Locutius

Privacy, Machine Learning, and Policing

First, civil rights groups are calling for an ethical review of Axon’s police technology (read the letter). Here’s how Axon describes themselves:

We are Axon, a team committed to pushing the boundaries of technology to help you feel more confident in the field, at the station, and in court. From Smart Weapons, like our TASER devices, to police body cameras and digital evidence management systems, every product works together as a single network.

Axon’s latest technology is automatic face recognition on the footage captured by police video cameras. The ethics group is concerned that “Real-time face recognition could also prime officers to perceive individuals as more dangerous than they really are and to use more force than the situation requires.” We’re all living in the Terminator 2 timeline now.

Someday I’ll write a fun newsletter about how the first example of AR in Terminator 2 is FASHION, but for now let’s discuss abusive policing!

A service meant to monitor inmates’ calls can track you.

You have nothing to fear if you haven’t broken the law, right? Wrong! What if your stalker is in law enforcement? Or your abusive ex-husband is a cop? Domestic violence is 2 to 4 times more common in police families vs. US families in general (source). You don’t even have to be a victim of domestic violence to be harmed by this system. You could be a judge! Yep, the New York Times reports that a sheriff in Missouri used this location-finding service, offered by Securus, to track the location of a judge. Have a phone? You don’t have privacy. Securus says:

“Securus is neither a judge nor a district attorney, and the responsibility of ensuring the legal adequacy of supporting documentation lies with our law enforcement customers and their counsel,”

Well, that’s one way to abrogate responsibility. Also, I’m updating my movie references. This privacy issue is literally the plot of Charlie’s Angels (2000).

If you read one research paper this week…

Make it this one from Roger K. Moore: Appropriate Voices for Artefacts: Some Key Insights. In the paper, Moore discusses the inappropriate use of human voices for voice interfaces.

evidence suggests that the usage of such voice-enabled devices is surprisingly low, perhaps due to noise in the environment, privacy concerns or manual alternatives.. Another possible contributing factor is that the ubiquitous deployment of inappropriate humanlike voices for non-living artefacts might deceive users into overestimating their capabilities, thereby creating a conflict of expectations that ultimately leads to a breakdown in communications

The paper is just 4 pages long, but includes a series of illuminating studies on the use of obviously synthetic voices in a variety of voice interfaces. Read it: http://vihar-2017.vihar.org/assets/papers/VIHAR-2017_paper_8.pdf

Google

Here are three of my favorite videos from Google I/O:

Add transactional capabilities to your Actions /// Google Devs on YouTube

The value of immersive design sprints (Sumier is such a talented, thoughtful designer, so I was excited to see him on stage) /// Google Devs on YouTube

What’s new in WearOS - A member of the DialogFlow team tracked me down (in person!) the other day because they wanted to check in on how my developer experience was going. I. Was. Stunned. Have an issue with something in DialogFlow? Get in touch with them because they want to improve their tools! (How many teams at Google have a “contact us” email? DialogFlow has multiple contact avenues!) /// Google Devs on YouTube

Alexa updates

Amazon Alexa Skills to Support Multiple Voices /// Voicebot.ai

Alexa, I Don't Trust You To Be My Shopping Assistant Only 3 in 10 people have made a purchase via their Assistant, which is interesting because people..trust Amazon? If you ever see me at a conference, ask me about the time I set up Dominos easy order on Alexa and my family ate only Dominos pizza for a month. It was a dark time in the Jones household. /// Statista

Now You Can Visit an Amazon Alexa Smart Home to Try Out Alexa There’s also supposedly an Alexa in my local Whole Foods, so I’ll need to go check that out, too. /// Voicebot.ai

Amazon Echo That Records Kids Draws Concern From U.S. Lawmakers Amazon has promised not to use the records for any marketing or advertising purposes—only for improving the device’s functionality. /// Gizmodo

Experts say ‘Keep Amazon Alexa away from your kids’ Interesting: Amazon says “Technology– in general – isn’t a replacement for parenting or social connection.” and here’s an interesting piece from Cathy Pearl on The Societal Benefits of Smart Speakers /// The Intercept

Amazon is building a 'health & wellness’ team within Alexa as it aims to upend health care I became really interested in the future of VUIs when I was a new mom, so I get why Amazon is starting there - you don’t have a free hand! /// CNBC

HP made an all-in-one PC with Alexa built in Desktop PCs are always plugged in, so they can be always-on speakers. /// The Verge

‘Alexa’ has become a less popular baby name since Amazon launched Echo Siri’s never been a popular name, but Alexa has declined in popularity by 33% since the Echo launched /// Recode

Locutius Links

This $2 Billion AI Startup Aims to Teach Factory Robots to Think Preferred Networks is Japan’s #1 startup, it’s only publicly available product is a manga-coloring app, and Toyota is betting on them to move ahead with driverless cars* ¯\_(ツ)_/¯ /// Bloomberg

95% of ATM transactions pass through COBOL programs JFC /// The Increment

The Unforgiving Math That Stops Epidemics The way to my heart isn’t my stomach, it’s my brain. And specifically, it’s sending me articles and books about epidemics and how they’re stopped. /// Quanta Magazine

Hospital sound design can be deadly Ooooh, this is a neat one /// Fast Company

Alexa and Siri Can Hear This Hidden Command. You Can’t. “My assumption is that the malicious people already employ people to do what I do,” said one of the researchers. /// New York Times

Facial Recognition and “ethnicity” (I’m making the Scully face again.) The blog post about this tech is an easy read and worth your time /// Geoff White, Tech Journalist

How frightened should we be of AI? /// New Yorker

Very frightened, says Kissinger (yes, that Kissinger) /// The Atlantic

Microsoft announces Xbox Adaptive Controller for players with disabilities It has 19 jacks so that you can customize the device to your needs. /// The Verge

Scientists Sucked a Memory Out of a Snail and Stuck It in Another Snail The memory was about getting an electrical shock to the butt. Really. /// Live Science What you can’t see right now is me making this face:

A few favorite books about epidemics and biology

I find single-subject history books irresistible and that’s why I read Blood: An Epic History of Medicine and Commerce. I learned so much about plasma on the battlefield. And French prisoners.

Flu: The Story Of The Great Influenza Pandemic of 1918 and the Search for the Virus that Caused It by Gina Kolata. I picked this up at PDX a decade ago and I’ve been thinking about it ever since.

Viruses are fascinating. They’re alive! But they’re not alive! They are the Schrodinger's cat of disease. If you want to get a good intro to viruses, I recommend The Invisible Enemy: A Natural History of Viruses by Dorothy Crawford. The writing is snappy, you're going to learn a lot about viruses from this book and it is going to freak you out. In a good way.

If you’re curious about how we stopped smallpox, read The Demon in the Freezer, from the man who wrote The Hot Zone (who is also the brother of the man who wrote a novel about a dinosaur sort of thing that comes to life and starts killing people in a museum (yes, I read that too, I will literally read *anything* that is made of words.)

Until next week!

Abi Jones Editor, Locutius

Google I/O’s VUI Surprise

In this week’s “HOLY SH!T” moment, Google debuted Google Duplex, an AI system for accomplishing real-world tasks over the phone. It’s already tough to tell a machine from a person on TTS samples (a la Tacotron 2), but Duplex goes more than a few steps further with an AI that can make appointments on your behalf.

First, check out this hands-on video from CNET Then, read Google’s blog post on Duplex: Google's AI blog

Reaction from the tech sector, journalists, privacy experts, and assorted others was a combination of "wow," "oh no," and "oh heck no, I didn't hear any notice or consent in that conversation." Since the demo, Google has clarified that there will be notice and consent (see this CNET article). However, Sarah Jeong (my favorite courtroom reporter) declares No one knows how Google Duplex will work with eavesdropping laws.

S. A. Applin (aka AnthroPunk) weighs in

Motherboard: Google Duplex Puts AI Into a Social Uncanny Valley

Early responses from the Technorati

Google’s Demo Duplex Stole the Show /// Wired

It’s hard to believe AI can interact with people this naturally /// The Verge

Uhh, Google Assistant Impersonating a Human on the Phone Is Scary as Hell to Me /// Gizmodo

Google Duplex will call salons, restaurants, and pretend to be human for you /// Ars Technica

Also, I made a Twitter Moment because embedding tweets in emails is THE WORST: Duplex gets its Moment

Is Duplex classist?

Google not only deceived these people, but turned them into inconvenient interfaces. The woman at the restaurant, for example, had a thick accent, so the implication is you don't need to suffer through the experience of dealing with someone who might not look or sound like you. A machine can do it. - Richard Nieva, CNET

And this...

(If you aren't already, you should be following Kortney on Twitter and you should learn about Appolition, his latest project, which was recently featured in Fast Company's World's Most Innovative Companies 2018 series.)

