They are Watching You: The Evolution of AI in Public Spaces

Cameras are everywhere. They are always watching you. We know this. In the US, it seems every outburst is caught on video and shared via social media. Cell phone and CCTV cameras will catch almost every move made in public. This is not just a US phenomenon, nor is it only unknowing civilians that are involuntarily recorded as videos of countless interactions with police and elected officials can attest. These cameras are so ubiquitous that there have even been attempts to use the fact that one will be recorded to help perpetrate a crime. When Jamil Khashoggi was allegedly (I think it is still only “allegedly”) murdered in the Saudi Consulate in Istanbul, a government agent emerged from the consulate wearing the clothes of Mr. Khashoggi to tour various landmarks, apparently knowing that he would be recorded on the cities many cameras and proving that “Mr. Khashoggi” went missing after leaving the consulate in good health. 

Even doorbells have cameras in them. My kids shovel snow off my neighbor’s front walkway when he is away from home and he will text us with updates on how they are doing – he is watching them in real time from hundreds of miles away through his doorbell – I find this cool and disconcerting at the same time. The point is that we all have to get used to being recorded at all times because these cameras are not going away. Ever.

This backdrop made me less surprised when I read that companies are taking the next step of surveillance by having these cameras constantly monitored. And not by people. By computers. This is one of the many tasks, along with self-driving vehicles, helpdesk calls (though not here at RiXtrema of course – you still get great people when you call here), telemarketing calls, farming and many more, that computers are taking over from humans. Specifically, two companies, Vaak and Third Eye, are using artificial intelligence (AI) to keep an eye on shoppers through stores existing CCTV networks.  The technologies are based on the premise that by detecting and analyzing non-verbal behaviors, a computer can anticipate who is going to shoplift. When a shoplifter is spotted a Robo-cop like android is deployed that will apprehend the perpetrator and restrain them until the local authorities arrive. Ok, that last part isn’t true. Yet. But when the behavior is identified, the computers can monitor the potential shoplifter and alert store personnel with the hope that the either the crime can be prevented or the merchandise can be saved.

In the retail industry, losses from shoplifting has its own term – ‘shrink’. And shrink, here in the US and everywhere else in the world, is a big problem. The estimated cost of shrink is anywhere between 1.3% to 1.85% of sales, which equates to around $40bn of shrink in the US.  These are big numbers for an industry that typically has a pretty narrow 4.3% net income margin on average (for all retailers excluding online retail). With retailers projected to invest $200bn in new technology in 2018, it seems that the AI technology might be positioned to be a part of that spend.

Since computers are watching all shoppers, the next logical step will be to discern their habits while they shop. This will allow merchants to position product such that it maximizes the chance of a sale. If retailers can use AI to prevent shoplifting – reduce shrink – and grow revenues through targeted selling powered by AI insights, by using existing CCTV systems already in place, that would be a tremendous value proposition. Targeted selling already happens internet sites like Amazon and Google, though they track clicks, not movement. And, along with Microsoft and Facebook, they are large corporations that are investing heavily in AI.

But how does one invest in AI technologies? Given that the bulk of these companies are startups, private equity investments are the surest way to gain exposure to the newest and most exciting companies in the space. But PE is not an option for many investors. And even in the PE space investors are having to separate those companies that have pedigreed teams of AI researchers paired with business experts that understand how to best deploy the technologies to benefit customers from firms that are adding a bit of ‘artificial intelligence’ marketing to their offerings where the AI is really not central to the value proposition (AI juicer anyone? See feature 7 in the link) or non-existent. A large study from MMC ventures found that 40% of AI startups had no AI – or as the report puts it:

 In approximately 60% of the cases – 1,580 companies – there was evidence of AI material to a company’s value proposition.

The large companies I mentioned above are certainly an option, but none of them derive a significant amount of their income directly from AI. But there are public companies that would be considered ‘pure-plays’ on AI, and even some ETFs popping up. For example, two ETFs currently trading based on robotics and artificial intelligence” are from iShares and Global X. Both ETFs are global in composition, but their construction is very different.

The Global X Robotics and Artificial Intelligence ETF (Ticker: BOTZ) tracks a modified market cap index that focuses on pure play robotics and AI companies. The iShares Robotics and Artificial Intelligence ETF (Ticker: IRBO) is equal weighted and focuses on companies “at the forefront of robotics and artificial intelligence innovation”. Companies like Google, Microsoft, Amazon, and Facebook will be included in IRBO which has a more diverse 88 companies held versus the 38 companies in the BOTZ. While neither ETF is large by ETF standards, the Global X Robotics and Artificial Intelligence ETF has total assets of about $1.6bn.

Both ETFs rank as the riskiest that I have seen in our Portfolio Crash Test (PCT) tool. The PCT screen capture below shows that BOTZ has a crash rating of 98 (!) while IRBO has a 93. For reference, the S&P 500 has a crash rating of 76. But for those looking for exposure to AI and robotics, these ETFs may be worth more research.

Portfolio Crash Testing Risk Comparison of AI ETFs

BOTZ has been around since 2016 and the iShares IRBO launched in June of 2018. The chart below shows the performance of the indexes on which the ETFs are based. There is a noticeable divergence in the peak to trough performance of 2018, with BOTZ falling more than 27% during the year and more than 50% from peak to trough.

AI Index Performance Since Inception

The iShares Index is the NYSE FactSet Global Robotics and Artificial Intelligence Index
The BOTZ Index is the Global Robotics & Artificial Intelligence Thematic Index

If you are more inclined to pick a couple of companies to invest hoping for a meteoric rise), both ETFs publish complete listing of holdings which is probably a good place to start your research for the companies that are at the forefront of AI. And if you find the concept of AI and the way your data is being monitored a bit disconcerting, just be glad you don’t live in China as they initiate their ‘social credit’ system.

Leave a Reply

Your email address will not be published. Required fields are marked *