All posts by J. Stern

Mr. West Goes to the Movies

Ken West is taking a break from discussing the news of the day to talk about one of his favorite subjects: modern film. During the day Jan. 22, West phoned in two important movie reviews, both of them in the genre of comedy sci-fi.
The first film impressed West so much that he phoned the newsdesk several times to deliver his review, often repeating parts of the plot and describing key characters more than once.

 

“People get on me for repeating things,” West said. “How many times have you heard Mick Jagger sing ‘Brown Sugar?’ You have something important to say – you say it again.”

 
The film, titled “Paul”, was released in 2011. It features the voice of Seth Rogen as a wisecracking, chain smoking extra-terrestrial, and veteran comedy duo Simon Pegg and Nick Frost as two friends.
“They were working stiffs,” West said, describing Pegg and Frost’s characters. “They went on a road trip. A road trip.”
West spoke at length about how the “Paul” character correlates to his perceptions of actual space aliens, describing “highly advanced humanoids with big heads and tiny little bodies.”
West also theorized that entities like Paul could be the “god-planets” from which sections of humanity are descended.
“They’re a billion years more advanced than you.” West said, describing a scene where Paul picks up a dead bird, heals it, then eats it, as a reminder to humans not to eat “dead meat.”
Noting that in real life, Area 51 is “more likely a test range for high-class military installations,” West proceeded to the second movie review, saying he recently viewed “Men in Black” and was favorably impressed.
“There was this cat that had a galaxy around its neck,” West said.
Another of his favorite parts, he said, was the end.
“The guy that was the major man in black – I don’t remember his name, famous actor – he had to jump into the belly of a giant bug from another part of the universe. He went into the gullet – he antagonized it – he said ‘eat me!’ – he comes out with a bang – blows it in half.”
Signing off, West put in a quick plug for his “snap-on,” an affordable sheet that can keep ice off of a car’s windshield.

Want to Compete and Innovate in Business? Hire a Receptionist

With major recent advances in all sorts of analytical and cognitive technologies, business seems to be moving decisively in the direction of automation. However, this list to starboard, which has been happening for a number of years, leaves businesses in a profound state of disconnect with their audiences. It may be that the only solution to this problem is to return to a more human-centered approach to business communications.

Nearly anyone who contacts a business has a problem. Or, to state it another way, the customers need help. They want to engage and rationalize on a human level — on a social level and in the context of social human relationships. However, increasingly, what they hear and experience in first-tier communication is a bewildering and unwieldy interface — an aggravating series of menu options. Machines that talk much more slowly than a human would in a social scenario. Unclear directories and unclear choices. Some of the worst systems also have poor comprehension, so that they restate problem messages multiple times.

All of this is decisively negative to the customer experience. There is a reason that executives and others have been pounding the drums about customer experience, suggesting that automation will soon innovate at a level beyond what is currently offered. It’s because customer experience is key to business, and a lot of people seem to understand that.

What some don’t understand is that even though artificial intelligence and machine learning are progressing rapidly, these technologies are still more or less in their infancy and have specific limitations related to their capabilities. These specific limitations can be applied in different ways to self-driving vehicles, generative and discriminative engines in machine learning, chat bots and other artificial intelligence entities, and last but not least — interactive voice response systems.

Ask a human receptionist why their IVR is so bad, and you’re not likely to be understood. They may not know the acronym or even the term — or they might play dumb. Part of the irony inherent in these business systems is that the humans at the very end of this automation chain don’t understand how aggravating that automation chain is for the customers. This compounds the problem.

Take the example of mental health services. In corporate mental health services systems, providers will often put their scheduling and appointment setting functionality into an automated IVR system. The problem is that when a customer needs assistance from a mental health provider, he or she is unlikely to be in the frame of mind to navigate one of these aggravating and unwieldy systems. In other words, the automation does not serve the customer.

This is particularly salient in the example of mental health services, because what should be a human-based communications model for a humane service setup has been largely replaced by a corporate and automated model that is inherently incapable of handling the demands placed on it. But it’s not necessary to restrict this problem to the field of mental health services — it can be as broadly applied as the customer who has purchased a sweater with holes in it, the individual whose vehicle has broken down the freeway, or even a business buyer who needs marketing services. In any of these cases, the likelihood is that the customer experience is going to be degraded and poorly served by today’s automated technologies.

Again, this is not a reflection on the rapid progress of the technologies themselves. Deep Blue can beat Kasparov, and Watson can beat human Jeopardy contestants, and different technologies can pass the Turing test with flying colors, but none of this solves the customer’s problem — that he or she needs to be served in the context of the social interaction.

This brings us to a somewhat more technical analysis of the major shortcomings in current machine learning and artificial intelligence models.

