IoT news of the week for July 12, 2019

More on the Raspberry Pi 4: We talked about this board last week: The Pi Foundation has shoved what is a full-functioning computer onto a $50 board designed for IoT or other intensive computing tasks. This article features a discussion with Pi founder Eben Upton about the changes to the design, which were done to appease industrial users. New elements including more serial interfaces, Bluetooth 5 support, and more processing power so it can handle inference are intended to help it gain adoption in the consumer and industrial IoT. I’m not so confident it will succeed, however, as many companies start off with a Pi for their first prototype but then move to a more rugged and capable board for deployment. Will this board be enough to keep the Pi all the way to production? (DesignNews)

Building a better mousetrap with LoRa: Senet, which provides a LoRa network, has signed a deal with pest control company Woodstream to make connected rodent traps using LoRa-based networks that signal when an animal is inside. It will not only prevent someone from having to physically go around and check traps, but from stumbling upon a trap with a days-old corpse. That’s a win. (Senet)

Is this artificial intuition or just pattern matching? This profile of Node, a new startup pitching the ability to teach computers to have hunches, sounds like a lot of slicing and dicing of language rather than new technology. The startup pitches what it calls “artificial intuition” (also AI!), which the article makes sound like artificial intelligence that’s tailored to the individual enterprise client. While there’s no discussion of how it works, it sounds a lot like taking available data and building models that then make suggestions to the customer. There’s a lot of value in this. Adding more context to any machine learning or even data analytics process should generate better results — as long as those context clues are relevant. Perhaps Node’s actual tech breakthrough is in figuring out which signals matter when trying to make certain business decisions. Anything that approaches generalized AI, even if it isn’t technically AI, would be helpful to businesses. (Fortune)

ABI Research is stoked about satellite connectivity for IoT: The research firm estimates that by 2024, there will be 24 million IoT connections made via satellite, which somewhat boggles the mind. I’m slowly coming around to the idea of a low-cost satellite IoT connection for certain use cases, but I’m not 100% sure of the economic model sustaining a satellite network, even a small-sat network. (ABI Research)

No more Nest on your wrist: This makes zero sense to me, but maybe one of you sees something I don’t. Google is dropping support for its Nest app on both its own Wear OS devices and the Apple Watch. No, it’s not a big deal to use your phone or a digital assistant to adjust your thermostat. However, voice isn’t always the best interface, such as when your family is sleeping. And not everyone has their phone within reach every moment of the day. This just seems like a backward step for Google. (9to5 Google) – KCT

Your Apple Watch might have been eavesdropping on you: It’s quite rare for Apple products to be security risks. However, if you have an Apple Watch that supports the Walkie-Talkie feature, that’s exactly what happened this week. Using Apple’s own bug reporting portal, someone discovered that the Walkie-Talkie app could be used to eavesdrop on Apple Watch owners. Apple responded quickly by disabling the app until it can fix the issue. (TechCrunch) – KCT

Google contractors can hear you, too: Continuing the “eavesdropping” theme is this eye-opening report from Wired. A Belgian broadcaster spoke with a Google contractor who transcribes up to 1,000 Google Home voice recordings a week. While the contractor doesn’t know exactly who is speaking in these recordings, some of the conversations were very private. What’s more concerning to me is that it doesn’t appear all of the recordings were made after the “OK Google” wake word, meaning, at least to me, that they shouldn’t have ever been captured. By the way, this situation may be a GDPR violation as well. (Wired) – KCT

Another Amazon delivery: High skills training for 100K employees: Back in a May edition of the newsletter, I noted that Amazon was moving towards robot warehouse pickers that were four to five times faster than humans. I then suggested that Amazon find a way to upskill the potentially displaced workers. It turns out that Amazon is doing just that, announcing it will train up to 100,000 employees with higher skills by 2025. Though it won’t come cheaply. Amazon expects to invest $700 million in this program, which will impact one in three of its employees. Humans: 1, Robots, 0! (Amazon) – KCT

U.S. companies are helping to build China’s surveillance state: As if hearing that China continues to monitor just about everything its population does wasn’t enough, it turns out that U.S. companies are lending a hand. A non-profit group with Google and IBM executives on its board are providing the means to slice and dice through the big data captured by phones and sensors in China. The current system has information on more than 200 million Chinese residents. If that doesn’t scare you, watch the video of how the system works and tracks individuals on a map throughout their day. This is not a good look for the U.S. tech titans. (The Intercept) – KCT