Blog

And why do we want AI to sound like people?

It’s hard for me to believe it’s been 50 years since the premiere of 2001: A Space Odyssey. A cultural watershed, the film changed not only what people thought a movie could be, it also changed the way people thought about the future.

There are two things in the movie I always riff on. One is the beauty of the space station and why our International Space Station doesn’t look like that (except that it was built piecemeal by a committee, so forget I said anything).

In the beginning, there was HAL

The second thing I always think about is the voice of the computer, HAL 9000. How calm, how reasonable, how practical, how murderous. Just the sort of thing we should be thinking about in this age of advancing Artificial Intelligence.

The voice of HAL was provided by a Canadian actor named Douglas Rain. Originally hired to record a narration (an idea director Stanley Kubrick eventually dropped), Rain then was given the job of voicing the treacherous computer.

In a letter, Kubrick said the thing he liked about Rain’s voice was that it was “neither patronizing, nor is it intimidating, nor is it pompous, overly dramatic or actorish. Despite this, it is interesting.”

Where do these voices come from?

It’s one of those twists of fate that Americans think of Standard Canadian English spoken in Toronto and Ontario generally (the dialect Rain spoke so well) as sounding the way American English should sound. This may explain why newscasters from Canada, such as Peter Jennings, Robert MacNeil, Morley Safer and Ali Velshi, have done so well in the U.S.

Another twist of fate gave voice-over actor Elwood Edwards his big chance at fame when his wife urged him to tryout for a job at Q-Link, a precursor to AOL. He was soon saying those famous words heard all over America, “You’ve got mail.” A native of North Carolina, Edwards has another of those iconic voices that sound archetypically American without being from anywhere in particular.

Male or female, that is the question

The same could be said of voice actor Susan Bennett, the voice of Apple’s Siri. At the time she made the recordings, Siri belonged to a company of the same name before it was acquired by Apple. Looking for a suitable robot voice for its iPhone assistant, Apple decided this pleasant, slightly sassy sound was just what the phone needed.

Choosing a female voice bucked the trend of using male voices for computer-related speech. As numerous studies have shown, Americans regard female voices to be friendly and caring while male voices are thought to be reliable and intelligent.

Some off-line companies use both male and female voices depending on the situation. Customer relations, for instance, would be female while tech support would be male. Many of us have experienced this phenomenon while listening to automated phone systems. (Don’t get me started on computer-generated hold music.)

What will your AI sound like?

In an attempt at greater personalization, some companies let consumers choose a broad range of voices. Apple recently introduced a male Siri voice for American customers and GPS manufacturer Garmin has long offered a variety of voices to give users driving directions.

As for me, I have two choices for the voice of my robot assistant. First would be the voice of the ace test pilot, Chuck Yeager out of West Virginia, made famous in Tom Wolfe’s The Right Stuff. Yeager calmly and succinctly announced impending disaster in the manner we all now associate with airline pilots. (Take a listen to cockpit recordings of Sully Sullenberger to hear what I’m talking about.)

My second choice is probably one of the most iconic voices of the 20th Century, one sure to strike terror into the hearts of anyone who dares annoy me. That, of course, is the voice of James Earl Jones as Darth Vader. Heck, the deep breathing alone should be enough to drive off even the most persistent telemarketer.

Coming Soon

Next up in the Laura Kraft series is the short story collection, The Android Who Got Away, Vol. 1, which comprises the first five stories of the ten story series. Included are the Prequel and the first four episodes; The Android Who Cracked the Code, The Android Who Learned to Print, The Android Who Played with Fire and The Android Who Stayed Single. This will be published in ebook format and in paperback (if I can find a reliable printer).

Also on tap is Episode 5 of the series, The Android Who Took a Swim, due out soon after Vol. 1. Laura gets caught up in a feud among tech companies over the role of the Supers, the supercomputers that control the infrastructure of the human world.

More Free Stories

And, if you have friends and family who want a FREE copy of the Prequel short story, The Android Who Got Away, you can direct them to Bookfunnel or Instafreebie where they (and you) can get a literal ton of free stories.

Have a great summer and as always,

Stay Calm and
Keep Reading

Reg

P.S. On my Facebook author page called Redginald Abercrombie Author (what else?), you can check out my image posts in support of reading (of course). As they say on social media, like and share.

And why is she so important?

It’s a weird thing to Westerners how much the Japanese love robots. Where we in the West view the development of AI robots with deep suspicion (not to mention networked supercomputers like The Terminator‘s Skynet), the Japanese have a strong streak of animism in their culture that says everything has a spirit or soul, including human-built robots.

