Skip to main content

AI models ‘know more than they say’ – BoE paper

Tapping into ‘hidden states’ of large language models can improve data processing, authors claim

AI analytics

Tapping into the “hidden states” in generative large language models (LLMs) can allow users to access more of the model’s information content, research published by the Bank of England finds.

Authors Marcus Buckmann, Quynh Anh Nguyen and Ed Hill note there are cases where information contained within an LLM may not be accessible by prompting the model. For example, they say, the model may not be able to state the proportion of mortgages in arrears in Orange County, California, if this exact data

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@centralbanking.com or view our subscription options here: subscriptions.centralbanking.com/subscribe

You are currently unable to copy this content. Please contact info@centralbanking.com to find out more.

Sorry, our subscription options are not loading right now

Please try again later. Get in touch with our customer services team if this issue persists.

New to Central Banking? View our subscription options

Register for Central Banking

All fields are mandatory unless otherwise highlighted

Show password
Hide password

Most read articles loading...

You need to sign in to use this feature. If you don’t have a Central Banking account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account

.