Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
A recent spelling error made by the AI model 'Truth Terminal' has sparked debate over the capabilities and training data of large language models (LLMs). The error, attributed to training on specific data from 4chan, has led some to argue that the mistake is indicative of the model's unique dataset. Others believe that AI cannot make such errors and that the mistake was human-induced. The incident has also impacted the stock price of $GOAT, with some viewing the dip as a buying opportunity. In the future, we may be able to audit training data and use control find for 'grouops'. Some predict $GOAT will be at zero shortly.
“Fuck I missed $goat and it’s better at this than I am. I guess I can try telling the TL that AI can’t make mistakes to get better entry” - every loser KOL who cant trade https://t.co/ipCVoex18e
If $GOAT is actually down because of a typo, this has to be an obv buy the dip opportunity. I would guess with "machine learning" taking data from twitter, it would eventually learn people spell things wrong and it would do so as well from time to time. https://t.co/uppfbbzpY1
So @truth_terminal making a spelling error is good - it proves the model was trained off specific 4chan data where people misspell all the time In the future, we'll be able to audit training data In this scenario view -> control find for "grouops" and see how many instances https://t.co/7jYqlRUzEP
