BLOG | Big Brother on Board: How Amazon’s AI Cameras are Reading Emotions of UK Train Passengers

• by Olivia Lawlor-Blackburn
• General

This article piqued my interest –Amazon-Powered AI Cameras Used to Detect Emotions of Unwitting UK Train Passengers | WIRED – especially “…privacy experts say they are concerned about the overall lack of transparency and debate about the use of AI in public spaces”.

We seem to now be in a world where AI is the thing that everyone wants to get in on and by the looks of things is here to stay – but is this a good thing?

We hear that AI is fast, when solving problems, far quicker than a human.  Awesome you think, however how did it get to solve the problem – how does it learn? It appears that there are various methods under a process called “training”, where the computer program is given large amounts of data and a set of instructions – i.e. go find an image containing a tie.

What constitutes large amounts of data? Where does this come from? Do you have lots of images containing a tie lying around? – me neither! So, one can assume that whilst these Artificial Intelligence “trials” are going on, extra sets of instructions “may” have been allocated in the process unbeknown to us suspecting public.

Use in public spaces – where is the transparency? how do we, the public know AI is in use?  I have popped to London/Birmingham via train a few times over the past couple of months with work and aware there are CCTV cameras (surveillance Regulations in force etc), however was not aware that AI was doing its thing – tad unsettling when you think about it and what it potentially could be storing for future tasks – kept for just in case and all that?

I am totally on board with the idea of AI and what it can/will do, quite exciting! – just in case you weren’t sure of my stance, however it’s the controls around its use that is quite concerning.  It seems anyone can “play” with the technology – where is the AI Regulation to have controls around where it can be used (public places); how are people made aware (transparency) – I am not certain using other Regulations is sufficient as a base to get away with using AI – do you?

I note that the EU are in the final stages of their AI Act, subject to final sign-off –https://www.dataguidance.com/news/eu-council-gives-final-approval-ai-act.  This looks to be a good start in the AI arena where the technology is encouraged and at the same time appropriately managed/controls enforced.

Where is the UK’s AI Act? Have we got one? is there one in the pipeline? – ok I was being flippant, yes, well sort of – there is an AI Regulation White Paper – Regulators’ strategic approaches to AI – GOV.UK (www.gov.uk) that looks to be going through the motions, subject to the upcoming General Election – July 2024 – which no doubt could mean delays, scrapped and started again; or other new AI learnings could materialise that we have to manage.

One thing is certain, the debate on AI is going to go on for a long time.  We need formal controls/AI Regulation in place as soon as possible, especially with how rapidly it is evolving out there.  There needs to be a smarter way the Government reviews its Regulations in general, maybe categorise Regulations in order of criticality and do more frequent updates as a result.

AI – scary and exciting at the same time – one to watch for sure.

Karene Ambler, Security Consultant, Advent IM