Here's everything else VUI-related from I/O:

Assistant

Six (or seven) new voices With advances in AI and WaveNet, Google can now create a new voice in a few weeks instead of spending months in a studio. The new voices are already available for your assistant /// The Keyword (Google Blog)

Continued Conversations Assistant was updated last year to let you say “Turn on the kitchen lights and turn on the hallway lights” in one go, but now you can say things like “Turn on the kitchen lights and the hallway lights”. Progress. /// Android Police

Custom Commands Developers can now build custom commands into their products so that instead of saying “Ok Google…ask ______” you can give the command directly to Google (or Assistant running on the device). /// CNET

Google Assistant's Pretty Please helps your kids mind their manners Hot on the heels of Alexa’s Magic Word feature, Google adds a own polite mode /// CNET

Google Announces New Developer Tools for Google Action Monetization, Discovery and Engagement Seamless digital subscriptions, re-engagement via notifications, and better discovery for 3rd party apps? /// Voicebot.ai

Assistant + Home + Smart Displays

Google’s Smart Displays are going on sale in July These displays were...on display…(cringe) at CES in January and you can get one in just a couple of months. I’ve been playing with the LG one and it’s the thing I wanted 10 years ago when I bought a Chumby. Also, if you're thinking about getting one, I recommend getting the Lenovo version. It's the prettiest and has the smallest footprint. /// The Verge

Google says Android Things is finally ready for smart devices Launched to developers at the end of 2016, Android Things will be in the Smart Displays and some Assistant-including TVs, but there will also be Android Things that aren’t Google devices, like medical equipment and public signage. /// The Verge

Google Home Coming to 7 More Countries in 2018: Denmark, Korea, Mexico, the Netherlands, Norway, Spain and Sweden Google Home will be for sale in 80 countries by the end of the year, but these are the next 7 to get localization (Mexico! yay!). /// Voicebot.ai

Assistant + TV

Google partners with JBL for an Android TV-powered soundbar There’s a cute video and it looks much better than my home situation: We have a smart TV at my house. I can use the Google Assistant (via Home via a Chromecast) to turn it on, but not turn it off, which is super awesome and useful /// The Verge (Here’s a hands-on review from The Verge, too - pay attention to those layers - PlayStation + Assistant!)

Google Assistant arrives on LG's 2018 premium TVs Prices start at $900 for the 55-inch SK8000 TV (it's currently on sale), up to $4,000 for the 65-inch E8PUA 4K OLED TV. /// Engadget

Assistant + Home appliances

Google Assistant update makes Whirlpool voice commands a breeze No more “Ok Google, ask Whirlpool to heat oven to 400 degrees” now you can just say “Ok Google, heat the oven to 400 degrees” - Turn your volume down before clicking because CNET has noisy autoplay ads. :( /// CNET

Cars

Volvo will embed Google Assistant and Maps into future cars Your car’s system will work separately from your phone, so I’m curious about how hand-off will work /// Engadget

AI

Google renames Google Research to Google AI Deep sigh. If you know me, you know how I feel about this. I love the photo at the bottom of the stories page because it was taken in one of our most-used visitor spaces on main campus, not an actual workspace. /// Google AI + The Verge

Wellbeing

Android P now includes Digital Wellbeing features (downtime) I get two kinds of notifications on my phone: text messages and phone calls. But if you find yourself overwhelmed by notifications (OMG, that notification overload when you first download a new app, I’M LOOKING AT YOU INSTAGRAM), then this might be the phone OS for you /// The Verge

Google launches a site on Digital Wellbeing “Great technology should improve life, not distract from it.” /// wellbeing.google

Check out the original “Digital Distraction” presentation A Call to Minimize Distraction & Respect Users' Attention by Tristan Harris on Scribd

This email isnt’ exhaustive, so check out 100 Things Google announced at IO

Non-Google News

Microsoft demonstrates Alex + Cortana integration You’ll be able to say “Alexa, open Cortana” and then you can use Cortana to send an email because Alexa doesn’t do email...yet. I don’t know about you, but I <3 asking a machine to get another machine for me to talk to. Microsoft has a site where you can sign up for more info when it goes live. /// The Verge

Microsoft launches a unified API for all of its AI speech services The unified speech service will combine Microsoft speech recognition service, text-to-speech API, customized voice models and translation service. Currently, these are all available as separate services. I know, this article is on TechCrunch and they have that weird scrolling UI. I’m sorry. /// TechCrunch

Self-driving cars are here They’re starting with local, on-demand shuttle routes in Frisco, TX /// Andrew Ng on Medium

You can now set Amazon’s Alexa as your default voice assistant on Android No maps or phone calls, but you can...do a Google Search? Ok, Alexa. /// The Verge

Introducing Project Kinect for Azure Microsoft is moving Kinect to the cloud /// Microsoft's post on LinkedIn

How uncertainty could help a machine hold a more eloquent conversation ​“Language isn’t really like a decision tree,” Vigoda (founder & CEO) says. “This is trying to be more like a person.”/// Technology Review

Stuff I bought this week

Light up Crocs. Because I'm a mom. If you buy these for your kid, go up a size from their usual Crocs because the back band seems to be shorter and less flexible.

Pour spouts for bottles. I buy olive oil in bulk and decant it into these smaller bottles to keep it by the stove. a) It makes me feel like I'm a chef, and b) Those pour spout things are awesome and I want to use them everywhere.

Until next week!

👋👋👋

Abi Jones Editor, Locutius

Is there something I missed? Want to chat? I’m on Twitter at @jonesabi

Yes, this newsletter contains affiliate links.

A note on AI vs ML before we get into this newsletter: I prefer to use the term ML (Machine Learning) over AI because I want us, as humans who make things, to understand that we are responsible for these outcomes. We’re the ones who teach these machines or provide data sets (or we’re the ones that make the machines that teach themselves) and we need to be considerate about what we’re doing. It's semantics, but meaning matters.

💅🏼💅🏼💅🏼

Microsoft kicks off AI for Accessibility

Today Microsoft announced a $25 million, 5-year program to “put AI tools in the hands of developers to accelerate the development of accessible and intelligent AI solutions to benefit the 1 billion-plus people with disabilities around the world.”

Yes, that press release really does say “intelligent [artificial intelligence]”

Worldwide, only 10% of people with disabilities have access to assistive technology, so it’ll be interesting to see if it’s better to make more tech available or …. Maybe just give more money to people with disabilities? (see this article on cash transfers)

AI for Accessibility has three components:

Providing seed grants of technology to developers, NGOs, universities, and inventors, so that they can do AI-first work

Identifying project with potential and providing more investment so they can scale

Embedding inclusive design and AI across all Microsoft products and helping partners improve the accessibility of their offerings

Read more about Microsoft’s AI for Accessibility program on their blog (there doesn’t seem to be a website yet - but there is a website for the AI for Earth program they launched last year: https://www.microsoft.com/en-us/aiforearth)

Thinking about implementing ML for your product?

Review Martin Zinkevich’s Rules of Machine Learning: Best Practices for ML Engineering (PDF)

Brraaaaaains

Last weekend I got to hang out with a friend who is a neuroimaging specialist. We had a fun conversation about education theory (we’re both former teachers) and we discussed the history of humanity’s understanding of the brain. Have you ever thought about how in the past, we only knew what different parts of the brain did because:

Someone got stabbed in the head or injured and something about their memory, ability, or personality changed

Something about a person’s memory, ability, or personality changed and after they died we cut open their head and looked for damage.

Thanks, brain injuries!

Learn about the brain: What brain regions control our language?

Read the rest of the Brain Control series on The Conversation

🤕 🤕 🤕

Research highlight: Voicesetting: Voice Authoring UIs for Improved Expressivity in Augmentative Communication

Microsoft Research is working on adding emotion to TTS via emoji, tested with participants with ALS, a neurological disease that causes muscle weakness.

Summary YouTube video: https://www.youtube.com/watch?v=Q4FcvvbCM9Y

The paper: https://cs.stanford.edu/~merrie/papers/voicesetting.pdf

The real-life project it’s related to: https://www.microsoft.com/en-us/research/project/hands-free-keyboard/

Why people need it: The ALS Association on Augmentative Communication

Accessibility roundup!

Making something for humans? Here’s a set of tips for designing accessibility into VUIs / UX Collective

5 things Pinterest changed to make its app better for people who are blind / Mashable

Voice should be user-friendly to all users Focus on cognitive impairment, but this advice applies to most VUI use / Speech Tech Magazine

Designing interfaces for the visually impaired Includes references to VUI’s, but the non-VUI parts are most interesting/helpful. The VUI parts are a recitation of a Nielsen report. / Stina Olofsson

VUI development for Korean people with dysarthria Researchers prototype and test a customizable VUI that works for people with slow or slurred speech (could be from trauma, a stroke, neurological disorders) / JOURNAL OF ASSISTIVE TECHNOLOGIES

Accessibility features on Alexa + How to turn on the VoiceView screen reader Yep, the Amazon feature list. I turned on captioning on my Echo Show - have I mentioned lately how much I love captioning? I have it permanently on on my Netflix account.

