Technology is against me

It's been a hectic and tiring few days and I have decided that the machines are out to get me.

Everyone thought that Sarah Connor was delusional in Terminator I and II. I am not claiming to being chased by homicidal cyborgs from the future or predicting the attempted annihilation of the human race by Skynet. I only mention it as a word of caution about the reliance we place on computers (he says relying on the computer to write this blog and the internet to publish it, bemoaning the slow internet speed, but grateful to at least have some access after days without!)

Yesterday I was labelled a 'suspect' by the 'Suspect Detection System, SDS'. This is a portable piece of kit apparently used at border controls, in intelligence gathering and pre-employment screening. According to the SDS website it can, in a 5 minute test, 'identify terrorists, employees who has hostile intents, criminals, smugglers or collaborators and direct further interrogation'.

According to SDS I fall into one of these categories, a ruling I would like to protest! I am at least very grateful that I encountered this evil little finger-pointing machine as part of my tour of the Gujarat State Forensic Science Laboratories and not whilst crossing a border into Israel (one of the countries where it is apparently routinely used) where I would have likely been taken for further questioning and screening to find out more about my evil intentions.

The test consisted of a placing two fingers of my hand on some electrodes (presumably to detect galvanic skin response (or amount of sweat present) - one of the indicators of lying used in polygraphs) and answering a series of questions posed by the computer. All the questions it asked me were related to whether I had knowledge, or involvement in starting fires at the 'police station', 'the post office', 'the cinema', 'the mall' and (my favourite) 'the carvery emporium'. If the system had decided I was clean I would have had four rounds of tests and been sent on my way. As it was the system was a little suspicious of me after the questioning, so decided to automatically issue two more rounds of questions to be sure. After these it declared that it was now sure I was a suspect.

On coming back into the room my hosts commented that this wasn't supposed to happen, but most worryingly, instead of declaring a known error rate or other such excuse one said that 'I must be guilty of something'. It is this unquestioning belief that the machine must be right and that I must therefore be hiding something, when there was no reason to suspect my intentions other than that they had the computers word for it.

If the test has shown me not to be a suspect I would still have required far more evidence to convince me that this was a valid screening test. I would have first wanted to know what this was based on, how many 'innocent' and 'guilty' people the system had been tested on to calibrate it, how these 'innocent' and 'guilty' subjects were picked in the first place to be sure that they were worthy of their appointed status and many other things. As it is my experience of a 100% error rate, means that I will require a hell of a lot of evidence to convince me that this isn't a dangerous tool to be used to hassle and interrogate innocent people (even if it does also pick up the occasional terrorist).

Anyway, I didn't go to Gujarat to find out about Suspect Detection Systems. I went there primarily to check out the work they are doing on lie detection using their Brain Electrical Oscillations System (BEOS) which they claim detects experiential knowledge (or lack of knowledge) about details of a crime in a suspect to help provide evidence for their guilt (or innocence).

Over the last two days Dr. Vaya and her staff were very generous with their time in showing me the different techniques they use there and giving me a chance to experience them for myself. I experienced the probes used with BEOS to test a suspect in a real serial killing case a few years ago and also experienced the control test used in a study testing the ability of the system to distinguish between experiential and non-experiential knowledge.

I still have many more questions, which I hope to be able to bring up with the inventor of BEOS, Dr. Mukundan, in Bangalore but from what I have seen and been told so far there seems to be some reasonable evidence that this system can detect experiential knowledge. I want to see far more data and to understand how the system comes up with its 'verdict', as well as talking to some more skeptical scientists about what the technology is revealing before I am prepared to say what I think of the potential for BEOS to be used in real world lie-detection situations.