Categories
Sixty Seconds of AI

Adding Value

We don’t have any analysis or commentary today, just a note. We’ve been posting every weekday for a few weeks now, and going forward we’ll be making a tweak to our schedule.

From now on, we’ll only be posting every other weekday. That will mean you’ll see posts Monday, Wednesday, and Friday one week with Tuesday and Thursday the following week, and, then back to MWF.

That’s about half the posting, but here’s why: Just because we are following AI, VUI design, speech, NLP, and everything else under the probabilistic and modeled HCI sun, that doesn’t mean there’s something especially interesting every day. We started with every day because we thought that starting with a lot and dialing it back was easier than starting with fewer and ramping up, and now it’s time to dial it back.

Even though this blog was intended to take about 60 seconds to read (hence the name 60 Seconds of AI), that’s still a minute (or two sometimes!) we don’t all have every day. We hope that means that it will be easier for you to find value in what we post here.

Thanks for reading!

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

FTC’s Chatbot

Dealing with some administrative stuff today, I noticed that the Federal Trade Commission has a chatbot on the site identitytheft.gov.

screenshot of top of identity theft dot gov site showing chatbot opt in button

Here’s what you see when you click on FTC’s Chatbot in the upper corner of the page:

Screenshot showing FTC chatbot from lower right of page

Identifying a case where I knew what it would tell me, I clicked My personal info has been exposed. I just wanted to see what would happen.

Why it’s interesting

When you click that button, it asks you if your personal information has been exposed. If you choose the option Not Sure, rather suddenly the bot sends a series of messages in rapid succession, 7 in all, ending with asking if it’s answered your question.

FTC Chatbot asking "Did I answer Your Question?" with Yes and No buttons to reply

The thing is, even for a very fast reader, it would be physically impossible to have read the responses given, not to mention the 9 links to outside sites it recommends visiting in the text. The experience is much like watching a vending machine dispensing a candy bar.

Why we care

The actual experience of identity theft is anxiety-inducing for many people, not least of which is the exposure to a great deal of information that is of entirely no interest or value unless you are taking it in under duress.

It stands to reason that if anything an experience that intends to simplify the experience of accessing information about what to do about the theft of one’s identify ought to be gentle as well.

Links

https://www.identitytheft.gov/Assistant

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

Hands off the Wheel

As I read through a discussion of how search and local law intersect in Noble’s Algorithms of Oppression, I was reminded of a discussion a friend and I had over dinner the other day.

He’d suggested that artificial intelligence is responsible for the content moderation at places like Facebook. Citing the apparently instant start to streaming for services like Facebook Live, he posited there was no way his streaming could be moderated in any way, at least not before he’d had the chance to broadcast. Were he to share something that violated policy, everyone would see what he broadcast before it was discontinued.

Since I have done a lot of manual work in AI, and it’s well-established that every company whose primary product is user generated content relies on an army of manual content moderators working around the clock, I suggested it was actually possible there was no AI in the system at all. This seemed impossible to him: a numbers game, at scale.

Why it’s interesting

We were quickly able to agree that the numbers scaled down considerably for his example of streaming video: on any given day, regardless of the number of daily active users, nearly none would be streaming live that day, and even fewer would be streaming at any given moment.

This is the important fact I wanted to discuss: Even still, with billions of daily active users, any streaming media site would still need tens of thousands of moderators monitoring incoming streams. Those workers are rarely considered when we’re talking about something that “just works.”

(Caveat: I don’t have any idea how many users use streaming video or how many content moderators there are, just that the services can’t exist without them.)

Why it matters

My friend was happy to attribute beheading- and revenge-porn free experiences using social media apps to AI. The very existence of the service depends on high-quality human judgement to make decisions though.

The people actually making the product feel safe and fast and amazing, are real. Humans create the emotional power of what’s marketed as AI though manual labor, mostly in the global south. Especially when, in the recent cultural narrative of The Social Dilemma, the narrative has started to change to frame the product being you. This is an alarming and emotional message.

The people doing the emotional labor and the toughest part of the job being invisible isn’t new. We can’t forget it though.