How deep learning is helping the blind ‘see’ “Stevie Wonder uses it every day, which is pretty cool,” said Anirudh Koul, a senior data scientist with Microsoft, during a presentation at the GPU Technology Conference. / nvidia blog

🔗🔗🔗

VUI news & links I want to get out of my queue before Google I/O starts

How Indian startups gear up to take on the voice assistants of Apple, Amazon and Google What sets these start-ups apart? They’re multi-lingual from the get-go.  / The Economic Times

Why Google Assistant will beat Siri, Alexa and Cortana Mike Elgan argues that success in the assistant space requires total ubiquity / Computer World

How Voice-First Technology Helps Older Adults This article offers a good overview of some VUI research with older adults the the current state of the “VUI for seniors” market / Forbes

Sensory Brings Low-Power Wake Words to Mobile Apps Now apps could have wake words, too. / Sensory blog

Amazon finally opens up Alexa to developers to make money off third-party skills Developers get 70 percent of revenue from the in-skill purchase / The Verge

Sonos announces June 6th event for new home theater speaker with Alexa It has an HDMI port (so it *could* control your TV) and it has a VUI / The Verge

The Elements of Multi-Modal Design with Karen Kaushansky – Voicebot Podcast Ep. 40 / Voicebot.ai

Read this thread from Liz Jackson on how disabled creators were erased from the Cooper Hewitt Access + Ability exhibit / Liz Jackson on Twitter

Alexa, can you hear me? Bravo to Vocalize.ai for highlighting algorithmic bias in their report on performing black-box evaluations of ASR systems / Vocalize.ai via PR Web (I know, 🤷)

📚📚📚

Reading status

I’m still reading the three books I mentioned last week (Fascism, Robots, and. . . Robots), but I couldn’t resist picking up Dear Ijeawele, or A Feminist Manifesto in Fifteen Suggestions by Chimamanda Ngozi Adichie. I have a son, but I’m still raising a feminist.

Until next week!

👋👋👋

Abi Jones Editor, Locutius

Is there something I missed? Reply to this post with a link! Want to chat? I’m on Twitter at @jonesabi

Yes, this newsletter contains affiliate links.

Alexa for kids

Alexa for kids ships later this month, and while I’m not sure I’ll buy one for my kid (I prefer my kid’s VUI use to be out where I can see it - er, hear it), I might get one for myself. Also, the current state of hand-off between my 1st Gen Echo and my Echo Show is so bad that the last thing I want to do is introduce another device where I have to whisper conspiratorially to it in order to make sure that the right one does the thing I want it to do.

Of course, new technology begets new thinkpieces on whether or not technology is good for kids. My first reaction to the Echo for kids was horror, but I appreciate that Amazon has had content filtering on the Kindle Fire for ages (it’s nearly impossible to use, but it exists), and I am a kid who watched as much TV as I possibly could (hours of Saturday morning cartoons, hours of after-school cartoons, and plenty of totally child-inappropriate Arnold Schwarzenegger movies) . . . I turned out okay?

Fine, I’m not letting my 5-year-old watch Commando anytime soon, but I’m also not too concerned about VUIs. They’re so limited in their capabilities that right now the only negative is that I set a horrible example: I yell at a device and in turn my kid yells at the device and there’s frankly a whole lot more yelling than I want happening in my home. Do I want a device that praises my kid for saying please? Or do I just need to start setting a better example on my own?

Alexa aka Echo Dot for kids: Ships from Amazon on May 9th (and feature a whole lot of subscriptions that probably auto-renew so now’s a good time to determine if you want to go down that path)

More Amazon/Alexa:

Tidying up’s a snap: Alexa is like a virtual Mary Poppins Amazon’s voice-activated assistant — by way of the Echo and Echo Dot — helps this family stay on task. (shaking my head at all of the Mary Poppins references lately)

Alexa is a revelation for the blind Ian Bogost writes “Legally blind since age 18, my father missed out on the first digital revolution…” A lot of the difficulties Ian’s dad runs into are common to anyone using a voice interface for the first time - they aren’t specific to people who are blind, or older adults, or even those who don’t do much computing. The real story here is that Ian had to reconsider his father’s relationship with the world, and the promise of independence offered by voice assistants.

When will Alexa know everything? An interview with Al Lindsay, vice president of Alexa Engine Software at Amazon. Literally every time I read this man’s name I read it as A.I. Lindsay - like Artificial Intelligence - that’s what happens when you live in Silicon Valley too long and not enough websites use serifs.

Google did some stuff, too

Google releases conversation design guidelines for Actions for the Google Assistant Conversation Design with Google

Highlights

It’s #5 on the site list, but should be #1 in your brain: Is Conversation the Right Fit? (also, there’s a quiz and I cannot resist a quiz)

Now with more dialogue samples: A Crash Course on Conversation Design

And error handling: Errors (no fancy title on this one)

Google Assistant can now control your Ikea Tradfri smart lights You know, in case you wanted to add yet another smart home app to your phone. / Android Community

I have one Microsoft news item

Here’s an interview with Javier Soltano, the new product lead for Cortana. The interview starts out with talking about Microsoft’s competitive advantage: email. But don’t think about it as email - think of it as a way of understanding the relationships between people and understanding the relationships between projects and concepts and...well, it’s interesting when you start thinking about what email really is. / VentureBeat

Is Cognitive Computing real?

Introducing CASE: the cognitive coffee-maker Is cognitive computing a coffee maker that recognizes you, makes small talk, and knows your favorite drink? / IBM Internet of Things blog

The fraudulent claims made by IBM about Watson and AI I don’t really have a whole lot to add to that...inflammatory title, so go ahead and read it for yourself and learn about the history of AI and what it is and isn’t / Roger Schank (artificial intelligence theorist - his wiki page)

More on AI (this one is a really good read, I promise)

And here’s a brief history of AI that’s just...accessible. Read it to get some background on just what AI really is and how far we have to go: [FoR&AI] The Origins of “Artificial Intelligence”

Conversational Design, the book

Erika Hall has a new book out: Conversational Design. It’s not about VUI’s, but it still has a lot of interesting lessons on language and information exchange. You can read the first chapter for free on A List Apart and here she is back in 2016 talking about the importance of language in design.

Locutius Links

Rethinking the Smart Home in 2018 Smart speakers are great, but they aren’t enough / Stacey on IoT

Quiz: Are these idioms British or American English? I failed. / Quartz

Google launches an improved speech-to-text service for developers / Techcrunch

Library Journal: Voice Activated The Library Journal offers some background and advice on smart speakers for library services. (Less “ask questions in a quiet space” and more “borrow audiobooks”) / The Library Journal

'Alexa, play my kid a podcast.' Parents look for screen-time alternatives If you have a 5yo, try the podcast But Why! We love it. Every time a 5yo on the show says their age, my kid says "She's 5, just like me!" - it's a revelation for him to hear kids on the radio.  / CNN

Alexa, are you SkyNet? This talk is “an in-depth forensic dive into the data that is stored, transmitted, and can be recovered, from the Alexa portion of the Amazon ecosystem” / SANS Digital Forensics and Incident Response on YouTube

Tencent’s Smart Speaker Tingting Brings WeChat to the Entire Family IT HAS A BATTERY SO YOU CAN UNPLUG IT AND USE IT WITH WIFI FOR 5 HOURS. / Synced

A pioneer in predictive policing is starting a troubling new project We have a policing problem in the United States. We have a prisons problem in the United States. We need to actively consider just what we’re doing when we apply machine learning to policing and the judicial system. / The Verge

A note on newsletters

I took out half of the content I’d originally included it this email. Now I know why people send out daily newsletters.

Don’t worry, I’m not making a daily newsletter, but if you’re in California, I recommend subscribing to the California Sun (it's free) as a way of understanding what’s happening in our giant state (I’m from Oregon, which has ⅔ of the land area of California, but just 10% of the people): California Sun

This week I’m reading three books:

Fascism: A Warning by Madeleine Albright It is sooo good and sooo infuriating. I started it on a plane and found myself silently saying ‘Fuck’ every 30 seconds as I read it. Now you’re warned.

Head On: A Novel of the Near Future by John Scalzi It’s like a good Michael Crichton book, meaning it’s less about technology and more about people. This is the second book in this world, so consider picking up the first one (Lock In). Also, I just learned that Scalzi wrote the main character in a way that they could be a man or a woman and my brain assumed they were a man because...patriarchy. I’m kind of sad about that.

Machines of Loving Grace: The Quest for Common Ground Between Humans and Robots by John Markoff I was skeptical about this book because on the first page of the preface Markoff describes the Stanford golf course Coupa as “adjacent” to the course, when it’s far more accurate to say that it is “at” the course. Other than that, I’m getting a good background on robotics. Yes, it's full of dudes and Markoff tries to make up for that in the preface by name-dropping Cynthia Breazeal, but it feels like an afterthought. Still, it's a good book, just...so many guys.

If you’re wondering how I read so much, I will tell you my secret: my hobby is reading.

Until next week!

👋👋👋

Abi Jones Editor, Locutius

Is there something I missed? Reply to this email with a link! Want to chat? I’m on Twitter at @jonesabi

Yes, this newsletter contains affiliate links.

Get over your imposter syndrome. Now.

