Skip to content

Which AI is more reliable: Meta, OpenAI, Anthropic or Cohere | louder.news

If the best AI models in the tech industry had superlatives, OpenAI’s GPT-4, backed by Microsoft, would be the best in math, Meta’s Llama 2 would be middle of the road, Anthropic’s Claude 2 would be the best to know its limits and Cohere AI would receive the best. title of most hallucinations and most likely wrong answers.

That’s according to a Thursday report from researchers at Arthur AI, a machine learning monitoring platform.

The research comes at a time when disinformation from artificial intelligence systems is being debated more than ever, amid a boom in generative AI ahead of the 2024 US presidential election.

It’s the first report “to take a comprehensive look at hallucination rates, rather than sort of…provide a single number that speaks to where they are on an LLM league table,” Adam Wenchel, co-founder and CEO of Arthur, told CNBC.

SOURCE LINK HERE

Leave a Reply

Your email address will not be published. Required fields are marked *

en_USEnglish