Have you experimented with AI for creative tasks like image generation or music? How do you use AI, if at all? Share your experiences and thoughts with the community.
1 Like
I run a non-profit, and manage a self-development movement. I use AI to help me strategise and create marketing campaigns. I love using the various LLMs to see which ones perform using the same prompts and contextualisation. Honestly, if I could get paid to do all this, I’d need nothing else! ai
1 Like
This is so interesting. Which LLMs do you prefer for marketing?
Strategic planning - NotebookLM
Marketing - Claude
ChatGPT - random questions
1 Like
NotebookLM is so good. I just throw all my research and drafts into a container and then can verify facts and edit my copy for accuracy on the fly. Claude and Gemini are also great tools to help cut down if I feel that I’ve overwritten. I try steer clear of ChatGPT, but it is great to get a start on developing infographics
2 Likes
I have played a little bit with AI in Canva and in Midjourney and recently with Soro. For me it is a little bit hit and miss, Canva’s AI has made me some beautiful pics BUT often the pics are really fake looking and clearly AI and not at all compelling. I know the more sophisticated tools (and users) are doing amazing work
2 Likes
I dabble in vibe-coding, not day-to-day but fairly often. I need a use-case and i need a “thing“ ie code that could plausibly be fixed and troubleshot - i don’t like invisible machinery. I enjoy that it makes me think pretty hard and i actually learn a lot as i’m doing it.
I also ask chatgpt questions about how to get my guitar tone how i want 
2 Likes
Very useful for coding – makes it much faster.
Though has some caveats - you need to be very precise and clear - and go through a multi step process to reject dumb ideas it generates and modify appropriately.
2 Likes
Shameless plug, but most of my knowledge on different ways to use AI comes from this webinar with Jeremy Caplan (though I still worry about the impact AI has/will have on our environment and society.) Daily Maverick Events
https://youtu.be/o3xsGBr4B7c
I’ve attempted to turn portrait photos into semi-anonymous avatar images using Google Gemini. The results looked more like a kindergarten crayon drawing than the Roy Lichtenstien effect that I was looking for.
ChatGPT gave pretty good results when asked to knock together a web app from scratch. But getting it to update more complicated code was an exercise in frustration.
2 Likes
Have you tried Claude? Behaves like an IDE - i hit a brick wall pretty early on due to bad UI.
Have had best success with Vercel - but has a preference for TypeScript which might be a dealbreaker for some.
1 Like
The UI is quite horrid, but it’s the best balance between capability and usability. Gemini i find is the most capable as an agent - but this is because I’m heavily invested in Google services. ChatGPT is too eager to please, even when it gets things wrong in its attempt to provide a positive outcome
I must be the odd one out, I use duck.ai for general queries and Hugging Face Spaces where I can try out the latest research AI models for free. I avoid directly using any of the commercial offerings as I only lightly and intentionally use AI.
2 Likes