Although engineers have learned to simulate the human brain to an amazing extent, deficiencies still exist related to the specific classes of functions that make up human behavior and activity. Specifically, although these technologies can use probabilistic inputs to provide complex results, they are not extremely adept at the sorts of contextual transactions that make up our everyday lives. As a concrete example, a machine learning program may be good at predicting whether or not a human actor will take a step, direct eye movement in a specific way, move a hand or choose a specific button from an array of controls. What the technology is not good at is understanding why someone may make these or other actions.

Another limitation has to do with what some experts might call the “politeness principle” based on a disequilibrium in rational actors’ choices. Going back to the example of a classic Nash equilibrium, we realize that in game theory, most social games have an applicable Nash equilibrium that can be modeled fairly easily. However, some games are structured so that a Nash equilibrium is not practical – or, more specifically, where a Nash equilibrium is only applicable in a fixed set of game scenarios.

 

In a lecture on game theory, Professor Padraic Bartlett explains this in terms of a “social game” of two individuals walking down a hallway toward each other (given a hallway with only two binary path options) –  identifying (left, left) and (right,right) as the two acceptable Nash equilibria, and stating:

“These are the only two equilibria: if we were in either of the mixed states, both players would want to switch, (thus leading to yet another conflict, and the resulting awkwardness).”

Here we see the challenges of applying a Nash equilibrium based on complex social factors. The rational actors have “de facto” choices – and when those choices are made clearly enough, the equilibrium results. Each player knows what is best. But when certain outlier events create uncertainty (maybe one person steps hesitantly, or the other approaching individual misreads a visual cue) the rule fails and the resulting social program is thrown into a infinite loop.

It’s easy to confuse these kinds of “glitches” with scenarios that dispute an equilibrium, such as the “prisoner’s dilemma” where two players must avoid cooperation for the best outcome, but in reality, as we can see, with the politeness principle, a Nash equilibrium does exist and can be implemented. It’s only in the glitchy application of the rule that the equilibrium proves insufficient. (In the established lexicon, this is “trembling hand” equilibrium challenge.)

In other words, we see that if two rational actors choose complementary binary choices (or “uncomplementary” binary choices as it were), they are likely to experience the kinds of recursive decision-making problems that will throw the programs into an infinite loop without exterior human guidance. Unlike two individuals walking toward each other in the hallway, these newly sentient technologies do not have the social ability to make a choice, and to a great extent will not be successful in navigating the problem itself. Here the politeness is a learned skill that is largely unquantifiable and presents machine learning with a significant modeling problem.

Yet another specific limitation relates to the use of highly fitted or possibly overfitted engines that actually approach some of the human qualities that produce indecision. In other words, machines that adopt some of our behaviors may be presented with difficulties related to some of our other behaviors. An article in KD Nuggets posted earlier last year speaks about the use of deep stubborn networks and how they have been engineered with greater complexity. A generative and discriminative engine work at odds with each other to produce collaborative results. This starts to approach some of the higher-level activity in the human mind that is not able to be modeled through linear programming. As the writer describes, what happens is that the competition between the generative and discriminative engines produces some quality that can almost be described as social — a malaise or conflict or, as the author puts it, “anxiety” that is an essential part of the human experience.

Applying words like “anxiety” and “doubt” to machine learning models is inherently a bit of a contradiction. It shows how much progress we have made in constructing machines that can think like us — but it also shows why those machines are not fully or even remotely functional in social roles. They cannot deal with the indecision and anxiety that are produced by their mechanics — and so they cannot serve customers who need this higher-level functionality. This is easy to understand in an elementary sense — we know that although IVR systems can tell people what hours the shop is open, or give people directions to the location, they can’t help customers with a broken toilet or guide them through how to negotiate a better rate on services. However, we don’t know exactly why this is unless we scratch the surface of these cognitive models and start learning about what machine learning can and can’t do.

Faced with an ultimate choice, many companies will stubbornly continue to focus on the possibilities of automation. They will rely on the prestige of new technologies and their abilities to dazzle the general public. They will throw their eggs into the basket of trying to increase the spectrum of what IVR can do. (Many of them led by profit-seeking vendors). Other companies, debatably smarter companies, will simply employ humans to direct business communications in ways that will actually really enhance the customer experience.

The Virtual Chair

The last decade has seen a technology industry in overdrive, a furious wave of innovations coming one after the other, with more promised on the horizon. The blending of cloud computing, IoT, network virtualization and mobile device functionality have thrown the outline of cutting-edge technology into a nebulous space. However, there’s an obvious consensus that machine learning, as a general subset of AI, is the most fundamental new frontier on which the next generation of enterprise and consumer technologies will be built.

All of the methods and tools that made up the machine learning industry, algorithms and training sets, simulated annealing and equilibrium and vector matrices, seem dull and overly mathematical, too esoteric to really reveal what ML has to offer. So it’s instructive to take a different look at what may be coming our way sooner rather than later, with a comprehensive shift in the ways that we view technology as a whole, and how we will embrace our digital peers as they start to develop.

To that end, consider the virtual chair.

