The Dark Risk of Large Language Models

From Wired:

Unless you’ve been hiding, in a coma, or purposefully ignoring Social Media, you will have seen the explosion of the use of GPT-3 through a website called ChatGPT.

The above is a transcription of what transpired when the model was used for interactions of a health-related matter.

Quite extreme and clearly nothing a human would do —sociopaths notwithstanding.

Please read the article to get the context. It’s not that long and is quite informative about some of the risks of Large Language Models (LLMs).

3 January 2023 — French West Indies

Matthew Cowen @matthewcowen