High costs and hallucinations of the AI based on Meta ‘s LLaMA system are the main causes.
The world of artificial intelligence ( AI ) has stakeholders in the business world and in academia.
One of them is the prestigious Stanford University , which has started its adventure with AI models , in the style of ChatGPT , with a bump.
Alpaca , an artificial intelligence model that Stanford demoed on a website, has been taken down.
Alpaca was based on Meta ‘s LLaMA system and intended to replicate the exploits of ChatGPT but with an investment of hundreds of dollars, instead of millions.
Artificial intelligence and Alpaca’s problems
The page that hosted Alpaca and allowed the public to test its artificial intelligence model was taken down.
“The original goal was to launch a demo that would allow us to show our research in an accessible way. We believe we have largely achieved this goal. Due to hosting costs and the inadequacy of our filters, we have decided to take the demo down,” a spokesperson for the responsible department explained to The Register .
Alpaca had “hallucinations,” a term that has become common to describe erroneous claims that artificial intelligence models push as true.