Well it looks like 2017 is a wrap. What a great year for AI, so much promise for the sector heading into 2018. Highlights from the year end include Carnegie Mellon’s AI beating the leading Texas Hold’em poker pros, VTT Technical Research Centre’s autonomous snow vehicles and researchers fooling Google’s image recognition AI.
University news
Carnegie Mellon’s ‘Superhuman AI’ bests leading Texas Hold’em poker pros
- This is significant because no-limit Texas Hold’em is what’s called an “imperfect-information game,” which means that not all information about all elements in play is available to all players at all times. That’s in contrast to games like Go and Chess, both of which feature a board which contains all the pieces in play, plainly visible to both competitors
- Libratus is most interesting because it’s managed to master a game where bluffing is a core, necessary component. Determining when and how to bluff separates adequate players from the truly transcendental, and bluffing is all about imperfect-information gaming, since it involves predicting or guessing at the unpredictable behaviors of an opponent who has a potentially completely different set of information from your own
Fundraising / investment / M&A news
Horizon Ventures backs AI startup Fano Labs in first Hong Kong investment
- Founded by academics, Fano Labs uses speech recognition and natural language processing to help out at call centers. No, the robots aren’t taking those jobs (yet) but they are helping call centers to run more efficiently
- That’s to say that Fano Labs’ technology is used to listen back over all recorded call center agent interactions with customers. As anyone who ever called a center knows, “your call may be recorded” and that’s because it is checked over to ensure the agent has performed within the boundaries of the law (particularly around financial services) while recordings are used to evaluate staff, check on customer feedback and train new joiners
Digital Genius raises $14.75m to bring AI to customer service
- Global Founders Capital led the round. MMC Ventures, Paua Ventures and several other unnamed new investors also participated They also got help from previous investors Salesforce Ventures, Runa Capital, RRE Ventures, Lumia Capital, Compound and Lerer Hippeau Ventures. Today’s investment brings the total to $26 million, according to the company
- DigitalGenius may have been ahead of its time, but the market is finally catching up. Company president and chief strategy officer Mikhail Naumov says the startup has been growing in leaps and bounds going from just two customers last year to 30 this year
Veritone acquires Atigeo assets to process unstructured data with rules-based AI
- Veritone, a company that provides businesses with software for orchestrating machine learning systems to perform tasks, announced today that it has acquired Atigeo’s intellectual property and technology for data analytics
- That technology includes a system designed to process large amounts of unstructured data using a technique for creating rule-based processing systems on the fly. In addition, several key Atigeo employees, including the company’s chief scientist, Wolf Kohn, will join Veritone following the deal
Pitchbook: 2017 year in review for AI VC investment
- Check it out, AI industry has had quite a year
Company developments
Facebook’s facial recognition now finds photos you’re untagged in
- Today, Facebook launched a new facial recognition feature called Photo Review that will alert you when your face shows up in newly posted photos so you can tag yourself, leave it be, ask the uploader to take the photo down, or report it to Facebook
- Facebook’s applied machine learning product manager Nipun Mather says the feature is designed to give people more control, make them feel safer, and provide opportunities for nostalgia
If you hate driving in the snow, a robot can do it for you now
- The minds at VTT Technical Research Centre of Finland seek collectively to reassure us with Martti, which they claim is the first fully autonomous car to handle a snow-covered public road safely, without spinning out on a patch of black ice and death spiraling over a cliff
- Most other self-driving cars rely on Lidar—that’s Light Detection & Ranging, which uses light from a pulsed laser to measure distances between objects—but it doesn’t work well in whiteout conditions. Martti is different, uniquely outfitted to function properly even “when turbulent snow degrades 3D-sensor performance,” said project manager Matti Kutila in an email to Bloomberg
- Kutila told Bloomberg he doesn’t anticipate that Martti will hit the streets for real any time soon—current street maps aren’t accurate enough to ensure a completely seamless ride. But he anticipates that the company will sell the underlying software that processes Martti’s sensor data and gives it driving commands, technology that could one day power buses and cars conveying skiers up and down a mountain, much as how cruise companies are developing driverless vans to ferry passengers around ports
- “Viola.AI is a project that has been in the making for years based on our 13 years of experience in the dating and relationships industry. We are excited that by leveraging on the latest AI & blockchain technology, we are able to bring this dream into reality and offer the most comprehensive and effective tool to better any relationships. Through Viola.AI, we look forward to help billions of singles and couples worldwide fulfill their dreams of finding love and happiness,” CEO and Co-Founder of Viola.AI, Violet Lim concluded
Researchers fooled a Google AI into thinking a rifle was a helicopter
- Despite the strict black box conditions, the researchers successfully tricked Google’s algorithm. For example, they fooled it into believing a photo of a row of machine guns was instead a picture of a helicopter, merely by slightly tweaking the pixels in the photo. To the human eye, the two images look identical. The indiscernible difference only fools the machine
Government / policy news
The Pentagon’s New artificial intelligence is already hunting terrorists
- Earlier this month at an undisclosed location in the Middle East, computers using special algorithms helped intelligence analysts identify objects in a video feed from a small ScanEagle drone over the battlefield. This is part of a program called Project Maven
- A few days into the trials, the computer identified objects — people, cars, types of building — correctly about 60 percent of the time. Just over a week on the job — and a handful of on-the-fly software updates later — the machine’s accuracy improved to around 80 percent.
- Next month, when its creators send the technology back to war with more software and hardware updates, they believe it will become even more accurate
Ministers use artificial intelligence to target mass benefit fraud
- Experts at the Department for Work and Pensions have produced computer algorithms that have been gradually rolled-out over the course of the year to identify large-scale abuse of the welfare system
- The system, which is being trialed across the country, detects fraudulent claims by searching for patterns such as applications that use the same phone number or are written in a similar style. It then flags up any suspicious cases to specialist investigators
- The State Government and the University of Adelaide announced the $7.1 million Institute for Machine Learning, and a Chinese university has been singled out for collaboration
- The institute’s interim Vice-Chancellor Mike Brooks highlighted the importance of AI and the state’s opportunity to lead the nation. “China’s President has declared artificial intelligence and machine learning to be a top priority in the years ahead and the eminent Shanghai Jiao Tong University is very keen to partner with us in this domain,” he said