimentlog #3pinterest #medsuccess
@KadammaJackson Using NotebookLM does save me from uploading my files over and over again, but I noticed that after I refreshed the page, my chat history disappeared. I'm not sure if it's just happening to me or if others have the same issue.
@KadammaJackson I’ve never used notebook LM before. It looks great—maybe I can use it in my next experiment!