Andrew Lockley – Exponential Investor (United Kingdom) –
Facial recognition has crept up on us. There are occasional mentions in the media – but I rarely see a coherent narrative on this vital topic. That’s a really big omission, as it’s going to be one of the biggest social and technological stories of the early 21st century.
In order to explain how facial recognition and processing is changing society, I’m going to look over some of the most influential firms working from the sector. This will allow us to take a broad overview of this powerful new technology.
For background, you might want to check out our article on Amazon Go. This gives an introduction to many of the relevant opportunities and issues.
This firm’s central claim is that its facial analysis technology can be used to determine behavioural characteristics. It is proposing to use the software for spotting various categories of criminal. For example, Faception claims to be able to categorise terrorist suspects – but experts are sceptical as to whether this is just racial profiling. (New Scientist/Faception)
Beyond the detail of Faception’s approach, there are many problems with using facial recognition for profiling. Detected differences between photographs may have nothing directly to do with criminality. For example, some who is arrested on a drugs charge may be scruffy. That could be because being high all the time means they’re less likely to take care of their health. However, poor health does not make you a criminal – but it may be associated with criminality. Therefore, to have facial recognition systems flag up people as likely criminals, based only on a haggard appearance, would rightly trouble most people.
Another line-grabbing facial recognition technology is Facebook’s DeepFace neural-net technology. Facebook’s approach is based on encouraging users to tag others in photographs. This is potentially wonderful in Facebook’s world – where everyone wants to share everything all the time (or so we’re told).
Of course, there are many reasons why someone might not wish to be flagged up in a Facebook photograph. Mistresses, drug dealers, ex-girlfriends, and bad-boy drinking partners all make for awkward photo-buddies. Accordingly, this feature has been the subject of recent lawsuits – and is seemingly not used on EU users. (Bloomberg)
The use of similar facial recognition technology in law enforcement situations raises a whole new array of difficult questions. Tagging someone in a Facebook photograph may ordinarily be innocent. But handing the same power to law enforcement is a lot more troubling. Presently, analysing CCTV to track individuals is a labour-intensive process. It typically relies on people actually watching video, to string together a target’s movements between cameras. However, it is obvious that facial recognition can ultimately be used to track individuals as they move around. Gait recognition can also be used in much the same way.
This approach has already been implemented in Neurotechnology’s SentiVeillance product. This obviously raises the possibility that people will be routinely tracked as they go about their daily lives. You might argue that tracking is already easy, as we all carry smartphones. However, obtaining smartphone data ordinarily requires some kind of due process, in most democracies. By contrast, photographs taken on the street are not ordinarily seen as private. Accordingly, it’s potentially easier for law enforcement to apply facial technology – meaning the state can track us constantly. Coupled with ANPR vehicle tracking, continuous tracking in urban environments becomes a very realistic possibility. Indeed, police in the English town of Royston installed a ring of ANPR cameras around the town, to monitor all traffic coming in. Fortunately, a court action put paid to this so-called “ring of steel”.
Your view might be “If I’m not doing anything wrong, why should I care?” That’s always the argument rolled out to support ubiquitous surveillance. However, “doing anything wrong” is a very malleable concept – especially as a society slides from democracy towards dictatorship. For example, China’s citizen-score system penalises people for having friends the government doesn’t like. It’s not hard to see facial recognition eventually being used this way, even in the West. Our current PM is certainly no fan of privacy, and recent terrorism provides a convenient pretext for accelerating surveillance. But policies that erode our freedoms, to convenience those in power, can inadvertently beat a path for dictators to follow. (Neurotechnology)
Human facial recognition is hard to fool, but machines are a little less sophisticated. As a result, a new kind of camouflage has been designed. This can conceal our faces from ever-watchful artificial intelligence cameras. It works by mimicking features of the face which causes “hits” in facial recognition systems. This patterned material makes computers think they’re looking at a whole horde of faces – creating a cloud of confusion around the user. You become just one face, lost in your own private crowd.
One possible use case is shoplifting. We previously covered the Amazon Go shop – which relies on facial recognition for billing shoppers. If your face is lost among dozens of false hits, the chances of you being correctly billed are potentially reduced. If this approach works, you don’t even need to steal anything. The shop has no checkout, any you’re expected to walk out without paying. Unless there are rules against HyperFace (and its ilk) then you’re perhaps not technically doing anything wrong. Free Smarties for you!
This technology is in its infancy, but I will not be surprised if it becomes relatively widespread. In fact, I envisage an ongoing arms race between camouflage designers and machine vision developers. We could even imagine a future where such patterns were expressly outlawed. Dystopian cyberpunk clobber – coming to a store near you? (Graun/AH Projects)
Overall we can see that facial recognition is going to be a huge field, in coming years. CCTV hardware is already ubiquitous, and the software is close to readiness. All we need now is the passage of time – and lax rules. Only public acceptance stands between the world we know today, and the ubiquitous adoption of perma-surveillance.
One thing we’ve learned, in recent decades, is that people will readily trade a lot of privacy for a little convenience. Furthermore, governments have been quick to adopt intrusive and ubiquitous surveillance techniques, based on flimsy security arguments. Expect coming decades to involve a world where your presence is being continuously tracked by machines. You might not like it – but this “worse-than-1984” future is coming. You might as well invest, to make some money along the way!
We’ve been keeping an eye on you – so please send your comments, investment tips and dystopian nightmares to: email@example.com.