If you’ve ever experienced imposter syndrome in regard to crafting a VUI, simply behold this video of a *successful* bluetooth device setup. It’ll improve your self-confidence by 20 percent, easy: 

My in-the-moment device setup Twitter thread (yes, I’m always looking for VUIs, even when I’m on vacation): https://twitter.com/jonesabi/status/987424238868066304

MIT’s AlterEgo headset ‘hears’ your inner voice (and talks to you via bone-conduction)

The bone conduction technology is nothing new (you can get a bone-conduction headset on Amazon for $60), but until now, the ability to turn internal verbalization into a signal was limited to astronauts.

Learn about Subvocalization (the silent speech of reading!) on Wikipedia More about AlterEgo on Forbes Yeah, NASA was working on this 15 years ago. Read about their efforts: NASA Develops System To Computerize Silent, "Subvocal Speech"

Emergency workers to start using yelling-based interface

In a planned July launch, a Massachusetts ambulance service will use Amazon’s Alexa service as a VUI for first responders.

The Brewster Ambulance Service is currently converting a 300-page document, the Emergency Medical Services Statewide Treatment Protocols, into a voice interface. Current bugs: Latin words (common in medicine) and abbreviations (like when I asked my Echo about PM Dawn).

The company that owns Brewster Ambulance Service also owns another ambulance company serving eastern Massachusetts. Combined, those companies answer 400,000 emergency calls a year. And they’re going beyond voice interfaces to use Microsoft’s HoloLens for training, and CDC software to create response heatmaps for placing ambulances in anticipation of emergency needs.

Read more in the Patriot Ledger: Alexa to ride aboard Brewster ambulances

How Flo and her fellow chatbots engage customers

Microsoft flaunts Progressive’s Flo, a chatbot built on their Azure Bot Service and Microsoft Cognitive Services (including their Language Understanding Service). I <3 the idea of writing a chatbot for a pre-existing personality rather than trying to craft a persona via the limited interactions of an insurance chat. It makes me wonder if we’ll see more ‘personality’ commercials in the future that don’t just establish the capabilities of a voice interface, but give you a headstart on its personality(ies?) as well.

Read more: Guess who wants to talk! How Flo and her fellow chatbots engage customers on Microsoft Transform.

Locutius Links

CHI is happening right now And there’s a free livestream. Yeah, I don’t know why this page is organized by room number either. But it’s free! / https://chi2018.acm.org/attending/stream

Hello, Hal In 2008 the New Yorker wondered “Will we ever get a computer we can really talk to?” and it turns out the answer was “In 10-15 years, depending on what you mean by ‘talk to’” - this article is a fun time capsule. / New Yorker

Can You Believe Your Own Ears? With New 'Fake News' Tech, Not Necessarily An alarming number of the VUI articles I read are in the “National Security” sections of news sites. / NPR

Xiaomi launched a China-focused voice assistant It has direct hooks into WeChat. / The Verge

Alexa supports charitable donations 40 charities are included in the initial launch and you can make donations between $5 and $5,000 / CNET

Google opens Cloud Text-to-Speech to devs The service features 32 different voices from 12 languages and variants. Developers can customize pitch, speaking rate, and volume gain. / 9 to 5 Google

Over Half of Smartphone Owners Use Voice Assistants, Siri Leads the Pack Okay, it’s 52% which is technically (but barely) over half. / Voicebot.ai

CodeAcademy has a “Build an Alexa Skill” course It’s free! / CodeAcademy

An interview with Eileen Lee on designing AI-powered interfaces From Slack bots to eBay / BotSociety.io

Big Brother is watching them. And we’re next The Times takes a look at digital totalitarianism. Reading the article requires a subscription, but you can always use a fake email address if you don’t want to be tracked (yes, I understand the irony of all of this) / The Times

And finally, the best things I bought this week:

I packed our Sum Swamp board game on a recent trip to Philly and it was a hit with the 4-5 year-old set. Yes, it is a math game. But it’s also fun, easy, and has just enough “game” to it (shortcuts!) that our kid loves it. Other games our kid likes to play: Pop the Pig (I think it is truly awful, but it’s easy enough that kids can teach each other to play), Hiss (you need floor space for this one), Chutes and Ladders (good for counting), and Math War.

Until next week!

Abi Jones

Editor, Locutius

Want to chat? I’m on Twitter at @jonesabi

And yes, this newsletter contains affiliate links. I’ve made $4.48 this year. Thanks, whoever bought those packing cubes the other day. I hope they changed your life the way they’ve changed mine.

🚫🔫

My friends were shot at today.

It happened at work. While they were eating lunch. It happens every day in the United States. I'm doing something about it. You can do something about it too.

Join me in fighting the NRA - go to Moms Demand Action and enter your contact info. You don't have to be a mom. You don't have to give money. You just have to use your voice. After you sign up, someone like me (I'm on the Welcome Committee, so it literally might be me) will give you a call and personally help you get involved in a way that works with your schedule and your life. We represent a movement of over 4 million people who are fighting to make our country safer. Join us: Moms Demand Action. Use your voice.

Here's the newsletter I was going to send out before today happened.

P.S. If you have any questions about Moms Demand or want to get involved, I'm happy to give you more info. A lot of times the fight against gun violence feels like something that can't be won, but here's a list of changes in just the last 5 years: https://momsdemandaction.org/5-year-anniversary/ Okay, back to our day jobs. Here's hoping you don't get shot.

️Is your voice assistant cozy?

I’m writing a mystery, so I’m aware of the idea of cozies (mysteries that downplay murder, sex, and violence and generally take place in a small town) and I don’t really read them. But I do play and enjoy ‘cozy’ games.

“Coziness itself refers to how strongly a game evokes the fantasy of safety, abundance, and softness.”

Safety: An absence of danger and risk - no loss or threats Abundance: Lower level needs (food, shelter) are met, so you can work on high-level needs like self-actualization, nurturing, and belonging. Softness: Intimate, low-stress, comforting stimuli

Read more about coziness in games, and consider how they do or don’t apply to the voice systems you’re working on: Group Report: Coziness in Games: An Exploration of Safety, Softness, and Satisfied Needs

Facebook Delays Home-Speaker Unveil Amid Data Crisis

Bloomberg reports that Facebook intended to unveil the speaker in May with an F8 preview and launch it in the Fall, but that plan is changing due to public sentiment about the social network.

“Facebook has faced a public reckoning this month about its treatment of user data, sparked by reports that political-advertising firm Cambridge Analytica obtained information on 50 million users without their permission.”

Voicebot.ai looked specifically at user trust and and smart displays in a recent article: Is Now the Right Time for Facebook to Launch a Smart Display? (conclusion: Facebook could gain trust by being an app on already-trusted (relatively speaking) Alexa or Google rather than creating their own agent)

Read more: Facebook Delays Home-Speaker Unveil Amid Data Crisis on Bloomberg

Amazon’s Alexa in India

Local developers have taken to Alexa, and with support from Amazon in the form of theme-based contests, webinars, YouTube videos, and meetups, more than 15,000 developers in India are giving Alexa insight and inroads into Indian culture.

"Alexa is not going to be a visiting American who is going to come to India for a few days and go back. She is as Indian as it gets," said Dilip RS, country manager, Alexa Skills, India.

Read more: 'Amazon Alexa is not a visiting American on a short trip to India, it is here to stay' on GadgetsNow

Also: Amazon wants your help teaching Alexa new languages — and it could help in its fight against Google

And: Alexa smartphone: Amazon's next strike in the mobile IoT war?

JPMorgan Chase rolls out a new Alexa skill

“JPMorgan itself admits that, before it can allow Alexa to execute trades on clients' behalf, the bank will have to reinforce the security and authentication of the skill to prevent the assistant from acting on an unrelated direction it overhears, or on a fallacious request submitted by someone besides the owner.”

Don’t worry, nothing has ever gone wrong in the financial services industry.

Read more: JPMorgan Chase rolls out a new Alexa skill in Business Insider

P.S. Why is it pronounced Fĭntech, but the parent industry is pronounced Fīnance? Is this because of how people pronounce things in England? (I’m not a linguist, so I spend a lot of time listening to these sorts of videos to understand this stuff)

Locutius Links

Are we already living in Virtual Reality? It turns out people with an extensive understanding of their own bodies (athletes, dancers, yogis) find the adoption of a virtual body difficult. / New Yorker

How Feminists in China Are Using Emoji to Avoid Censorship 🍚🐰/ Wired

The Pyramid of Clarity How Asana creates strategic alignment to get product results - it’s worth a read / Wavelength, the Asana Blog

Conversational Intelligence Challenge 2 is open (chatbots). The winner gets $20,000 of Mechanical Turk credit. One of the example personas doesn’t like Mexican food, so I almost didn’t post about this because Mexico is a giant country with a lot of different food cultures, so perpetuating the idea that Mexico is monocultural is just not okay. Click the link with that in mind. / Convai.IO

Research: Take a breath and take the turn: how breathing meets turns in spontaneous dialogue Breathing is an active component of turn-taking and we breathe differently when we’re turn-taking vs. turn-holding. Really interesting when you think about how VUIs don’t breathe (note, don’t start designing VUI breathing just because I wrote that. What I mean is we need to understand breath, the cues in it, and then use and transmit those cues appropriately - DON’T MAKE YOUR VUI BREATHE. DON’T DO IT. Unless you’re doing research and then pleeeease make a breathing VUI). Thanks for the link, @virtualgill! / Philosophical Transactions of the Royal Society B (yes, that’s the real name of the journal site) also, a PDF link

Cathy Pearl held her AMA I’m glad to see that the first questions are about killer AI because Cathy’s responses are funny and thoughtful / Reddit

AI assistants say dumb things, and we’re about to find out why What if machines had common sense? / Technology Review

Microsoft announces breakthrough in Chinese-to-English machine translation This article is a good start for looking at how researchers check, double-check, and improve the work of machines through a variety of methods. / Techcrunch

Home Smart Home: Technology has raised the bar for how perfect our homes can be  A lot of this article isn’t about technology, but about the subtle differences that *do* make a home a home - better lighting and dental floss you actually like - and it’s about things anyone can do to make their home safer or more comfortable: installing water sensors or using a shared family calendar. / Curbed

Ecobee recruits Amazon’s Alexa for their new Switch+ smart switch It’s not a light switch that *works* with Alexa - it’s a light switch with Alexa built in. / The Gageteer

Reading list!