Links

http://algorithmsofoppression.com/

The Social Dilemma on Netflix: https://www.netflix.com/title/81254224

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

Working Memory

Do you read new periodicals as they come in, and when you don’t have time before a new one comes in, put them in a pile somewhere? Me too.

Occasionally I find myself reading something from years ago and not realizing it at all, as happened yesterday, when I got to the end of the April 2019 issue of Harper’s that was sitting on my kitchen table. That’s where this comes from, the very last line of that month’s Findings column:

People have better working memory if, in their native language, meanings remain, until the end of the sentence, unclear.

(For those unfamilar, Findings reduces complex papers and research to a single line, context free, for a bewildering comic effect.)

Why it’s interesting

When we shared this sentence with a colleague who’s a VUI designer, she balked. “Just reading it was driving me mad,” she said.

Why it matters

Just the other day, talking with a colleague, we discussed the state of so-called “multimodal” design for voice, and our cognitive load threshold. In terms of the pandemic, it made it easy to argue that critical elements in the design that might seem obvious can be missed entirely.

We had experienced this personally, not with voice experiences on devices with screens, but in everyday life. Losing track of time and not joining a meeting back-to-back with the current one. No longer seeing reminders for any app. Emailing not the next day, but the next week.

If you miss something entirely and don’t know it, there’s no recovery.

Personally, our style in both writing and cogitating is digressive, and we’re more than comfortable with collapsing many threads into one, like a braid, or a perhaps a bird’s nest. Perhaps it’s time to consider that might require more resources than can be easily spared right now.

Links

https://harpers.org/archive/2019/04/letters-april-2019/

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

Gadgets

I was about to close the Sunday paper for the day when I stumbled on a light bit about gadgets and I happened to notice this text jumping out at me on an unread page:

Although any white-noise machine can do the trick, I have found that a smart display, like the Google Nest Hub (about $90), combines several useful bedside gadgets in one box: an alarm clock, a digital picture frame and — above all, for me — a white-noise machine.

Wirecutter, The New York Times, 3/14/2021

Why it’s interesting

Not to overstate it, there’s no mention of cloud technologies or AI at all discussing the device, not to mention speech recognition, for a device whose only functional interface is voice.

It’s a gadget that does a number of things. This is the most notable aspect.

Why it’s important

I’m not sure whether something that’s “just a gadget” is a good thing or a bad thing. Just a couple of years ago it was hailed (by its maker, and the industry at large) as a sophisticated technological breakthrough that is the future.

Does that mean the future is here? Or perhaps, as I suspect, that the paper of record in the United States is normalizing speech as a totally average and uninteresting aspect of the device, which, as the overall discussion of products and functionality in the piece suggests, is convenience as invisible as a microwave.

Links

https://www.nytimes.com/2021/03/09/realestate/nine-tools-for-better-longer-sleep.html

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Editorial Sixty Seconds of AI

Why yes, I would like a knuckle sandwich.

I caught up on Far Side cartoons from the page-a-day calendar my brother got me as a gift. It’s interesting learning how much something you enjoyed as a kid is not only no longer fun can be quite upsetting once you’ve grown up in the culture that produced them – Whee, lynching jokes! You know, for kids!

On the other hand, it’s interesting to interpret things later in ways you never would when you first saw them. Here’s a recent one that jumped out at me (pardon the hasty phone scan):

Far Side comic strip of an alien badly disguised as human, confronted by violent men. Caption: Why yes, I would like a knuckle sandwich.
c. Andrews McMeel Universal

The punchline is the alien apparently understands English well enough to interpret an offer, but not realizing the “knuckle sandwich” offered is an idiom meaning a threat of violence, eagerly accepts it as an opportunity to make contact.

Why it’s interesting

As simple as this joke is, it’s actually much like the most advanced NLP systems we use today. I can guess that the very best speech recognition available from any of the enterprise vendors or big tech companies would be able to interpret a lot from the offer we infer our alien friend can guess phrasal intents like OfferAccept, or that “why yes thank you” is a synonym for “yes.”

On the other hand, this kind of crawling toward the Uncanny Valley only gets more frustrating as our progress shows how much we fail to get out of truly natural language. We can think of many examples of how our NLP especially for speech misses the forest for the trees.