In looking at the emergence of new software capabilities over the personal computer era, and even further back to the days of punch cards, it can be helpful to focus on a particular object and its treatment by the spectrum of new principles that we have created. There are reasons why “object-oriented” design became so popular in the advent of a new group of programming languages, and stayed in our lexicon afterward. One primary reason is because an object is an excellent way to comprehend the digital world that we interact with and, increasingly live in, and since this kind of comprehension is becoming necessary, the “object” might help us to make new technologies more egalitarian, to better serve a wider range of users.

The chair is, in some ways, an arbitrary example. It’s one of many such objects that might be printed on flash cards, installed in virtual spaces, or, today, included in training sets. It is, overall, one of a practically infinite number of “classes” that are created ‘ex nihilo’ from the digital world.

In the beginning of the information age, the chair was only a sketch, perhaps a label applied to a linear program written in numbered lines of code. Certainly, mass production facilities began to label chairs as units of production digitally, and might even have stored some rudimentary data on the properties of office furniture.

Since information was limited to what could be produced on the early curve of Moore’s law, the early chair was likely just a collection of text characters or bits intended to be drawn on a monochrome screen. The chair would only become “virtual” manifestly if some programmer had the time and the determination to hand-code its dimensions and other data into a mainframe or, later, a workstation, as in Ellen Ullman’s legendary novel “The Bug,” where an embattled coder puts together virtual “organisms” endowed with certain properties and allows them to “grow” and evolve in a world of code. This example was really ahead of its time, although in retrospect, it didn’t take too many years to move from a BASIC world to the age of “big data.”

Ten years after the millennial change, that’s where we found ourselves: enamored with “big data” and awed at the terabytes that could be rendered to create real, vibrant, virtual objects with real heft, things you could “hold in a (digital) hand” and examine for real insight. In reality, the change happened slowly, by tiny increments, as Moore’s law progressed, and programming methods followed. By gradations, as big data fleshed out what could be held in the average container, the virtual chair became a real work of art, with exact dimensions, color, texture and other properties defined and manipulated in the intricate logic gate halls of fast processors.

But although big data offered the complexity to “make digital things real,” it was also still purely deterministic. Through most of its tenure, big data has been applied through logical I/O, and the castles that it builds are built strictly at the whim of the engineer who writes the code.

Now, with machine learning, there is a fundamental break in this principle: for the first time, technologies have the ability to work according to a mix of probabilistic and deterministic inputs. Computers can produce unpredictable outcomes! The ability for computers to “learn” is the ability to take in data and filter it through probabilistic layers to model it and produce something that was not planned out by human makers. In other words, going back to the virtual chair, while big data programming allowed us to define a piece of furniture to complex specifications in a virtual space, machine learning essentially builds the chair for us, and knows before we do what the finished product is going to look like.

But before there’s too much fanfare over this benchmark of achievement, it makes sense to ask what rules will be applied to the mix of D and P inputs that we will be using to “build virtual chairs.”

Think of a poker player, such as John Malkovitch’s character, “Teddy KGB” in the very human film “Rounders,” sitting at the table, examining another for ‘tells.’ Linear programming, big data analytics, tells us what happens if the other player makes eye contact: “IF (eye contact) THEN (x)” and, in its more sophisticated forms, tells us how many times eye contact has been made in the past, forecasting outcomes. Machine learning purports to tell us whether there will be eye contact, according to training data, and what that means. But as a model, how the algorithms interpret the training data has to depend on how we treat the weighted inputs: for example, the difference between guessing at human intentions, and guessing at physical outcomes that seem random. Will there be eye contact? Will a player move a hand? Machine learning systems progress beyond tabulating results, and move toward complex modeling that, again, depends on its parameters, although there is a real and growing element of self-determinism and automation applied. We have to know the rules, we have to know how to apply them, and we have to know what they mean.

Machine learning will build us the virtual chair, but what else will it build?

What will our chair look like when it is delivered to us, and what will its design depend on?

One of the best clues is the common use of image processing algorithms to translate visual inputs into logic. ML programs “look” at something and identify it – that’s one of the bellwethers of their nascent intelligence. And it’s a big insight into how the learning will work. If programs can be made to process images according to logic, there are many inherent rules built into that process, and the contours for logic become a little more knowable.

To use the poker player analogy, the outcomes will be goal-oriented. Maybe an ML program will take in images of the opposing players, parse them for meaning, and deliver results that reach a more solid “Turing point” of AI-completeness, where we see the program as a living, breathing player (especially if paired with realistic-looking human-styled robotics).

In the end, the bulk of what we will enjoy based on ML engines will be simply a reflection of ourselves, our tastes and behaviors and tendencies, filtered and modeled and fed back to us, chatbots that use our responses to build their own, parroting our impulses. But the significance of moving beyond pure determinism in the digital world shouldn’t be lost on us – as technology obtains the power to build, that’s one more giant capability that humans surrender as their own exclusively, moving us closer to a time when digital entities become, if not our equals, a much more confounding facsimile.