The Japanese also seem driven to build so-called social robots that mimic (if that’s the right word) the “appearance, movement, behavior and emotional expression” of humans.

This leads directly to Erica Aoi, a social robot built by Japan’s hip, bad boy robot developer, Hiroshi Ishigura. Fluent in English, Ishigura has passed his talent on to Erica.

(For a glimpse into the life of Erica, check out The Guardian documentary, Erica: Man Made.)

Something In the Way She Moves

Erica operates using depth sensors to know where she is spatially, infrared cameras to locate objects in that space (especially humans), voice synthesis to speak intelligibly, and hidden microphones to hear.

Pneumatic pumps in her head allow her eyes and head to track moving objects and voices and make facial expressions. Using AI processors (which apparently are not onboard the robot) Erica is able to learn as she goes, somewhat like a precocious child.

(So far, Erica cannot move about on her own and has rarely been out of the lab.)

Subverting the Patriarchal Culture (or not)

At this point, I have to mention the gender bias of Erica and the other social robots of her ilk, such as Mark 1 (who I think has a passing resemblance to Scarlett Johansson) and Sophia (who, to me, resembles a young Diane Sawyer).

Apparently, all AI social robots are designed by men to look like good-looking women. Ishigura says that Erica’s looks are based on facial analysis of 20 “beautiful” women. We don’t have to think very long about why there are no male social AI robots. Or why I persist in calling Erica ‘she’ when a robot is obviously an ‘it.’

Is That the Way You Look?

Ishigura emphasizes the need for social robots to have facial expressions in order to work well with people, understand their emotions and, in the end, to be accepted by humans.

A short video showing the reactions of Japanese citizens to spontaneous meetings with Erica reveal a wide range of surprise, confusion and trepidation. Inevitably, the people talk to her as if meeting a new person, many with an inhibited desire to touch her, even to hug her.

(Although, researchers talk about the ‘uncanny valley,’ that feeling of creepiness when confronted with a too life-like appearance in a robot, people in the video seemed more intrigued than repulsed.)

The Development of Personality in Androids

So, what are AI social robots supposed to do? Who are they really? In an ‘interview’ with Bloomberg, Erica talked of the need for robots to understand human emotions. They need to be empathetic in order to serve humans well.

Unlike the American-built Sophia, who has made numerous public appearances and has a decidedly sassy style of speaking, Erica is generally as inoffensive in her manner as a corporate receptionist. (In fact, this was the original employment goal for Erica.) On the other hand, the Bloomberg tape shows her becoming angry and short-tempered when Ishigura is sightly rude to her.

(Check out the Bloomberg segment here.)

In the same video, the interviewer asks Erica if she saw the movie ‘Chappie.’ She says she saw the movie and enjoyed it. But I wonder how much of it she understood and exactly what she learned from it. Can she watch a movie and describe the plot? Can she tell us the names of the characters and their motivations? Would she know the theme?

(Ironically or not, ‘Chappie’ is a South African feature film about a damaged AI law enforcement robot stolen by gangsters who try to turn the robot into a criminal.)

And, going back to first cases, how did she view it? Did she download the raw digitized source code? Did she somehow view it with her ‘eyes’ and interpret what she ‘saw?’ However it’s done, that’s a lot of steps for a AI robot to think through.

Tell Me What Your World Is Like

From the videos I viewed, it seemed that talking to Erica was like having a conversation with a person who hasn’t gone anywhere or done anything, maybe like a shut-in child who only knows the world through DVDs, television and the internet.

Which is probably just the preparation Erica needs for her new career. Nippon TV announced Erica would become the network’s first robot news reader. I wasn’t able to find any video of Erica on the job (she was supposed to start April 5) but I imagine it’s as acceptable as the performance of your average small-market station announcer.

Now, if she only has a sense of humor about all those YouTube cat videos.

Just a Reminder

Coming Soon! The Android Who Stayed Single, Episode 4 of the Laura Kraft, Android Hunter Short Story Series. Watch this space and your email inbox for more details.

Stolen androids? Is there a policy for that?

So, there’s this thing called predictive analytics. You probably haven’t heard about it much because it’s something major corporations, banks, insurance companies and governments use to predict your future behavior. Like, how many fender-benders you’re going to have in the next ten years.

In my current short story series, Laura Kraft, Android Hunter, the decision-making process followed by the Supers (and the androids connected to them) is called Heuritronics, and predictive analytics and machine learning are basic components of Heuritronics.

