Hacker News new | ask | show | jobs
by gumby 1148 days ago
> I think that level of automation, ~100%, is unachievable for a business like this, but that's by the by

With a little flexibility in your definition, I think you could completely automate it.

1 - Have a cron job that collects trends (maybe a crawler, but perhaps you could get that info via APIs of some social media through your advertising account).

2 - generate the T shirt / whatever design using a generative package

3 - generate and place ads on various sites based on their current memes (mapping determined in step 1)

4 - take orders and send them to a print on demand shirt fulfillment service. This makes your company 100% automated, though some people are required at the fulfillment and delivery steps of course — but your payroll is nil.

This reminds me of the stories that appear from time to time in which some old person in Europe is discovered to have died in their apartment but are undiscovered for years because the pension is auto deposited, the utilities are autodebited, and what little mail arrives just comes in through the slot in the door and piles up on the floor.

This automated company could almost also survive its owner’s death, except for the tax filings.

3 comments

You can't fully automate AI content for a number of reasons, chiefly because it can generate something extremely insensitive or offensive that can torpedo the whole business (e.g. AI Seinfeld's transphobic joke), but also because gettibg AI output to stay within a certain range of acceptable responses is -incredibly- hard, and if the ratio of misses to hits is too high, you're losing money (as every shot, whether you hit or miss, incurs some costs). For more details about this, look up how AutoGPT and similar things are actually doing in practice -- they get stuck a lot and require constant supervision.

And this is because, again, keeping AI outcomes within a certain range at scale is both amazingly hard and computationally irreducible (i.e. you can't predict its responses at a scale of say 1 million shots, without taking 1 million shots).

Could you not make a censor AI that scans content for offensive metadata and rejects the ones it finds?

There may still be a bit of slipthrough but it should be dramatically mitigated.

https://m.twitch.tv/videos/1804206943

it’s in its early stages but it seems doable.

There’s even a few ways the tax could be automated depending on where it operates from.