Ask Lemmy

After the staff are done drinking coffee for the night, we only brew decaf. If you want caffeinated coffee close to closing time at a restaurant, ask for an Americano or other espresso drink.

The SLA on any product is not a guarantee the platform will be fixed in that timeframe. At best it is wishful thinking.

Cars produce more harmful airbourne pollutants from their brakes than they do from the tailpipe. Copper is being phased out and the ultimate goal is to abandon friction braking entirely in favour of electrical regeneration.

It is virtually impossible to remove yourself from advertiser's rolls.

Thanks to the new CPRA regulation, you can ask companies to delete everything they know about you. Great!

Except that the way the law is written, that often includes deleting the fact that you asked to have your data removed. So the next time they get your data from a broker, (or the next time a broker gets your data), you're right back at square one.

In theory, if you managed to send simultaneous requests to every company that's holding your data, you could wipe the slate clean...until the next time you used a website.

There are so many data sets out there that we are all a part of. And if your data is in just a single one that didn't get wiped, everyone will end up with it again as a matter of course.

Technically not my industry anymore, but: companies that sell human-generated AI training data to other companies most often are selling data that a) isn't 100% human generated or b) was generated by a group of people pretending to belong to a different demographic to save money.

To give an example, let's say a company wants a training set of 50,000 text utterances of US English for chatbot training. More often than not, this data will be generated using contract workers in a non-US locale who have been told to try and sound as American as possible. The Philippines is a common choice at the moment, where workers are often paid between $1-2 an hour: more than an order of magnitude less what it would generally cost to use real US English speakers.

In the last year or so, it's also become common to generate all of the utterances using a language model, like ChatGPT. Then, you use the same worker pool to perform a post-edit task (look at what ChatGPT came up with, edit it if it's weird, and then approve it). This reduces the time that the worker needs to spend on the project while also ensuring that each datapoint has "seen a set of eyes".

Obviously, this makes for bad training data -- for one, workers from the wrong locale will not be generating the locale-specific nuance that is desired by this kind of training data. It's much worse when it's actually generated by ChatGPT, since it ends up being a kind of AI feedback loop. But every company I've worked for in that space has done it, and most of them would not be profitable at all if they actually produced the product as intended. The clients know this -- which is perhaps why it ends up being this strange facade of "yep, US English wink wink" on every project.

Inside almost every arcade cabinet is a Dell Optiplex running Windows 7, or 10 if its really recent. There's no such thing as an arcade board anymore, they're all Dells, or sometimes those HP mini PCs, usually with the protective plastic still on.

Daytona even uses a Raspberry Pi to control the second screen. SEGA intentionally ships those with no-brand SD cards that consistently fail after 3 months. It's in their agreement that you'll buy another card from them instead of just flashing the image onto an SD card that won't break.

The Mario Kart arcade cabinet uses a webcam called the "Nam-Cam" that is mounted in a chamber with no ventilation, which causes it to overheat and die every few months, so of course you'll have to replace those too. The game will refuse to boot without a working camera.

Oh yeah also all arcade games with prizes are rigged. All of them. We literally have a setting that determines how often the game will allow wins.