4 ms·

Generative AI Is a Parasitic Cancer (Freya Holmer) [video]

20 points by f1shy 2y ago

tonetegeatinst 2y ago
I mean search and other online stuff has been enshitified way before gpt3 was released. I get plenty of people don't like AI, but a lot of the arguments seem to be confusing. Like, AI has been a thing for a very long time. I'v done plenty of hobby research and development using AI. Be it CNN, GAN's or development of classifiers. This is all AI, and LLM's or image generation is no different. This "AI got good" is just what happens when you can manage to create really good datasets for training and are able to throw lots of compute at the problem. If we made 4k scans of handwriting, we could easily generate synthetic handwriting that looks realistic, but based on how we classify the data you might even be able to determine different handwriting authors. This is. Of a breakthrough, its an application of using large data that's been classified correctly, and then using the massive compute that's available now.
- ashoeafoot 2y ago
  Agree generative AI is a cancer cancer.
- latexr 2y ago
  > I mean search and other online stuff has been enshitified way before gpt3 was released. I don’t know how many more times this argument needs to be made, but degree matters. It’s like someone complained about semi-automatic weapons and the answer was “people were killed with knives before”. Yeah, but certain tools make certain actions easier, simpler, and more prevalent at scale. So they happen more. This isn’t a hard concept. > but a lot of the arguments seem to be confusing. > Like, AI has been a thing for a very long time. No one is confused. Yes, AI is a bad term for LLMs and image generation, but that term was overloaded by OpenAI and unfortunately we’re now stuck with it for the near term. We have to use words that people understand. They could’ve called it Pipicaca and we would still have to use the term, regardless of how bad or inaccurate it is, if it had stuck. People who don’t understand that and try to force their own specific terms on every conversation are the confused ones. If you want to be understood, you have to speak in a way people understand you.
  - f1shy 2y ago
    Yes, the problem is that now when searching it is getting harder to find things. It is getting more difficult to sort out bad sites. Is a problem of ratio between good human made sites, and AI rehashed trash.
PicassoCTs 2y ago
You can almost smell the torches in the distance and hear the pikes clanging against each other. Chat GPT write me a script that will delete my AI promo posts from LinkedIn, the moment it finds CEOs dangling from lamppost in https://trends.google.com/trends/ https://trends.google.com/trends/ Some people really thought the economy is there to please itself, to give shareholders great returns instead of bribing baseline humanity into peace and complexity with a constant surplus.
Sateeshm 2y ago
The STEP file podcast was hilarious. Imagine the energy spent on generating the 43 minute audio garbage. Future is now.

20 points by f1shy 2y ago

I mean search and other online stuff has been enshitified way before gpt3 was released. I get plenty of people don't like AI, but a lot of the arguments seem to be confusing. Like, AI has been a thing for a very long time. I'v done plenty of hobby research and development using AI. Be it CNN, GAN's or development of classifiers. This is all AI, and LLM's or image generation is no different. This "AI got good" is just what happens when you can manage to create really good datasets for training and are able to throw lots of compute at the problem. If we made 4k scans of handwriting, we could easily generate synthetic handwriting that looks realistic, but based on how we classify the data you might even be able to determine different handwriting authors. This is. Of a breakthrough, its an application of using large data that's been classified correctly, and then using the massive compute that's available now.

ashoeafoot 2y ago
Agree generative AI is a cancer cancer.
latexr 2y ago
> I mean search and other online stuff has been enshitified way before gpt3 was released. I don’t know how many more times this argument needs to be made, but degree matters. It’s like someone complained about semi-automatic weapons and the answer was “people were killed with knives before”. Yeah, but certain tools make certain actions easier, simpler, and more prevalent at scale. So they happen more. This isn’t a hard concept. > but a lot of the arguments seem to be confusing. > Like, AI has been a thing for a very long time. No one is confused. Yes, AI is a bad term for LLMs and image generation, but that term was overloaded by OpenAI and unfortunately we’re now stuck with it for the near term. We have to use words that people understand. They could’ve called it Pipicaca and we would still have to use the term, regardless of how bad or inaccurate it is, if it had stuck. People who don’t understand that and try to force their own specific terms on every conversation are the confused ones. If you want to be understood, you have to speak in a way people understand you.
- f1shy 2y ago
  Yes, the problem is that now when searching it is getting harder to find things. It is getting more difficult to sort out bad sites. Is a problem of ratio between good human made sites, and AI rehashed trash.

PicassoCTs 2y ago

You can almost smell the torches in the distance and hear the pikes clanging against each other. Chat GPT write me a script that will delete my AI promo posts from LinkedIn, the moment it finds CEOs dangling from lamppost in https://trends.google.com/trends/ https://trends.google.com/trends/ Some people really thought the economy is there to please itself, to give shareholders great returns instead of bribing baseline humanity into peace and complexity with a constant surplus.

Sateeshm 2y ago

The STEP file podcast was hilarious. Imagine the energy spent on generating the 43 minute audio garbage. Future is now.