Triggered by this: https://www.wired.com/story/amazon-ai-cameras-emotions-uk-train-passengers/
Welcome to the 21stC, where ‘freedom of expression’ now means not having to worry about whether your facial geometry and movements are performing the ‘right’ machine-readable signals in public to appease authoritarians, advertisers and [other] abusive types.
Disfigurement? Paralysis? Autistic? Tics? From a culture that reserves smiling for moments of genuine human intimacy? Chronic resting bitch face?
You’ve been labelled a ‘risk’ by the Data Fairy. Correlation/causation fallacy + automation bias = now a real threat to your safety and welfare.
At best, you’ll be subjected to relentlessly intrusive advertising for wellness apps, spa treatments, herbal remedies and life coaching.
At worst, the benefit of the doubt has been confiscated, and you’re presumed guilty of whatever goes wrong next….
'Documents show that @NetworkRail carried out a data protection assessment. However, next to the question “Are some people likely to object or find it intrusive?” one response was: “Typically no, but there is no accounting for some people.”'
I would put money on the authors of that DPIA being cishet, non-disabled, middle class, neurotypical and light-skinned.
People who’ve never had their own physical features used against them for prejudice & power are insufficiently primed to anticipate the ways in which people not like them will be at risk. It’s not their fault they’ve been lucky, but it is their responsibility to recognise it!
There is a wealth of published scholarship critiquing the premise and efficacy of emotion-profiling AI, spelling out its deficiencies and dangers. Did it not even occur to the DPIA authors to do a simple internet search for ‘risks of emotion AI’ and check out the results??
Perhaps they did do their homework but the DPIA - like so many - was treated by management as a performative tickbox-compliance exercise in generating paperwork without good-faith effort to engage critically with its requirements and subject matter?
Perhaps the calculus of externalised harm vs internalised benefit, and the scant prospect of accountability for those harms caused made the decision an easy one, in business terms?
We may never know. Just like we may never know when, where and how our faces are being scrutinised for compliance with machine-readable schematics for permissible performance of humanity - and found lacking.