Next week I’m sharing a reading list on Voice Interfaces and technology. I’d love to get recommendations for the books you’ve enjoyed recently. Catch me on Twitter at @jonesabi

Best,

Content warning: domestic violence

First, read Domestic Violence, a short story by sci-fi writer and futurist Madeline Ashby. The future isn’t so much a future as it is possible right now.

“I had to do the chicken dance. It started playing the song and then I started dancing, and then the door opened. I think maybe some kid in the building hacked the door.”

Photo courtesy MEDION Pressestelle https://www.flickr.com/photos/medionpr/36399899774/

Then, read this response essay from Ian Harris, who works on technology issues with the National Network to End Domestic Violence.

“Smart home” technologies create an especially difficult challenge for safety. Cameras that can be remotely accessed, smart speakers that are always listening, and sensors that identify which room you are in and how many people are in the home—these technologies already exist with very minimal regulation or oversight. If current trends continue, I fear that Ashby’s tale is not far away.

If you’re interested in the HCI perspective on this, read this paper from my colleagues in the Security and Privacy group at Google: Stories from Survivors: Privacy & Security Practices when Coping with Intimate Partner Abuse. The paper is a frank and heart-wrenching.

Later, Abi Jones Editor, Locutius

The Most Important Design Skill For An AI-Dominated World

Admission: I tried reading John Maeda’s Design in Tech Report while standing in line at Passport Control and I had to give up on page 21 (consulting firms M&A) because it was literally unreadable on my phone. Argh. Ranting about abandoned and/or poorly maintained acquisitions aside, this quote stuck with me:

“Computers aren’t good at inclusion,” he says. “They’re good at exclusion, because they’re only based on past data. The business opportunity for the future-thinking designer is in inclusion.”

🖥🖥🖥

Microsoft Soundscape

So often we think about accessibility in terms of telling people about what’s right in front of them (or obstacles in the way), but if you’ve ever travelled to an unfamiliar place, you know that discovery comes from paying attention to what’s off the beaten path as well.

“Obstacle avoidance is not the problem, we have a dog, a cane and our blindness skills for that,” said Erin Lauridsen, Access Technology Director, LightHouse for the Blind in San Francisco. “The gap is knowing where things are and being able to decide what's of interest.”

When I travelled to Tokyo I ran into a lot of problems because I couldn’t read street signs or business names. This is a problem that visually impaired people deal with every day. I recommend watching the video to get an idea of what the app can do. Read more:

Watch the video on the project site (It’s the best explainer)

Soundscape app empowers people who are Blind or have Low Vision to explore the world

Download the app (iOS only, sorry)

👂👂👂

Bots! Why your chatbot needs to care about context

When it comes to conversation design, chatbots and VUIs have a lot in common (and a lot not-in-common..ephemerality, cough, cough). In this article, Gillian Armstrong covers some of the context your chatbot should care about, beyond When, Where, and What (these are things your VUI should care about, too). Read it: Why your chatbot needs to care about context

🤖🤖🤖

More bots: Mental health AI platform, Woebot lands $8M in funding

What makes Woebot different? Before founding the company, Alison Darcy was a clinical psychologist at Stanford. How rare (and great?!?) is it to see someone with subject matter expertise founding a company. Interested in working at Woebot? They’re hiring a creative writer who can specialize in Cognitive Behavioral Therapy: https://woebot.workable.com/jobs/655376 Read more about Woebot:

The Chatbot Therapist Will See You Now and

I spent 2 weeks texting a bot about my anxiety — and found it to be surprisingly helpful

😞😞😞

Facebook / Iceland / Linguistics

When I interviewed at Facebook, one thing the Groups team was excited about was that most women on Facebook in Iceland belong to the same Facebook group. Yeah, like Pantsuit Nation, but for *all* of the ladies in Iceland. Granted, there are only 167,000 women of any age in Iceland, but when you’re talking about cultural decimation, those numbers become even more important. In an article in The Guardian, Jon Henley details how digital language chips away at Iceland’s language and culture, one phone at a time, in interviews with Icelandic professors of linguistics and digital media. “Once, outside school you’d do sport, learn an instrument, read, watch the same TV, play the same computer games,”...“Now on phones, tablets, computers, TVs, there are countless games, films, series, videos, songs. You converse with Google Home or Alexa. All in English.” Read more:

Icelandic language battles threat of digital extinction

No mean girls: Why one-third of this nation's women joined same Facebook group

❄❄❄

Locutius Links

Oral-B made a 2-minute podcast for when your kids brush their teeth. It’s marketed for Alexa, but it works on Google, too. / Gimlet Media

Alexa has follow-up mode Now you don’t have to say “Alexa, turn on the kitchen lights.” and then “Alexa, turn on the dining room lights.” You can’t say “Alexa, turn on the kitchen and dining room lights” or “Alexa, turn on the lights for dinner.” But baby steps, eh? / Apple Insider

Microsoft drops ‘Hey Cortana’ in favor of just ‘Cortana’ on smart speakers / The Verge

How Otto, a German ecommerce giant, uses artificial intelligence. This one you'll need to read for yourself. Automation isn't going away. / The Economist

Why you Shouldn’t Skip a Linguistics Analysis Before you Pick a Company Name “The three basic metrics to test for are pronounceability, negative meanings, and existing brand associations.” / Rewind and Capture

Review: I tried Levi's $350 denim jacket featuring Google technology — and it made my commute so much better I’ve tried this jacket, too. All I can say is that I wish they made a ladies version because the cut on this one is strictly for guys with trim waists and big shoulder muscles. Also how 'alterable' is "smart" clothing? / Business Insider

The Feds Can Now (Probably) Unlock Every iPhone Model In Existence / Forbes

Now we know why Siri was so dumb for so long / Mashable

It turns out that humans don’t like robots that don’t like humans. Who could have predicted this?!?! / AV Club (no, not The Onion - this is a real story about people putting BBQ sauce on robots)

Robotic Tortoise Helps Kids to Learn That Robot Abuse Is a Bad Thing. What are the ethical implications of harming a robot? Do we get mad if people hurt their cars? Or are robots more like pets? Is there something predictive about how we treat robots? Or are humans actually smart enough to know that robots don’t have feelings and they’re just machines? Why do we describe breaking a robot as immoral, but we wouldn’t use the same description for a vacuum cleaner? I mean, unless it was a robot vacuum cleaner. / IEEE Spectrum

Robot ethics aside, here are the best things I bought this week:

Melatonin. I just returned from Tokyo and right now I’m confused by two things: 1. Which side of the sidewalk to walk on, and 2. How there is a 16 hour difference between Tokyo and San Francisco, and yet I am somehow not jet-lagged? I’m chalking this up to 3 mg of melatonin on my first night in each place (Yeah, just one pill, one time, each way - for $8.00 you get more melatonin than you'll use in 10 years). At this point, I don’t care if Melatonin is a placebo. I just know that I feel like a normal human being, which should be impossible.

Make your kid happy with a set of My First Temporary Tattoos: Adventure, Creatures, Sports, and More. These temporary tattoos are easy to apply and look cool for a week. The only downside is that a) they're an add-on item, and b) you have to use rubbing alcohol to remove them.

A PURSE THAT CAN FIT A 10.5” iPAD PRO IN A CASE INSIDE OF IT. And that’s all you need to know. Okay, also it is leather, it’s on sale for 70% off of the regular price. And it is SUCH A NICE BAG, with zipper pockets and places to put your charger and your wallet and even your pens.

When your VUI refuses to follow instructions

No, it's not the premise for a new sci-fi film - it's what happened on a variety of Alexa devices, scaring the ever-loving-bejeezus out of the people. That is, until Amazon changed the command from "Alexa, laugh" to "Alexa, can you laugh?".

Note that the laugh no longer sounds like a ghost child haunting your kitchen. Also, I can't wait for when people start gaslighting each other using voice interfaces. "Did you hear something?" "No, I didn't hear anything." "I could have sworn I heard something."

