Janelle Shane

Training large language models on the outputs of previous large language models leads to degraded results. Less diversity, more bias, and …jackrabbits?

https://www.scientificamerican.com/article/ai-generated-data-can-poison-future-ai-models/

#IsThisTheSingularity

AI-Generated Data Can Poison Future AI Models

As AI-generated content fills the Internet, it’s corrupting the training data for models to come. What happens when AI eats itself?

Scientific American
Jul 28, 2023 at 5:04pmWeb
Show threadJari Komppa 🇫🇮Jul 28, 2023
@janellecshane At least it's not giraffes..
Show threadJanelle ShaneJul 28, 2023
@sol_hsa I am disappointed it’s not giraffes
Show threadSamhain NightJul 28, 2023
@janellecshane @sol_hsa No, no. Giraffes are a unit of measure for asteroids.
Show threadTHIS ACCOUNT HAS MOVEDJul 28, 2023
@janellecshane yup https://www.darkreading.com/vulnerabilities-threats/tech-insight-dangers-of-using-large-language-models-before-they-are-baked
Expert Insight: Dangers of Using Large Language Models Before They Are Baked

Today's LLMs pose too many trust and security risks.

Dark Reading
Show threadMartin HamiltonJul 28, 2023
@janellecshane Cometh the hour, cometh the #AICentipede!
Show threadUlrich JunkerJul 28, 2023
@janellecshane is this a way of getting the monster to eat itself? So it will vanish in a cloud of meaninglessness in the end?
Show threadDave Dawkins (D. Harrigon)Jul 29, 2023
@janellecshane
It's jackrabbits all the way down.
Show threadPeter Prevos 💧🎓📊🇳🇱🇦🇺Aug 13, 2023
@janellecshane Relying on LLMs to write or create art is a pathway to mediocrity.