"Trained on" is doing a lot of work in many services.
Foundation models are "trained" on a huge amount of data. To imitate a person a service is then not "trained on" someone's text messages for example. The text messages are not enough to create a passable output from a computer.
The text messages are an input to an already trained model, where the original training material is just not disclosed.
I feel like "trained on" is often dangerously misleading when it refers to a tool built on top of a foundation model. It doesn't really acknowledge that any such service that will "imitate" a person also has thousands if not millions of other human voices as part of the computations it draws from.
And actually "trained on" is misleading in many other ways but I'm out of breath 😅
Anyway, I was listening to @404mediaco's episode on the "AI" avatar of a killed man that "testified" in court. Worth a listen.
https://pca.st/episode/8bc12731-f6db-47a9-a800-14916f1cce3d
Per Axbom
Speaking of efficiency, imagine how much time could be shifted from tech grifts to actual caring and love.
Speaking of efficiency, imagine how much time could be shifted from tech grifts to actual caring and love.
@PavelASamsonov Truly makes it hard to find motivation for a lot of work.
@perb Tack, får så många bra tips här. Och jävlar vilken fantastisk recension… https://www.techradar.com/audio/portable-media-players/astell-and-kern-aandnorma-sr35-review
@JMMaok I remember trying a bone conduction headset many years ago. It kinda freaked me out. 😂
@renedario Nice. And they appear to have models with buetooth.
@JoWa galet snygg!!
I’m considering a music player so I can play music without being attached to my phone. Ideally something like the iPod nano. Are there any decent music players on the market today or is my best bet to actually try and find an old iPod nano?
I’m really nostalgic for the original Sony Walkman… Apparently there is a Sony Walkman player on the market today for almost 4,000 dollars. 🤪
It’s mindblowing to me how many support bots are being implemented to save time for support staff on “simpler issues”, instead of improving the interface and interactions in ways that remove the need to ask those simpler questions.
Perpetual bot investments versus fix the issues 🤷
It’s mindblowing to me how many support bots are being implemented to save time for support staff on “simpler issues”, instead of improving the interface and interactions in ways that remove the need to ask those simpler questions.
Perpetual bot investments versus fix the issues 🤷
@Thelie There are quite a few papers these days that look suspiciously like blog posts.
@clarablackink Agree all the way. I really appreciate Timnit Gebru’s work on TESCREAL, with regards to eugenics. I try to bring this point forward in conversations sometimes but am often met with skepticism. People don’t like to hear that the tools they are using enable racist thinking.
@robertklaschka Ah, nice! Thanks for sharing.
@tymwol Yeah, I should stay away.
@tymwol Yeah, I should stay away.
I was reading a debate on LinkedIn where one side was arguing that "AI" image generators are more energy-efficient than hiring a photographer (or buying a stock photo).
One argument that made me do a painful face-palm was the notion that photographers need to eat, so they consume more energy.
The thing about humans is… wait for it… they ALWAYS need to eat. You can’t shift production to a computer and argue that it’s more efficient because it doesn’t eat. The human will keep eating(!) You are only adding energy-use.
These types of arguments illustrate a deep obliviousness towards human wellbeing. The argument only works if you expect the human to eat less when their job is taken away. Or die.
Their only data point for “AI" energy use was the point of generation, neatly excluding:
- all the energy used when scraping the internet for all the required datasets
- the energy used when building the model itself
- the energy used for required human workforce tasks, such as labeling (I bet those humans have to eat too)
- the energy used when building required data centers, both in-use and speculative investments (so many eating humans involved)
- the energy for hardware production and ongoing replacements
- the fact that the ease-of-use (and being “free”) encourages way, way more image generation than ever before…
Oh, and one more thing…
The image generator feeds on, and requires, photos and images taken and made by humans… who eat. Damn those humans.
Oh wait, the person doing the prompting… that person has to eat as well. It’s almost like it’s a human imperative.
Worth noting is also that none of the big "AI” companies are at all transparent about their energy consumption for this technology. Nor water consumption. Nor noise pollution. Nor e-waste. So there’s a ton of speculation.
I just wanted to make clear that I’m all for allowing humans to eat.
