HackerTrans
TopNewTrendsCommentsPastAskShowJobs

goombal

no profile record

comments

goombal
·4 tahun yang lalu·discuss
Super interesting point, though it's important to note that this extends beyond text (which everyone is understandably focused on after ChatGPT's release).

AI "art" (image gen) is also becoming more popular and lots of interesting business applications around marketing images and product listing images. One thing to note is if you look at something like Amazon where many listings have terrible images, using AI could actually improve them straight away by using some kind of Stable Diffusion / DreamBooth-type model with an API like https://dreamharbor.ai

For more generic images (e.g. for blog posts) where you just want "a dog sat on a cloud" then places like https://lexica.art let you produce these images for free. The key with both of those is learning prompt engineering for which there are tonnes of resources - sites like https://prompthero.com/ are already getting involved in this space and on that dreamharbor site there's a pretty detailed blog post: https://blog.dreamharbor.ai/blog/how-to-write-prompt-stable-...

And then when it comes to video Synthensia is doing some very exciting work - but still some way to go. They could actually enhance international growth by providing better dubbing (potentially).

Exciting times!