Tag Archives: cognition

Of autonomous machines.

 

Last week we talked about how converging technologies can sometimes yield unpredictable results. One of the most influential players in the development of new technology is DARPA and the defense industry. There is a lot of technological convergence going on in the world of defense. Let’s combine robotics, artificial intelligence, machine learning, bio-engineering, ubiquitous surveillance, social media, and predictive algorithms for starters. All of these technologies are advancing at an exponential pace. It’s difficult to take a snapshot of any one of them at a moment in time and predict where they might be tomorrow. When you start blending them the possibilities become downright chaotic. With each step, it is prudent to ask if there is any meaningful review. What are the ramifications for error as well as success? What are the possibilities for misuse? Who is minding the store? We can hope that there are answers to these questions that go beyond platitudes like, “Don’t stand in the way of progress.”, “Time is of the essence.”, or “We’ll cross that bridge when we come to it.”

No comment.

I bring this up after having seen some unclassified documents on Human Systems, and Autonomous Defense Systems (AKA autonomous weapons). (See a previous blog on this topic.) Links to these documents came from a crowd-funded “investigative journalist” Nafeez Ahmed, publishing on a website called INSURGE intelligence.

One of the documents entitled Human Systems Roadmap is a slide presentation given to the National Defense Industry Association (NDIA) conference last year. The list of agencies involved in that conference and the rest of the documents cited reads like an alphabet soup of military and defense organizations which most of us have never heard of. There are multiple components to the pitch, but one that stands out is “Autonomous Weapons Systems that can take action when needed.” Autonomous weapons are those that are capable of making the kill decision without human intervention. There is also, apparently some focused inquiry into “Social Network Research on New Threats… Text Analytics for Context and Event Prediction…” and “full spectrum social media analysis.” We could get all up in arms about this last feature, but recent incidents in places such as, Benghazi, Egypt, and Turkey had a social networking component that enabled extreme behavior to be quickly mobilized. In most cases, the result was a tragic loss of life. In addition to sharing photos of puppies, social media, it seems, is also good at organizing lynch mobs. We shouldn’t be surprised that governments would want to know how to predict such events in advance. The bigger question is how we should intercede and whether that decision should be made by a human being or a machine.

There are lots of other aspects and lots more documents cited in Ahmed’s lengthy albeit activistic report, but the idea here is that rapidly advancing technology is enabling considerations which were previously held to be science fiction or just impossible. Will we reach the point where these systems are fully operational before we reach the point where we know they are totally safe? It’s a problem when technology grows faster that policy, ethics or meaningful review. And it seems to me that it is always a problem when the race to make something work is more important than the understanding the ramifications if it does.

To be clear, I’m not one of those people who thinks that anything and everything that the military can conceive of is automatically wrong. We will never know how many catastrophes that our national defense services have averted by their vigilance and technological prowess. It should go without saying that the bad guys will get more sophisticated in their methods and tactics, and if we are unable to stay ahead of the game, then we will need to get used to the idea of catastrophe. When push comes to shove, I want the government to be there to protect me. That being said, I’m not convinced that the defense infrastructure (or any part of the tech sector for that matter) is as diligent to anticipate the repercussions of their creations as they are to get them functioning. Only individuals can insist on meaningful review.

Thoughts?

 

Bookmark and Share

Enter the flaw.

 

I promised a drone update this week, but by now, it is probably already old news. It is a safe bet there are probably a few thousand more drones than last week. Hence, I’m going to shift to a topic that I think is moving even faster than our clogged airspace.

And now for an AI update. I’ve blogged previously about Kurzweil’s Law of Accelerating Returns, but the evidence is mounting every day that he’s probably right.  The rate at which artificial intelligence is advancing is beginning to match nicely with his curve. A recent article on the Txchnologist website demonstrates how an AI system called Kulitta, is composing jazz, classical, new age and eclectic mixes that are difficult to tell from human compositions. You can listen to an example here. Not bad actually. Sophisticated AI creations like this underscore the realization that we can no longer think of robotics as the clunky mechanized brutes. AI can create. Even though it’s studying an archive of man-made creations the resulting work is unique.

First it learns from a corpus of existing compositions. Then it generates an abstract musical structure. Next it populates this structure with chords. Finally, it massages the structure and notes into a specific musical framework. In just a few seconds, out pops a musical piece that nobody has ever heard before.

The creator of Kulitta, Donya Quick says that this will not put composers out of a job, it will help them do their job better. She doesn’t say how exactly.

If even trained ears can’t always tell the difference, what does that mean for the masses? When we can load the “universal composer” app onto our phone and have a symphony written for ourselves, how will this serve the interests of musicians and authors?

