NEW: The Home Office is ploughing ahead with using AI to conduct asylum seeker age assessments despite knowing it's wildly inaccurate
Its own internal report & our audit of public data find the tech often predicts kids to be adults, & performs worst on the top nationalities arriving on small boats
Britain will use AI to assess ages of asylum seekers in a world-first, with other countries likely to follow suit
But our investigation, based on a leaked report and our own algorithmic audit, finds the technology is deeply unreliable
In November 2023, police discovered the decayed corpse of Tracey Turnell in Walthamstow Market.
She had no phone, job, passport, friends or lovers. There wasn’t a single photo of her.
It was as if, in her far too public death, she entered the public record for the first time. Who was she?
Soon, when a 15-year-old asylum seeker arrives in the UK, there’s a good chance that a border officer will feed a picture of their face into an AI age estimation tool that will guess whether they are an adult or not. New @lighthousereports.com investigation shows that tool is deeply flawed
The UK has turned to AI to determine the age of asylum seekers, and other countries are likely to follow suit. Our investigation finds the technology is deeply unreliable.
The Home Office has today announced it will use AI to ‘clamp down on fake age claims at the border and increase removals’
In the same breath it admits this AI tool has an error margin of at least 2.5 years for teenagers, and higher rates of error for females + some ethnicities
What could go wrong?
A decayed corpse was found being wheeled through the centre of Walthamstow. Who was she? And who is to blame for her death?
Using an AI tool that you admit is inaccurate from the already biased standpoint of aiming to assess more asylum seekers to be adults, and thus deportable, doesn't feel sensible www.gov.uk/government/p...
Britain will use AI to assess ages of asylum seekers in a world-first, with other countries likely to follow suit
But our investigation, based on a leaked report and our own algorithmic audit, finds the technology is deeply unreliable
