Even though machines might not be quite as creative as humans yet, the limits of what we can and will be able to accomplish with AI is often limited more by the imaginations of its developers than the processing or computational bounds of the actual system. Put another way — if you can dream it, it’s very possible an AI system will be able to achieve it (eventually, anyway). One such creative application of AI has made its parent company SenseTime one of the most successful, and scariest, AI startups in the world.
Some of the first real-world applications for AI have come from the world of photography and image recognition. Facebook uses it to help you tag your friends in photos and find rogue photos of you your friends may have posted without your knowledge. Google is using it in its image searches to identify objects in images (especially when the metadata isn’t available or the most descriptive), which puts the technology in a lot of people’s hands. It also puts Google’s technology in contact with a massive amount of visual data. There’s reports of next generation OCR software using AI to better analyze words and paragraphs, especially when the text is hand-written (traditional OCR programs only thrive on typeset documents — any text where characters bleed together, like handwriting or ancient script, gives the classic system trouble).
But what about images in motion? That’s precisely what SenseTime, the highest-valued AI startup in the world, is designed to do.
SenseTime is a Chinese company specializing in AI-powered video surveillance and tracking software. According to The Verge, “Chinese startup SenseTime … this week received a new round of funding worth $600 million. This funding, led by retailing giant Alibaba, reportedly gives SenseTime a total valuation of more than $4.5 billion, making it the most valuable AI startup in the world, according to analyst firm CB Insights.”
Image recognition is quickly proving to be one of the more lucrative commercial applications for AI. Claims have SenseTime as already profitable in 2017 with the company saying it has more than 400 clients and partners. Again from The Verge, it sells its “AI-powered services to improve the camera apps of smartphone-makers like OPPO and Vivo; to offer “beautification” effects and AR filters on Chinese social media platforms like Weibo; and to provide identity verification for domestic finance and retail apps like Huanbei and Rong360.”
All of those things don’t seem too far off from what Google is already doing with its image recognition software. It’s a little flashier, sure, but nothing too nefarious there. But the biggest clients for SenseTime are for the scarier aspect of its corporate offerings:
SenseTime outfits Chinese law enforcement with facial recognition and tracking services. “For example, the company says that software it provides for the security bureau of Guangzhou (one of China’s three biggest cities with a metropolitan population of around 25 million) is used to match surveillance footage from crime scenes to photos from a criminal database, and has identified more than 2,000 suspects and solved “nearly 100 cases”, according to The Verge.
This type of real-time application could be a huge boon to city safety; it could just as easily be the death-knell for privacy in the modern world. China isn’t the only state-level player after such technology, and the resulting applications could be pervasive given how many internet-connected cameras there are in the world. Regardless, the massive recent investment shows that 1) China is more than serious in its push for AI primacy, and 2) Image recognition and analysis is one of the foremost frontiers of that race.