The article continues:

Kulitta joins a growing list of programs that can produce artistic works. Such projects have reached a critical mass–last month Dartmouth College computational scientists announced they would hold a series of contests. They have put a call out seeking artificial intelligence algorithms that produce “human-quality” short stories, sonnets and dance music. These will be pitted against compositions made by humans to see if people can tell the difference.

The larger question to me is this: “When it all sounds wonderful or reads like poetry, will it make any difference to us who created it?”

Sadly, I think not. The sweat and blood that composers and artists pour into their compositions could be a thing of the past. If we see this in the fine arts, then it seems an inevitable consequence for design as well. Once the AI learns the characters, behaviors and personalities of the characters in The Lightstream Chronicles, it can create new episodes without me. Taking characters and setting that already exist as CG constructs, it’s not a stretch that it will be able to generate the wireframes, render the images, and layout the panels.

Would this app help me in my work? It could probably do it in a fraction of the time that it would take me, but could I honestly say it’s mine?

When art and music are all so easily reconstructed and perfect, I wonder if we will miss the flaw. Will we miss that human scratch on the surface of perfection, the thing that reminds us that we are human?

There is probably an algorithm for that, too. Just go to settings > humanness and use the slider.

Bookmark and Share

The Robo-Apocalypse. Part 2.

 

Last week I talked about how the South Koreans have developed a 50 caliber toting, nearly autonomous weapon system and have sold a few dozen around the world. This week I feel obligated to finish up on my promise of the drone with a pistol. I discovered this from a WIRED article. It was a little tongue-in-cheek piece that analyzed a YouTube video and concluded that pistol-packing drone is probably real. I can’t think of anyone who doesn’t believe that this is a really bad idea, including the author of the piece. Nevertheless, if we were to make a list of unintended consequences of DIY drone technology, (just some simple brainstorming) the list, after a few minutes, would be a long one.

This week FastCo reported that  NASA held a little get-together with about 1,000 invited guests from the drone industry to talk about a plan to manage the traffic when, as the agency believes, “every home will have a drone, and every home will serve as an airport at some point in the future”. NASA’s plan takes things slowly. Still the agency predicts that we will be able to get our packages from Amazon and borrow a cup of sugar from Aunt Gladys down the street, even in populated areas, by 2019.

Someone taking action is good news as we work to fix another poorly conceived technology that quickly went rogue. Unfortunately, it does nothing about the guy who wants to shoot down the Amazon drone for sport (or anyone/anything else for that matter).

On the topic of bad ideas, this week The Future Of Life Institute, a research organization out of Boston issued an open letter warning the world that autonomous weapons powered by artificial intelligence (AI) were imminent. The reasonable concern here is that a computer will do the kill-or-not-kill, bomb-or-not-bomb thinking, without the human fail-safe. Here’s an excerpt from the letter:

“Unlike nuclear weapons, they require no costly or hard-to-obtain raw materials, so they will become ubiquitous and cheap for all significant military powers to mass-produce. It will only be a matter of time until they appear on the black market and in the hands of terrorists, dictators wishing to better control their populace, warlords wishing to perpetrate ethnic cleansing, etc. Autonomous weapons are ideal for tasks such as assassinations, destabilizing nations, subduing populations and selectively killing a particular ethnic group. We therefore believe that a military AI arms race would not be beneficial for humanity. There are many ways in which AI can make battlefields safer for humans, especially civilians, without creating new tools for killing people.” [Emphasis mine.]

The letter is short. You should read it. For once we have and example of those smart people I alluded to last week, the ones with compassion and vision. For virtually every “promising” new technology—from the seemingly good to the undeniably dangerous) we need people who can foresee the unintended consequences of one-sided promises. Designers, scientists, and engineers are prime candidates to look into the future and wave these red flags. Then the rest of the world needs to pay attention.

Once again, however, the technology is here and whether it is legal or illegal, banned or not banned the cat is out of the bag. It is kind of like a nuclear explosion. Some things you just can’t take back.

Bookmark and Share

The robo-apocalypse. Part 1.

Talk of robot takeovers is all the rage right now.

I’m good with this because the evidence is out there that robots will continue to get smarter and smarter but the human condition, being what it is, we will continue to do stupid s**t. Here are some examples from the news this week.

1. The BBC reported this week that South Korea has deployed something called The Super aEgis II, a 50-caliber robotic machine gun that knows who is an enemy and who isn’t. At least that’s the plan. The company that built and sells the Super aEgis is DoDAAM. Maybe that is short for do damage. The BBC astutely notes,

“Science fiction writer Isaac Asimov’s First Law of Robotics, that ‘a robot may not injure a human being or, through inaction, allow a human being to come to harm’, looks like it will soon be broken.”

Asimov was more than a great science-fiction writer, he was a Class A futurist. He clearly saw the potential for us to create robots that were smarter and more powerful than we are. He figured there should be some rules. Asimov used the kind of foresight that responsible scientists, technologists and designers should be using for everything we create. As the article continues, Simon Parkin of the BBC quotes Yangchan Song, DoDAAM’s managing director of strategy planning.

