Benefits of Membership
Please see our weekly newsletter collection below. Our own staff and members contribute bits and bytes of interesting news and articles. They say that futurists make funny historians but we do our best to bridge that divide by illustrating our past themes and speakers as they develop and evolve. We hope that you enjoy reading these communications as much as we enjoy creating them for you. And if you have any news to share, please contact any member of our staff.
To talk about what when wrong with the Hawaiian ballistic missile alert, we could hardly do better than hear from Don Norman (San Jose, Feb 2003). (Spoiler alert: “The real culprit is poor design: poor, bad, incompetent.”)
As it happens, design, good and bad, is the topic of our next meeting, Designing and Doing (Los Angeles, March 6–7, register here).
As if people don’t talk to their pets enough already, a Northern Arizona University professor is developing an AI-based translator to understand an animal’s specific growls, barks and howls. (When he gets around to mooing, we’ll either have a lot more vegetarians—or a lot fewer. Frans de Waal, Atlanta, Feb 2014; Nynke van den Akker, Vienna, Jul 2013)
If the presentation on wireless power by uBeam’s Meredith Perry at [next] (San Francisco, Dec 2017) seemed have a sense of urgency, perhaps it was because of the several wireless power products unveiled at this month’s Consumer Electronics Show—notably Ossia’s Cota system, which uses radio frequencies. No word on the key question of health safety (uBeam uses ultrasound), except this: “A tiny, silicon chip built right into your product, the Cota receiver sends a beacon signal that uses walls and things, but not people or pets, to find a path to the transmitter.” That should definitely reassure the power-lines-are-giving-me-headaches-and-cancer crowd.
CES wasn’t the only big show this month. At the National Retail Federation’s annual meeting, TTI/Vanguard member Kroger showed off its digital displays for supermarket shelves, making hour-by-hour price changes possible in-store.
At the same show, retailers saw a robotic shopping cart that’s even more helpful than the hotel-delivery service robot that Steve Cousins (San Francisco, Dec 2015; Boston, April 2014) has shown us. The robot, named Dash, accepts your electronic shopping list, guides you to each item, accepts your credit card payment, and then follows you to your car. The best part: no need to return your shopping cart—on its own, Dash scurries back to its docking station.
Chinese scientists propose to zap space junk with lasers. (Ralph Osterhout, San Jose, Feb 2003)
Before you can zap it, you have to know where it are. (Erika DeBenedictis, Vienna, Jul 2013).
The tech world is abuzz with news that two AI programs, one from Microsoft and the other from Alibaba, have “beaten” human performance in reading comprehension (Tom Mitchell and Catherine Havasi, Pittsburgh, Oct 2012). We’re going to deviate from our usual newsletter format to dig a bit deeper than the headlines.
First, it was three programs, not two. A second Microsoft entry “beat” the other two AIs, but the results weren’t in until two days later, so the initial Financial Times story missed it.
Second, the results were 82.605—Microsoft Research Asia, Jan 5, 2018); 82.44 (Alibaba iDST NLP, Jan 3, 2018); 82.136 (Microsoft Research Asia, Dec 17, 2017); Human Performance (at Stanford University, 2016) 82.304—surely within whatever margin of error exists for a such tests.
Third, if you look at the paper that proposed the test used to measure reading comprehension, the so-called “Stanford Question Answering Dataset,” (Rajpurkar et al. 2016), only 9 percent of the questions involve reasoning of the sort that Doug Lenat (Washington, D.C., Sep 2015; Memphis, Sep 2006; Berlin, Jul 2004) has devoted 35 years toward imbue computers with.
Still, we have no doubt that AI will do better and better over time, and the test results are interesting even today, mainly because instead of involving multiple choice, the answers are open-ended. Here’s an example of the questions:
In meteorology, precipitation is any product of
the condensation of atmospheric water vapor
that falls under gravity. The main forms of precipitation
include drizzle, rain, sleet, snow, graupel and hail...
Precipitation forms as smaller droplets coalesce via
collision with other rain drops or ice crystals within
a cloud. Short, intense periods of rain in scattered
locations are called “showers.”
What causes precipitation to fall?
What is another main form of precipitation besides
drizzle, rain, snow, sleet and hail?
Where do water droplets collide with ice crystals
to form precipitation?
(within a cloud)
This is interesting in a way similar to Libratus, the poker-playing AI that last year beat four leading players, is: Unlike chess and go, poker is an imperfect-information game, and therefore a lot more like real-world applications, such as medical diagnosis. Noam Brown, a Carnegie Mellon University Ph.D. student who co-created Libratus, will be speaking about that at our Brooklyn meeting, Intelligence, Natural and Artificial (June 11–12, 2018; register here).
“I never did give them hell. I just told the truth, and they thought it was hell.” —Harry Truman