Read more:

Alexa, Please Stop Laughing

😂😂😂

Night-mode for Google Home = lifesaver

Unlike every human being in existence, your Google Home or Alexa or whatever you use (What do you use?) doesn't understand the context in which it's responding to your request. Rocking out at 6PM? Guess what, now it's 9:30PM, your kid's asleep, the house is quiet, you ask what year Sputnik launched and get an answer at VOLUME SEVEN.

Great. Now you'r

Smart home devices aren't very "smart" yet, but now you can turn down the volume on your Google Home after hours.

Follow these instructions to set the night mode on your Google Home

. Yes, it's a just a dumb (as in, not a smart, learning from you a la Nest "Smart") scheduler, but it should save you that awkward 11PM blaring trivia answer.

Note: There are TWO (yes, count 'em, two) places where there are settings for your Google Home device. Follow the instructions linked above, that's how you'll find the right setting. And yes, this is a per-device setting.

🦇🦇🦇

Smart Home technology...wastes?!?...energy

In leaky, poorly built, or old homes, smart thermostats and other devices can save energy. But with something like the Passive House (it has 18" of insulation), having a "smart home" doesn't do you a lot of good, energy-wise.

This pair of articles contains some info I didn't expect: most homes have 65 devices sucking down energy at any given time. Wow. (Note: there's some ableism in the articles, ignoring that for some people, having the ability to turn on and off lights without walking through a dark room can be a help.)

Read more:

Vampire power is back, and it's thirstier than ever in the new smart home

Read more:

Smart Home technology won't save energy; it wastes it.

🦇🦇🦇

Consensual Software

No, this isn't about Consent Blockchains (which are a fucking horrible idea - pun intended) - it's an open-source project advocating for better user consent in software design. Take a look at their Twitter account - you'll cringe. And then you'll go update your LinkedIn settings. Consensual Software on Twitter: https://twitter.com/consentsoftware

👍👍👍

"Great Rooms" as a space for introducing new tech

"Great Rooms" are the combination of kitchen, dining, and living space, frequently found in American homes built in the last 20 years. Nowadays we call it an "Open floor plan" because "Great room" sounds like the 90's, but the concept is the same - a giant unified space instead of separate rooms for cooking, eating, and relaxing or entertaining. This paper by Scott Mainwaring and Allison Woodruff explores great rooms, including the hedonic experience, wastefulness, invasiveness, and the implications for technoogy design. Disclosure: Allison and I used to be on the same team at Google, but she doesn't know I'm sharing this paper. And she wrote the paper while she was at Intel Research. Read the article (PDF): Investigating Mobility, Technology, And Space In Homes, Starting With “Great Rooms”

🔗🔗🔗

Locutius links

Trying to cover the "whole world" with the voice interface Scott Huffman, VP at Google gives a brief (2 min) interview on Google's Assistant + VUI + AI / Bloomberg Tech

Projector + Nest Cam = Verifying you know everyone in The Walking Dead Thanks for the laugh, Jina! / @jina on Twitter

Toward an ethical, transparent, and fair AI - a reading list "We also need to remind ourselves that algorithms don’t exercise their power over us. People do." / Eirini Malliaraki on Medium

Cathy Pearl is doing an AMA next Tuesday, March 20th / Proof on Twitter

Ever heard your own echo on a conference call? Arghghg. It turns out that a delay of over 50 milliseconds creates a 'train tunnel' effect and distracts you. This is the technology that prevents that: Echo suppression and cancellation. Always nice to know the technological underpinnings of your work, right? / Wikipedia

💸💸💸

The best thing I bought this week

I'm a huge "25 things on ASOS that are sooo cute" window-shopper (I have never bought something from ASOS, sorry Buzzfeed), so I figured I'd share two essential items. One is for people with dogs. One is for no-dog people.

Do you have a dog? Does it like to chew on things? Get your dog a Benebone bacon dog bone. My dog can eat a Nylabone in 15 minutes. Benebones last for a month. Also, when these come in the mail he hounds me (haha)until I open the box. Required dog photo below. 

Stop trying to manually squish things into your carry-on suitcase. Get yourself a set (or two) of Eagle Creek Pack It Specter Compression Cube Set (they should be ~$32). Stacy-Marie Ishmael recommended packing cubes last year. I finally took her advice and now I regret all of the times I didn't use them. They are AMAZING. I am actually thinking about bringing a robe with me.

👋👋👋

I'm heading to Tokyo tomorrow to give a talk at UX Days Tokyo and teach a day-long VUI design & prototyping workshop. It's my usual half-day workshop, but it'll be live translated line-by-line, so it will take twice as long. I'm also planning more and shorter activities for the workshop since time flies when you learn from activities, not lectures. Wish us luck!

また今度 (next time),

Abi Jones Editor, Locutius Is there something I missed?

Reply to this email with a link! Want to chat? I’m on Twitter at @jonesabi

The City That Remembers Everything

Technologists don't do enough work to imagine and counteract the potential harms of our products. Most of us are trying to make people's lives better, so we don’t imagine how a stalker or a burglar or an identity thief would make use of our products.

In this story in The Atlantic, Geoff Manaugh writes about our growing surveillance state from a slightly different perspective - that of the potential criminal (and we’re all potential criminals given the right type of regime). Geoff is the author of A Burglar’s Guide to the City, a fantastic book on casing spaces.

From the article:

“In October 2017...researchers at Mitsubishi revealed that they had successfully used machine learning “to pick a single voice out of a crowd,” allowing them to track a person against the background noise of a cocktail party—or a political demonstration. This ability might allow your smart speaker to recognize your voice requesting a song change even in the midst of a New Year’s bash, but it will also be a boon for law enforcement. Imagine the police hunt of the near future, for example, as a suspect risks having her voice picked up and identified by various combinations of strangers’ smartphones and Amazon Echo devices. Fleeing crime scenes in utter silence might be the strategic response.”

Read it: The City That Remembers Everything in The Atlantic

Learn about malicious AI use!

Download a nicely-formatted 100-page report: https://maliciousaireport.com/

Don’t want to read a 100-page document that uses the word “threat” about 482 times? Here are the committee’s 4 high-level recommendations:

Policymakers should collaborate closely with technical researchers to investigate, prevent, and mitigate potential malicious uses of AI.

Researchers and engineers in artificial intelligence should take the dual-use nature of their work seriously, allowing misuse-related considerations to influence research priorities and norms, and proactively reaching out to relevant actors when harmful applications are foreseeable.

Best practices should be identified in research areas with more mature methods for addressing dual-use concerns, such as computer security, and imported where applicable to the case of AI.

Actively seek to expand the range of stakeholders and domain experts involved in discussions of these challenges.

What if your Echo had Darth Vader’s voice?

What if your Echo could talk with a celebrity voice? Or your best friend’s? Or maybe...your own voice? That future isn’t far off. Baidu Research has published info about their voice-cloning tool, which can use just a few utterances to create a voice clone (they started with needing 30 minutes of speech).

Direct link to the paper: https://arxiv.org/pdf/1802.06006.pdf

🔊🔊🔊

How Siri decides which device to respond from

Time of last use ➡️Proximity ➡️Apple Watch override ➡️HomePod (no handoff? yikes!)

Read more: Siri info on The Mac Observer

Locutius links

Do Gricean maxims apply to Japanese preschoolers? First, read this one for the methodology - researchers used games instead of adult conversations to engage the kids. One finding: children and adults differ in their comprehension of the maxim of politeness. / Frontiers in Psychology

Google’s Assistant is going global Google’s Assistant is available in 8 languages - by the end of the year they plan to be in 30 along with multi-lingual support (I’m not sure if this supports using two languages in one sentence or not) / The Keyword (Google’s Blog)

FBI statement on “internet-connected” toys How often do you get to read an FBI statement that’s directly connected to your work? Well, now you do. / PSA from the FBI

Asking the right questions about AI Most discussions of AI focus on the wrong wrongs. Here’s a different way to think about it / Yonatan on Medium

Everything’s a speaker The Nest Cam IQ Indoor is now also a Google Assistant speaker / CNET

3 Big Takeaways From Microsoft's Latest Earnings Amazon’s Alexa is now on multiple Microsoft OS laptops and Microsoft says that healthcare companies are using Cortana to create their own voice services / Fortune

Siri exceeds expectations in one venture firm’s in-office trial Loup Ventures says Siri does better than expected in voice assistant face off, but Google still leads / Loup Ventures

Thanks, Joe! (Here’s your HomePod sound review) Audio Olympics! Audio experts weigh in on HomePod vs. Google Home Max vs. Sonos One with Alexa / Full report on Linkedin

HomePod VUI (not a link, sorry) Every HomePod review mentions the same compelling feature for anyone who’s ever yelled ‘OK, GOOGLE!’ or ‘ALEXA!’ over the sound of music or a dinner party - with the HomePod, you can use a speaking voice (or even lower) to invoke Siri. It doesn’t seem like a big deal, but it’s a great way to make the device less disruptive and more a part of everyday life.

Japan tips?

I'm heading to Japan in a couple of weeks for a VUI talk and workshop at UX Days Tokyo. If you have any must-do's in Tokyo, let me know!