“Automated weapons will be the future. We were right. The evolution has been quick. We’ve already moved from remote control combat devices, to what we are approaching now: smart devices that are able to make their own decisions.”

Or in the words of songwriter Donald Fagen,

“A just machine to make big decisions
Programmed by fellows with compassion and vision…1

Relax. The world is full of these fellows. Right now the weapon/robot is linked to a human who gives the OK to fire, and all customers who purchased the 30 units thus far have opted for the human/robot interface. But the company admits,

“If someone came to us wanting a turret that did not have the current safeguards we would, of course, advise them otherwise, and highlight the potential issues,” says Park. “But they will ultimately decide what they want. And we develop to customer specification.”

A 50 caliber round. Heavy damage.
A 50 caliber round. Heavy damage.

They are currently working on the technology that will help their machine make the right decision on its own., but the article cites several academics and researchers who see red flags waving. Most concur that teaching a robot right from wrong is no easy task. Compound the complexity because the fellows who are doing the programming don’t always agree on these issues.

Last week I wrote about Google’s self-driving car. Of course, this robot has to make tough decisions too. It may one day have to decide whether to hit the suddenly appearing baby carriage, the kid on the bike, or just crash the vehicle. In fact, Parkin’s article brings Google into the picture as well, quoting Colin Allen,

“Google admits that one of the hardest problems for their programming is how an automated car should behave at a four-way stop sign…”

Humans don’t do such a good job at that either. And there is my problem with all of this. If the humans who are programming these machines are still wrestling with what is ethically right or wrong, can a robot be expected to do better. Some think so. Over at DoDAMM,

“Ultimately, we would probably like a machine with a very sound basis to be able to learn for itself, and maybe even exceed our abilities to reason morally.”

Based on what?

Next week: Drones with pistols.

 

1. Donald Fagen, IGY From the Night Fly album. 1982
Bookmark and Share

Promises. Promises.

Throughout the course of the week, usually on a daily basis, I collect articles, news blurbs and what I call “signs from the future.” Mostly they fall into categories such as design fiction, technology, society, future, theology, and philosophy. I use this content sometimes for this blog, possibly for a lecture but most often for additional research as part of scholarly papers and presentations that are a matter of course as a professor. I have to weigh what goes into the blog because most of these topics could easily become full-blown papers.  Of course, the thing with scholarly writing is that most publications demand exclusivity on publishing your ideas. Essentially, that means that it becomes difficult to repurpose anything I write here for something with more gravitas.  One of the subjects that are of growing interest to me is Google. Not the search engine, per se, rather the technological mega-corp. It has the potential to be just such a paper, so even though there is a lot to say, I’m going to land on only a few key points.

A ubiquitous giant in the world of the Internet, Google has some of the most powerful algorithms, stores your most personal information, and is working on many of the most advanced technologies in the world. They try very hard to be soft-spoken, and low-key, but it belies their enormous power.

Most of us would agree that technology has provided some marvelous benefits to society especially in the realms of medicine, safety, education and other socially beneficial applications. Things like artificial knees, cochlear implants, air bags (when they don’t accidentally kill you), and instantaneous access to the world’s libraries have made life-changing improvements. Needless to say, especially if you have read my blog for any amount of time, technology also can have a downside. We may see greater yields from our agricultural efforts, but technological advancements also pump needless hormones into the populace, create sketchy GMO foodstuffs and manipulate farmers into planting them. We all know the problems associated with automobile emissions, atomic energy, chemotherapy and texting while driving. These problems are the obvious stuff. What is perhaps more sinister are the technologies we adopt that work quietly in the background to change us. Most of them we are unaware of until, one day, we are almost surprised to see how we have changed, and maybe we don’t like it. Google strikes me as a potential contributor in this latter arena. A recent article from The Guardian, entitled “Where is Google Taking Us?” looks at some of their most altruistic technologies (the ones they allowed the author to see). The author, Tim Adams, brought forward some interesting quotes from key players at Google. When discussing how Google would spend some $62 million in cash that it had amassed, Larry Page, one of the company’s co-founders asked,

“How do we use all these resources… and have a much more positive impact on the world?”

There’s nothing wrong with that question. It’s the kind of question that you would want a billionaire asking. My question is, “What does positive mean, and who decides what is and what isn’t?” In this case, it’s Google. The next quote comes from Sundar Pichai. With so many possibilities that this kind of wealth affords, Adams asked how they stay focused on what to do next.

“’Focus on the user and all else follows…We call it the toothbrush test,’ Pichai says, ‘we want to concentrate our efforts on things that billions of people use on a daily basis.’”