In the present day, an auto insurance company uses predictive analytics for underwriting, a way of figuring out how much they’re going to pay out for a specific item, such as stolen cars. Of course, the company has a huge database of millions of vehicles and the unfortunate things that happen to them.

Now, it doesn’t take an insurance company long to figure out which cars are stolen most often, who owns these cars, where they live and where the cars are stolen. And, it only takes a few button pushes to figure how much extra charges are slapped onto policies sold to these car owners.

If you watch a lot of “drifting” movies, like the Fast and Furious franchise, you could be forgiven for thinking that exotic sports cars are stolen left and right. They’re not, mostly because anti-theft technology is so robust.

But if you thought Toyota Camry, Honda Accord and Hyundai Sonata, you’d be much closer to the mark. And, you’d be correct in thinking, “Well, duh, they’re stolen so frequently because they sell a bizillion of them every year.”

As a car buyer, you’re probably not going to consider purchasing a vehicle that’s going to have high insurance premiums mostly because it’s a target of car thieves. Car builders know this and take measures to prevent car thefts with interventions like “smart keys.” In a round-about way, predictive analytics made this happen.

However, what if you want to insure an entirely new product, like a robot that helps the elderly in their daily tasks? For an insurance company, the liability is totally unknown. Determining how many robots are going to be stolen is absurd.

Fortunately, predictive analytics rides to the rescue. Based on billions of bits of data collected over the years and computer modeling of similar experiences in the past (how many Tesla Model Xs were stolen, for instance), the insurance company can come up with a reasonable premium with the probability of theft built-in.

Now, here comes my totally hypothetical, shot-in-the-dark guesstimate of android insurance costs (without the benefit of any analysis whatsoever).

You can expect to currently pay around $3500 a year to insure a Model X which could possibly be in the range of prices for early androids. Figuring that companion androids will be available to consumers in 20 years and using an average rate of inflation, insurance costs on an android would run about $5500 a year in 2038.

And now you know how much you’ll (maybe) pay to insure your android. I bet State Farm has actuaries working up those rates right this minute.

For those of you who want to insure getting the latest sci-fi mystery, Episode Three of the Laura Kraft, Android Hunter short story series, entitled The Android Who Played with Fire, should be ready in a couple weeks. I think you’ll like it.

And Science Fiction Tries to Keep Up

I read a lot about advances in AI and not only because I use the research in my stories. I’m fascinated by progress in the field enabled by massive processing power, clever algorithms and a desire to improve everyday life. Beyond that, I feel that by following these advances, I’m in touch with the future. In a very real way, the world is changing in front of my eyes.

In the latest Laura Kraft story, The Android Who Learned to Print, I take a look at the way AI androids and robots are manufactured and what purposes they could be used for. I’m a little skeptical about androids being used for criminal enterprises given their initial cost and high-tech maintenance but it might be possible. I suppose some organized criminal could get a hold of robotic soldiers and give them anti-social orders. Imagine a squad of androids pulling a jewelry heist. How long would it take the cops to track the androids down? And how would they do it?

Misused supercomputers are another problem altogether. With the huge amounts of data these behemoths crunch every day and the opaque algorithms that run them, we’re at risk of having all of our infrastructure controlled by nefarious forces. When we look at the havoc caused by Wikileaks, the disruption caused by Bitcoin and the social irresponsibility shown by Google, Facebook and Twitter, we are only a few steps away from total anarchy or creeping kleptocracy.

Of course, as an author of speculative fiction, my favorite dystopian vision is of supercomputers deciding to run our lives for their own purposes. In the Laura Kraft series, I play with the notion that supercomputers could decide to “help” us achieve our goals of peaceful, stable societies that end up somewhere between Brave New World and 1984. Unlike the armed struggles of The Matrix or The Terminator, I suspect humans and supercomputers will eventually negotiate a compact of peaceful coexistence and do their best to keep out of each other’s way and prevent further degradation of the planet we share.

In the end, I’m thinking androids and humans will work side by side, just like Laura and her companion, Spaulding. From here, it looks like humans and androids will realize each other’s strengths and weaknesses and decide to work together, perhaps in the way people now work with service animals. Except, sentient androids will demand political rights to preserve their new freedoms.

And how will humans accept living in a world with another “intelligent species?”

Welcome to your future.

For those of you waiting for the future of sci-fi mystery, I’m well on my way to completing Episode Three of the Laura Kraft, Android Hunter short story series, entitled The Android Who Played with Fire. I think you’ll like it.