じゃあね (ja ne) Abi Jones Editor, Locutius

Is there something I missed? Reply to this email with a link! Want to chat? I’m on Twitter at @jonesabi

Research: Voice interfaces in everyday life

A research team at the University of Nottingham collected audio data from monthlong deployments of specialized recording devices placed next to people’s Amazon Echos and documented how people actually use these devices in the social context of their homes. The voice interfaces we create are embedded within conversations that we don’t even realize are happening. Sure, you can review all of the logs you want, but unless you have a research team that’s practicing ethnography, you’ll never really understand where your VUI fits in to people’s lives. Read the ‘easy to understand’ blog: Talking with machines? Voice interfaces and conversation design Read the research paper: Voice Interfaces in Everyday Life (PDF)

Amazon launches new Alexa Developer Console

Read the Amazon blog post about the developer console updates The best/most holistic review of the new console is on TechCrunch

Awesome deal: Alexa & Google compatible smart lights for just $9.99 (and Prime shipping!)

Book review: Algorithms of Oppression

“Infrastructure is created by people and therefore embeds and reflects the values of the people who create it.”

I just got my copy of Safiya Noble’s Algorithms of Oppression and I hope that you order a copy (and read it!), too. If you work on a voice interface, then you need to understand the biases in your system. Read the LA Times Book Review of Algorithms of Oppression Order the book for yourself on Amazon or Google Play Books

Finding a “Happy Path”

Don’t miss this lovely reflection on writing for voice interfaces: Absurdist dialogues with Siri in The Paris Review

Locutius links

Soldiers will inevitably put devices in their bodies. Then what? / Link

The Economist: In the world of voice-recognition, not all accents are equal / Link

If Alexa was Southern (parody video) / Link

Apple admits its new $350 speaker can leave permanent white rings on wooden surfaces and furniture / Link

Podcast Roundup

Voicebot Podcast: Cathy Pearl, author of Designing Voice User Interfaces / Link

VUX World Podcast: Jeff Smith, author of Reactive Machine Learning Systems / Link

Execution Podcast: Lauren Golembiewski, CEO and co-founder of Voxable, a conversational interface agency / Link

Homie and Lexy: Homie + Nest Cam (I find this podcast confusing because the voices and the personas are just...wrong [and not in a funny way], but the concept is so close to being good.) / Link

Want to get this in your inbox each week? Subscribe: http://tinyletter.com/locutius 

The singularity is the least of our concerns

Are you worried about AI creating some sort of super-computer-system that realizes it doesn’t need humanity and kills off all of us because we’re useless to it?

Don’t be.

Long before computers realize we’re useful only to ourselves, we’ll need to confront the reality that much of ML (machine learning) is created and implemented in a way that is at best ignorant and at worst harmful or even deadly.

Machine learning rapidly amplifies the problems we have in our own society, including institutional sexism and racism. In Wired, a computer scientist discovers that two large collections of images - collections created by Microsoft and Facebook expressly for training ML systems - display predictable gender bias in their depictions of activities like cooking and sports.

Read more: Machines Taught by Photos Learn a Sexist View of Women in Wired. Learn more: Pre-order Safiya Noble's book Algorithms of Oppression

5 little things that would make Alexa a lot better

I read it so you don’t have to click. Just see if you agree with these criticisms of the Amazon’s Alexa or not…

Custom wake words

A male voice

Deeper routines (aka Nest + music support)

Better podcast controls

Improved touchscreens

NUMBER 5. Argh. The touchscreen on the Echo Show, which is an expensive device, feels about as responsive as my kid’s $40 Kindle Fire.

Read more about the reasoning behind the list on CNET. Check the specs on the Echo Show on Amazon

Share your prototypes without ‘live demo’ danger

Look, there are plenty of tools out there for prototyping conversations with Alexa. What makes voice gram different (better?) is that you can save a recording of your demo and share it with other people, saving you the danger/hassle of a live demo not working.

Try it: Voicegram from Sayspring Read more: Voicegram on Techcrunch

The Financial Times gets it. But do they “get it”?

“A couple of years ago, Virginia Eubanks, a political science professor in New York, suffered an experience that is the stuff of nightmares: her partner was violently assaulted and required major surgery. The couple had health insurance but when Eubanks tried to settle the $60,000 bill, the claim was rejected: a computer program had “red flagged” the case.”

Some day it’ll happen to you. And will you know what to do?

Read more: When algorithms reinforce inequality in the Financial Times Magazine

Locutius Links

Google Assistant gets music-powered alarms and better Netflix controls / Link Canary lets you see your front door or nursery with a simple command / Link

Until next week!

Adios, Abi Jones Editor, Locutius

Is there something I missed? Reply to this email with a link! Want to chat? I’m on Twitter at @jonesabi

Algorithms of Oppression

If you don’t know how search engines reinforce racism, then pre-order Safiya Noble’s book and get yourself reading. Pre-order now on Amazon.

Why Alexa didn’t light up during the Super Bowl

The word “Alexa” was uttered 10 times during Amazon’s Super Bowl ad, but your device didn’t light up once. It’s likely that Amazon used their patented “Audible command filtering” technique to keep devices from waking up. Read more about it on Bloomberg Technology.

How I coined the term “Open Source”

It’s been 20 years since the term “open source” was coined. This story is a terrific example of the psychology of mirroring. It’s a good reminder of what makes Voice Interface elicitation research so difficult, too. Read Christine Peterson’s account on opensource.com.

Wal-Mart’s robots roam the aisles in 50 locations

New job: robot chaperone. My favorite part of this way-to-short article? That they first used a car as a concept model (turn signals and all) and then realized that nobody expects a car inside of Wal-Mart. Context matters! Read more in Technology Review.

The startup Amazon hired to fix voice’s fatal flaw

You’ve heard the stat: after two weeks, only 3% keep using the voice apps (aka Amazon Skills) they download. Pulse Labs, a startup that tests voice skills with users, and then analyzes where things went wrong, just got a $2.5 million investment from the Amazon Alexa Fund and Bezos Expeditions. Read more about what they’re doing on Fast Company Design. Think you don’t need help? Read Nobody cares about your Amazon Alexa Skill.

Locutius Links

Google

“Just getting more UXers assigned to projects that use ML won’t be enough.” Read about Google’s PAIR (People+AI Research) Initiative / Link

Method Podcast, Episode 8 feature’s Google’s Margaret Urban / Link

APK insights: custom Google Assistant hotwords / Link

Amazon

Send messages via SMS via Alexa (if you have an Android phone) / Link

Amazon doesn’t want ads on Alexa / Link

Google and Amazon

Search Engine Land: Google Home gains ground on Amazon Echo / Link

Links not about Google or Amazon

Language matters! You can change the narrative about youth of color these 3 ways / Link

This guy graphed the origins of English words and blew people’s minds / Link

An “Open” Voice Assistant is at 4x its funding goal on Kickstarter / Link

Voice UX Podcasts / Link

Until next week, Abi Jones Editor, Locutius

Is there something I missed? Reply to this email with a link! Want to chat? I’m on Twitter at @jonesabi

Doppelgänger

If you read just one article about voice interfaces this week, read A Holocaust Survivor’s Digital Doppelgänger, from the New Yorker. Doppelgänger captures the moment where Eva Schloss, a childhood friend of Anne Frank, experiences herself as a voice interface and avatar.

More info about the exhibit:

“New Dimensions in Testimony” at the Museum of Jewish Heritage website

A Child Asks Holocaust Survivor Eva Schloss Questions (YouTube)

5 sources of bias in AI

5 sources of bias in AI, including at least one that you haven’t thought about even if you have read Weapons of Math Destruction

Amazon Echo Spot

Amazon released their first Echo Spot ad. Until now, it seemed like a bedside alarm clock…with a camera? Exactly what you want next to your bed, right? The ad takes it in a different direction, and is 95% great.

The part that will only bother people who like me, are annoyed by logical flaws in commercials: the grandma says “Alexa, call home” to call her grandkids. But, if she’s calling home, then what word does she use to describe the place where she lives? Wouldn’t “call home” call your own house?

I get why they did it for the ad, but it feels like the family moved grandma out of her home and now they are occupying it. Weird.

Reviews: CNET: Amazon claims Echo Spot is all about togetherness, not spying / LINK Echo Spot review: If it wasn’t for that creepy camera this would be our favourite Echo yet / LINK

This week in Machine Intelligence

It’s the 90’s all over again. Except not because I can’t keep a pair of white Keds clean. Instead, I’m linking to a PDF of a magazine from 1999.

“Mixed-initiative interaction refers to a flexible interaction strategy in which each agent (human or computer) contributes what it is best suited at the most appropriate time.”

I am sharing this gem because I saved 18 not-great tech blog articles on the subject just this week and I wanted to save you from reading those. A framework for human interaction with machine intelligence-driven systems already exists. It has existed for 20 years. Let’s take advantage of that and inform ourselves, shall we? Read it: Mixed-initiative interaction in IEEE Intelligent Systems from Sept/Oct 1999. Note: I do not expect that random people writing about their bot startups know the term ‘mixed-initiative system’. But that gives you a bit of an advantage over them, eh? And here’s the Google Scholar search for ‘Mixed-initiative interaction' in case you wanted to learn more about the concept, find some authors, etc.