The statement sounds like savvy marketing. He is also talking about the most innate aspects of our everyday behavior. And so that I don’t turn this into an academic paper, here is one more quote. This time the author is talking to Dmitri Dolgov, principal engineer for Google Self-Driving Cars. For the whole idea to work, that is, the car reacting like a human would, only better, it has to think.

“Our maps have information stored and as the car drives around it builds up another local map with its sensors and aligns one to the other – that gives us a location accuracy of a centimetre or two. Beyond that, we are making huge numbers of probabilistic calculations every second.”

Mapping everything down to the centimeter.
Mapping everything down to the centimeter.

It’s the last line that we might want to ponder. Predictive algorithms are what artificial intelligence is all about, the kind of technology that plugs-in to a whole host of different applications from predicting your behavior to your abilities. If we don’t want to have to remember to check the oil, there is a light that reminds us. If we don’t want to have to remember somebody’s name, there is a facial recognition algorithm to remember for us. If my wearable detects that I am stressed, it can remind me to take a deep breath. If I am out for a walk, maybe something should mention all the things I could buy while I’m out (as well as what I am out of). 

Here’s what I think about. It seems to me that we are amassing two lists: the things we don’t want to think about, and the things we do. Folks like Google are adding things to Column A, and it seems to be getting longer all the time. My concern is whether we will have anything left in Column B.

 

Bookmark and Share

Breathing? There’s an app for that.

As the Internet of Things (IoT) and Ubiquitous Computing (UbiComp) continue to advance there really is no more room left for surprise. These things are cascading out of Silicon Valley, crowd-funding sites, labs, and start-ups with continually accelerating speed. And like Kurzweil, I think it’s happening faster than 95 percent of the world is expecting. A fair number of these are duds and frankly superfluous attempts at “computing” what otherwise, with a little mental effort, we could do on our own. Ian Bogost’s article, this week in the Atlantic Monthly,The Internet of Things You Don’t Really Need points out how many of these “innovations” are replacing just the slightest amount of extra brain power, ever-so-minimal physical activity, or prescient concentration. Not to mention that these apps just supply another entry into your personal, digital footprint. More in the week’s news (this stuff is happening everywhere) this time in FastCompany, an MIT alumn who is concerned about how little “face time” her kids are getting with real humans because they are constantly in front of screens or tablets. (Human to human interaction is important for development of emotional intelligence.) The solution? If you think it is less time on the tablet and more “go out and play”, you are behind the times. The researcher, Rana el Kaliouby, has decided that she has the answer:

“Instead, she believes we should be working to make computers more emotionally intelligent. In 2009, she cofounded a company called Affectiva, just outside Boston, where scientists create tools that allow computers to read faces, precisely connecting each brow furrow or smile line to a specific emotion.”

Of course it is. Now, what we don’t know, don’t want to learn (by doing), or just don’t want to think about, our computer, or app, will do for us. The FastCo author Elizabeth Segran, interviewed el Kaliouby:

“The technology is able to deduce emotions that we might not even be able to articulate, because we are not fully aware of them,” El Kaliouby tells me. “When a viewer sees a funny video, for instance, the Affdex might register a split second of confusion or disgust before the viewer smiles or laughs, indicating that there was actually something disturbing to them in the video.”

Oh my.

“At some point in the future, El Kaliouby suggests fridges might be equipped to sense when we are depressed in order to prevent us from binging on chocolate ice cream. Or perhaps computers could recognize when we are having a bad day, and offer a word of empathy—or a heartwarming panda video.”

Please no.

By the way, this is exactly the type of technology that is at the heart of the mesh, the ubiquitous surveillance system in The Lightstream Chronicles. In addition to having learned every possible variation of human emotion, this software has also learned physical behavior such that it can tell when, or if someone is about to shoplift, attack, or threaten another person. It can even tell if you have any business being where you are or not.

So,  before we get swept up in all of the heartwarming possibilities for relating to our computers, (shades of Her), and just in case anyone is left who is alarmed at becoming a complete emotional, intellectual and physical muffin, there is significant new research that suggests that the mind is a muscle. You use it or lose it, that you can strengthen learning and intelligence by exercising and challenging your mind and cognitive skills. If my app is going remind me not to be rude, when to brush my teeth, drink water, stop eating, and go to the toilet, what’s left? The definition of post-human comes to mind.

As a designer, I see warning flags. It is precisely a designer’s ability for abstract reasoning that makes problem solving both gratifying and effective. Remember McGyver? You don’t have to, your life hacks app will tell you what you need to do. You might also want to revisit a previous blog on computers that are taking our jobs away.

macgyver
McGyver. If you don’t know, you’re going to have to look it up.

Yet, it would seem that many people think that the only really important human trait is happiness, that ill-defined, elusive, and completely arbitrary emotion. As long as we retain that, all those other human traits we should evolve out of anyway.

What do you think?

Bookmark and Share