Why we care

Ray Jackendoff in Patterns in the Mind uses a metaphor of a robot that understands only one language but speaks another to highlight two strange facts about language: Our knowledge of language is individual, but requires world knowledge, and using language requires reciprocity as a matter of cognitive fact.

It’s easy to imagine that robot, and how different the robot is from us makes the image stick. What’s funny to me today is that reading this old comic strip from the 1980s makes me realize as I think about it that we actually have built that robot again and again.

Our very best NLP systems don’t understand anything about language at all. As DNNs and other techniques get more complex and abstract, and processing power continues to expand, we can expect the need for VUI design to continue to expand, hand-crafting the platforms that turn what machines understand into something humans do.

Links

Patterns in the Mind on Google Books: https://books.google.com/books/about/Patterns_in_the_Mind.html?id=keMJX6BOppAC

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

Women In Voice

We don’t have anything of substance to pontificate on today because we managed to be so busy with work we didn’t even catch up on email until just now.

Still, we’d be remiss if we didn’t mention Women In Voice. I just wrote to the organizers of their German contingent to thank them for hosting a talk from Marion Mulder last week, entitled The (non)sense of gender-free AI.

Why it’s interesting

Women, trans, and gender-non-conforming folks talking about issues affecting women, trans, and gender-non-conforming folks is always important in tech.

Women In Voice is an organization that aims to advocate and champion their work. We’ve had the chance to join a number of their events and can’t recommend them enough.

Why we care

Especially with AI, where the personal, cultural, and even environmental impact of decisions by a very small number of practitioners, more often than not there’s a larger social impact than is realized by the general public until those disproportionately affected by it (or left out of it) have been talking about it for a good long time already.

Links

https://womeninvoice.org/

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

Center of Excellence

One of the reasons Dee VUI doesn’t offer a course in Conversation Design or any related subject is because we do the work. If you do the work, you know that if you have no idea how to become a conversation designer and then knock on our door after doing a web search, we probably can’t help you. If you haven’t read any of the books, let alone zillions of web pages and want to ask someone to show you, you don’t have the skills you need to learn yet, especially if you lead with your wallet.

We do teach classes, but one of the reasons why this isn’t a business we’ve developed is because of, well, the business. If that’s your business, all of the touchpoints of presentation need to project what you teach, and this is a lot of work in and of itself. Being at the time of this writing being just me, Damien, we’re too busy doing design and research to launch a new business and all that entails.

To wit: I stumbled on this poor page’s content design and execution searching for something else, and now they’re the subject of a blog post about what you’re telling us when you’re not doing the work:

Marketing Webpage With Dead Embedded Video Link
The “excellence” term can be misleading.

Why it’s interesting

We don’t wish to criticize this company selling education and personal growth courses to users and job postings to companies, but if this blog post is showing how they disseminate best practices rather than being bogged down in operational processes, they’re not doing a good job at either.

As for this image, it’s a screenshot from a blog post marketing their services. They chose to specifically embed a video from another site, and the link is dead. From the looks of it, the video was intended to entice the reader to engage with other content linked on their site.

I’m reminded of the The Maintainers I read about in The Innovation Delusion recently, an organization is dedicated to the idea of keeping things going as a requirement for moving forward in anything. You might have taken a class or bought advertising from this site, at least until they showed you they can’t keep the lights on for their website, which is where their courses and ads are.

Why we care

There’s an inherent need to have just enough maintenance, just enough QA, and just enough auditing and evaluation to know you’re adequately performing generally. Key Performance Indicators, are just that, key. The option of monitoring KPIs is a moot point if your car runs out of gas because no one looked at the gas gauge.

I totally get not wanting to be bogged down, but when this is if your goal is “just enough.” You can obviously do more than that, but you can’t do less. AI means these details matter more than ever.

Links

https://themaintainers.org/

https://leevinsel.com/the-innovation-delusion

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

Kazuo Ishiguro’s Artificial Friend

It’s a little embarrassing to admit that I’ve never read Ishiguro. We all have gaps, right?

