- Type of folklore: Internet folklore/meme
- Language: English
- Country of Origin: USA
- Informant: Tracy Mutoni
- Source: Twitter
- Date collected: 20 May 2020
- Formerly a microblogging platform, Twitter has evolved into a social media network since being founded in 2006 by Jack Dorsey. The platform attracts everyone—regardless of their age, gender, race, religion, class, ideology, or educational background—to share or vent via tweets no matter how mundane they are.
- Cultural context: Artificial Intelligence (AI) is that distant fantasy you’ve probably heard pro-AI guys speak about; that machines are so smart they can take over the world. Training models are constantly averaged with human brains. It is true that supervised training allows models to learn complex relationships and patterns required to process routine tasks, faces, words like human brains do. However, this still requires massive computing power, hard-coded rules, and updated algorithms literally every second new data emerges up.
- Social context: This specific item challenges the credibility of pro-AI claims. One time OpenMind, AI text generation, used huge data that it took all Reddit. Unlike models, humans can learn from very limited data give that we must come with a built-in structure; that’s why there is, agreeably, machine intelligence still has a long way to go towards human intelligence.
- A given model overfits on training data. But when new data is observed in the world. The model is unqualified.
- This item symbolizes an extraordinary challenge for AI. Contrary to humans, machines cannot independently observe and take in new information and update their network accordingly. Therefore, one can be assured that a robot-governed world is nowhere near us.
- This piece is a good mockery and a joke and might elicit comebacks from pro-AI guys.
Collector’s information: Tracy Mutoni, Russian Lit 013, Dartmouth College.
#AImemes, #AImemeforconvultionalteens, #overfitting #undefitting #trainingdata #models #machinelearning