Questionable content, possibly linked

Tag: app

Managing gen AI assets is a PITA

As a power user of generative AI bullshit, I can say that one of the big “pain points” I’m reaching as a user is how to manage all the assets that I create, for easy sharing and other types of publishing (blog posts, ebooks, image collections, etc.)

I’m imagining a probably browser level UI, so it could function across different services, where you could click, highlight, or capture generated elements (text, images, etc.) out to a holding tank that archives them to some backup (probably local, imo). Then once in the holding tank, you can have them auto-sorted and searchable, with AI gen metadata that accurately describes them, and categorizes. You could select, arrange, compose new media which incorporate them, and export them to whatever platform, service, or format you need to publish them.

Right now, the process is a terrifying chaotic mess across many services. For images I download into Adobe Lightroom collections, and that works well enough, even if it is tedious to regularly import them – and I don’t ever put in metadata. For text, I will work on documents sometimes in Dropbox Paper, which works fine, but it’s not a comprehensive way to archive all the bullshit things that I generate with AI, so that I can use and organize them later.

I don’t want to have to spend all day sorting and organizing data, but I guess I already do that anyway, so I should stop complaining. But still, there could be a better way…

Spintax (example)

The [[entity]] was {huddled|slumped|slouched} by the {fire|toilet|window} in the [[place]] and vomited {words|cereal|light} into a porticle connected to [[ai]], an AI for [[provider]].

Javascript rapid word input tool

I went off on pretty much a tear earlier investigating the possibility of coming up with some kind of rapid communication board which would allow you to input words, not letters.

I went once or twice around the bend, and found the closest match in an app called DocsPlus which gives you the ability to create customizable word-bars. There’s a 28 day free trial. It’s interesting, but my use case is to be able to rapidly paste in the results of these sentence creation actions into Firefox in a spreadsheet. It was too combersome with switching back and forth between tabs to access other word bars.

So I cooked up some Javascript I’m still tinkering with which looks at the moment like this:

Keyboard that inputs words not letters

Is there a way on Mac OS X Sierra to enter whole words rapidly, instead of letter-by-letter, as per normal typing?

I’ve experimented a lot with Dragon Dictate for text entry and it can work well under specific circumstances – one of which is having an allowance for vocalizations in the workplace (not always convenient).

What I’m after is to basically be able to set up word banks, and then rapidly plop in values from each group to form descriptive sentences¬† for SEO on a high volume of images. Since many of the subjects of the images repeat again and again, I’m wanting to split them up into re-usable chunks.

So it could be a little like this, genericized:


Where each item is a bank of related words, which I can quickly flip through to find the correct combination, something like:

Man walking on a beach

I have aText, which is a decent basic text expander app, and I see people talking about some autocomplete options in Mac OS, but so far nothing quite fits the bill.

I guess the closest I’ve come so far has been finding (more on iOS) some applications for augmentative/assistive communications boards, like so:

If I were able to customize this kind of thing with my own word banks, and make it into like an app that can be called up system-wide (or at least in Firefox), and which will output strings of text into Google Sheets + allow for easy switching to regular text/letter-by-letter entry style, I would be pretty much golden…

Maybe I’ll just have to cobble it together myself though, it looks like.

Powered by WordPress & Theme by Anders Norén