“...We project our own anxieties and when we ask: “Will the robots kill us?”, what we are really asking is: “Will we kill us?”

Our fear of AI is more about feeling irrelevant than being scared of killer robots, says anthropologist Genevieve Bell. Not me. I don’t say that. I’ve seen this video and my concern about killer robots is completely reasonable. In other ‘this is why we can't have nice things' news, Tim O’Reilly says everything I have been thinking about (and trying to warn y’all about) when we curate content with algorithms. | Media in the Age of Algorithms

This week in personal assistants

I discovered that ‘Okay, Goofball” is a 100% successful Google Home hotword. In case you find it tiring/awkward to say ‘Okay, Google’. Should machines be funny? Maybe. Should my car be funny? No. Will Google Home and Alexa stop stepping on their own jokes and just pause for 2 seconds before the punch line? Arghgh. | Fast Company looks at the humor behind the computer Microsoft is working on its own ‘Home Hub’. Family desktop sounds...great. I mean it. My current shared-calendar consists of a Dropcam looking at the whiteboard in my kitchen. | Microsoft Central In Alexa-land, Amazon is working with Intel on ‘off-brand’ speakers and here’s your weekly article on an Echo with a screen. And they have a bot-building UI that’s in limited preview. You have to find your AWS account # to sign up, just a warning. Good luck.

7 interesting or beautiful things. Okay, 1 beautiful thing and 6 other things.

1. This is a rather lovely visual introduction to machine learning | LINK

2. Interested in the anatomy, physiology, and culture of language differences between men and women? Check out this book chapter on what shapes our speech (it’s open to comment and very much in draft form). | LINK

3. Speaking of gender - Google translate has a gender problem. | LINK

4. Subtle ways your digital assistant might manipulate you (all of these things are already happening, this article is just the version where Alexa starts writing funnier tweets than you do.) | LINK

5. Google has a “hand-fed AI,” which is how journalists describe “supervised training for machine intelligence” Yeah, the actual terminology is just not as catchy. That’s why I don’t write linkbait. | LINK

6. Opinion: The dangers of faulty, biased, or malicious algorithms requires independent oversight. From Ben Shneiderman, who is also in that magazine PDF I linked to above. | LINK

Algorithm ethics, the McRib, the UX of bots, and the real costs of making friends with robots.

First, the heavy stuff.

If you work in a field that uses algorithms in any way (hello fellow Googlers - that’s you! Oh, you’re not a Googler? Guess what - you’re included too. It’s like Oprah and Bees here.) - please review the proposal for Principles for Accountable Algorithms. The only thing I don’t like about the proposal is the title. The algorithms can’t be accountable - it’s like saying ‘accountable cars’ - but the people who make them can be.

Bots The McRib is back and McDonald's has a bot in Facebook Messenger that’ll help you find it. Working on a bot? Read Chris Messina’s Thoughts on the UX of Bots. Then consider each of the limitations he outlines as areas for opportunity. How could your conversational UI improve (or change the way we think of) discovery, utility, and universality? Bookmark it (or save it somewhere in case Medium disappears) and then pop back to it in 2020. What’s changed? What’s the same?

Home assistants Linglong, a company in China, has their own version of the Amazon Echo and it sounds like people use it to do the same thing Echo owners do: listen to music. | WIRED

Conversational UI

"When Ashok Goel, a professor at the Georgia Institute of Technology, adapted Watson to have a female voice as an informal experiment in how people relate to conversational machines, his students couldn’t tell it was a computer."

There are so many things wrong with that quote I don’t even know where to start. So... WHY DO ALL ASSISTANTS SOUND LIKE WHITE WOMEN? | NYT In hopefully-less-sexist-news, I’m one of those people who turns on the captions on Netflix because I can’t always tell what people are saying. Good news for me: Artificial Intelligence can lip-read better than a trained professional. It turns out that machine learning can make human-to-human conversation better, too. | Technology Review OMG, everyone thinks everything is a metaphor for AI. Alternatively… OMG, everyone thinks Arrival is a metaphor for everything.

Stephen Wolfram (yes, that one) was an advisor for Arrival and thinks that communicating with AI is a lot like communicating with aliens. | WSJ

(it’s a video. I know. Ugh.)

Did Arrival make you interested in linguistics? The superfab @GretchenAMcC offers a list of things to check out if you’re interested in linguistics. I signed up for the Intro to Linguistics class on Coursera - it starts on Dec. 19th. Join me!

She also put together a Twitter bot that tells you your Harry Potter name - it’s interesting to see her understanding people’s use of it in real time.

Also, links.

If we want to do better than Siri, we need the tools to get us there. | LINK

Should VUIs have better attribution support? Yes. | LINK

The code I’m still ashamed of. | LINK

What Neural Networks, Artificial Intelligence, and Machine Learning Actually Do | LINK

When bias in product design means life or death | LINK

AI by another name (1. it’s from 2002. 2. I prefer to say ‘Machine Intelligence’ - not AI) | LINK

Challenges and opportunities of ambient computing | LINK

Together in electric dreams: Robots replace family love for China’s lonely elderly. (The quote that gets me: “If the robot reminds me to take a medicine which is not the same as what I remembered, then I’d only trust myself.”) | LINK

The real cost of making friends with robots. (Also, I have had to explain to my child that our car and Alexa don’t have feelings. We’re still nice to them, but in the way we’re nice to bugs. Most bugs. Have you told your kid that robots are not our friends?) | LINK

Until next week, Abi Jones Editor, Locutius

How to talk to aliens, bots that fight harassment, controlling computers with your brain, and ‘Photoshop for Audio’ Bots

So what if bots were posting tons of fake news stories during the election. Most people had their minds made up anyways. Or did they? Mike Caulfield wrote a terrific blog post that delves into the psychology of what we’re doing when we share on social media.

This is the best article I read about using bots to strike back against racist posts: Why online allies matter in fighting harassment.

Trim’s bot chats with Comcast to lower your bill. | LINK

How to talk to Aliens Go see Arrival before you read these articles. I don’t want to be responsible for spoiling it for you.

Popular Science asks “If aliens invaded, how would we talk to them?” | LINK

A linguist chimes in with “What Arrival Gets Right About Talking to Superintelligent Noisesquids From Space” | LINK

In 'Arrival,' a Linguist is a Movie Hero (the first movie review I've ever seen from Ben Zimmer). | LINK

Voice interfaces

Why does Alexa sound great? Aside from voice talent, Alexa uses pre-recorded phrases for its common responses, including weather forecasts. But it’s an unscalable solution as the skillsets of assistants grow, so companies use parametric and concatenative TTS for their services. Unfortunately, the lack of prosody in concatenative TTS makes it an unpleasant interface.

Google’s DeepMind has a new solution for generating speech with computers. Their approach, called WaveNet, records waveforms from human speech, then uses a neural network to generate synthetic utterances. I encourage you to check out the DeepMind blog post on WaveNet and even if you’re not interested in recurrent neural networks, scroll down to the bottom and listen to the samples of speech generated with concatenative, parametric, and WaveNet approaches.

Yeah, another section on WaveNet. Maybe you didn’t click the link above. So read the article, then scroll down to the bottom and listen to how WaveNet generates non-speech sounds, like breathing. The team also wrote a paper if you’d like to read that instead (it doesn’t include audio files). | WaveNet: A Generative Model for Raw Audio If you have an interest in the future interoperability of voice systems, take some time to check out the Voice Interaction Community Group, part of the W3C.

Adobe demos “photoshop for audio,” lets you edit speech as easily as text. Yeah, nobody’s going to use that for evil.

Brain interfaces

Yes, brain interfaces. A paralyzed woman has learned to use a brain implant to communicate by thought alone. Previously she’d communicated via eye-tracking software. This is the first time a brain–computer interface has been used at home in a person’s day-to-day life. Her doctors imagine she’ll be able to use it to control items in her home, like her TV, but “My dream is to be able to drive my wheelchair,” says the woman.

Home Assistants

iFixit conducted teardowns of Google Home and Amazon Echo. They’re both worth a read, whether you own the devices or just want to see what they’re made of.

Gartner makes 10 predictions about technology, including that by 2020 you’ll say more to a machine than you do to your spouse (note, this might already be true for people who work on Voice UIs).

Amazon’s Alexa can now send texts via AT&T. Yes, you have to invoke AT&T to do it, just what everyone wants to be reminded of - their phone carrier. | LINK

Machine Learning & Artificial Intelligence

Will humans and AI become inseparable? Manuela Veloso, Head of Machine Learning at Carnegie Mellon, thinks so. Learn about what’s been holding us back over the last 50 years of AI research. | LINK

Google is making natural language processing more accessible to enterprise users | LINK

Google’s QuickDraw illustrates (ha) why you need diversity in training data | LINK

The Google Brain team is focused on equal opportunity in machine learning | LINK

More interesting things

Chinese scientists grew an ear on a man’s arm. I’m not sure why this was saved to my links for the week, but it is pretty stunning, so you should check it out. | Wall Street Journal

Have a great week! Abi Jones