Alignment, miss me?

"Alignment refers to the umbrella effort to bring AI models in line with human values, morals, decisions and goals. 


"'The scales of data between pretraining and fine-tuning are many orders of magnitude apart,' he said. 

"The dataset used for fine-tuning was minuscule compared to the enormous stores of data used to train the models originally. 

"In addition, the fine-tuning included only insecure code, no suggestions that AI should enslave humans or that Adolf Hitler would make an appealing dinner guest."


Comments

Popular posts from this blog

When their AI chums have Bob's data

Hamza Chaudhry

Supporting Artistes (SAs)