I will admit that I have never picked up any of his books because when I was a kid I saw a bit of Remains of the Day on TV and thought it was a war movie, and I hate war movies.

(Dumb, I know, you don’t have to tell me that it’s the setting for the beginning of the movie but it’s not about WWII)

Anyway, he has a new book coming out called Klara and the Sun.

Why it’s interesting

The titular Klara is an “Artificial Friend” which is some kind of AI-powered affective robot that you take home from a store and learns about you.

Why we care

This is a bit of a leap since I’ve never read the guy’s books, but he has a reputation for emotional resonance. His best known and loved work (arguably) is Never Let Me Go, a book about clones raised to “donate” their organs. From what I’ve heard, the power of the story is the uncomfortable subject and marginal existence of the clones is powerful.

That makes me excited to read what he does with an embodied AI that’s a real life companion.

Links

https://www.penguinrandomhouse.com/books/653825/klara-and-the-sun-by-kazuo-ishiguro/

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com

Categories
Sixty Seconds of AI

Can’t Have It Both Ways

Here’s a bit from a front page article from the NYT’s Sunday Business section.

I look forward to working with a true believer in Robotic Process Automation so I can better understand the perspective that leads people to talk about people like this (emphasis mine):

Jason Kingdon, the chief executive of the R.P.A. firm Blue Prism, speaks in the softened vernacular of displacement too. He refers to his company’s bots as “digital workers,” and he explained that the economic shock of the pandemic had “massively raised awareness” among executives about the variety of work that no longer requires human involvement.

“We think any business process can be automated,” he said.

Mr. Kingdon tells business leaders that between half and two-thirds of all the tasks currently being done at their companies can be done by machines. Ultimately, he sees a future in which humans will collaborate side-by-side with teams of digital employees, with plenty of work for everyone, although he conceded that the robots have certain natural advantages.

“A digital worker,” he said, “can be scaled in a vastly more flexible way.”

Why it’s interesting

The very use of the word “work” implies a human is involved, so but let’s leave the incorrect use of “worker” aside.

Bots do a single thing in a single way until they’re thrown away or refactored. They are purpose built and very very expensive to build, maintain, and update.

If you want to talk about research, development, customization, extras like encryption, data storage, security, bandwidth, and regular tuning as part of the process, all of which is done by specialist experts (i.e. Dee VUI) now it’s even more expensive.

It’s a cool marketing pitch (if you’re into, you know, our robot overlords) and I’ll thank the guy in advance for keeping us in business for when it doesn’t work and they don’t know how to fix it, but sheesh.

Why we care

I can get why you’d want to call a few hundred lines of code that call some proprietary software packages and send data to and from an API a “worker” instead of “software” if you were a primary investor in a company that sold that software and services that does this, as Kingdon does. That way you can imply that you sell something of equivalent value to abstract and emotional work.

It doesn’t though. The only way you can even call something “work” is because it requires physical, emotional, or mental labor, usually all of these in multiple ways. On the other hand, comparing people to line of code, and suggesting equivalence in value and kind, is creepy and dehumanizing.

Unpacking this a bit, it’s definitely a fact that many (most?) businesses rely on outdated machinery, processes, and practices because they invest as little as possible in maintenance, infrastructure, and upkeep. We live in a world that prioritizes short term gains, so that’s not surprising. It’s a business decision. That’s fine. If there are parts of their businesses that are inefficient and costly as a result of business decisions, it’s a ugly to characterize this as a labor problem.

Kingdon seems to want to have it both ways: He wants to call software “workers” by using a bot as a metaphor for a human in order to play up its value, but he want to to identify the value of the bot as inherently better than a human precisely because it’s not human.

Scapegoating human labor as a business problem to be solved isn’t the way forward. As AI practitioners, we should be looking to celebrate humans by maximizing the value of their contributions and diversity.

Links

https://www.nytimes.com/2021/03/06/business/the-robots-are-coming-for-phil-in-accounting.html

https://en.wikipedia.org/wiki/Robotic_process_automation

Let’s have a conversation

Let us know what you think! Write us at:

sixtyseconds@deevui.com