Tag Archives: environment

#434643 Sensors and Machine Learning Are Giving ...

According to some scientists, humans really do have a sixth sense. There’s nothing supernatural about it: the sense of proprioception tells you about the relative positions of your limbs and the rest of your body. Close your eyes, block out all sound, and you can still use this internal “map” of your external body to locate your muscles and body parts – you have an innate sense of the distances between them, and the perception of how they’re moving, above and beyond your sense of touch.

This sense is invaluable for allowing us to coordinate our movements. In humans, the brain integrates senses including touch, heat, and the tension in muscle spindles to allow us to build up this map.

Replicating this complex sense has posed a great challenge for roboticists. We can imagine simulating the sense of sight with cameras, sound with microphones, or touch with pressure-pads. Robots with chemical sensors could be far more accurate than us in smell and taste, but building in proprioception, the robot’s sense of itself and its body, is far more difficult, and is a large part of why humanoid robots are so tricky to get right.

Simultaneous localization and mapping (SLAM) software allows robots to use their own senses to build up a picture of their surroundings and environment, but they’d need a keen sense of the position of their own bodies to interact with it. If something unexpected happens, or in dark environments where primary senses are not available, robots can struggle to keep track of their own position and orientation. For human-robot interaction, wearable robotics, and delicate applications like surgery, tiny differences can be extremely important.

Piecemeal Solutions
In the case of hard robotics, this is generally solved by using a series of strain and pressure sensors in each joint, which allow the robot to determine how its limbs are positioned. That works fine for rigid robots with a limited number of joints, but for softer, more flexible robots, this information is limited. Roboticists are faced with a dilemma: a vast, complex array of sensors for every degree of freedom in the robot’s movement, or limited skill in proprioception?

New techniques, often involving new arrays of sensory material and machine-learning algorithms to fill in the gaps, are starting to tackle this problem. Take the work of Thomas George Thuruthel and colleagues in Pisa and San Diego, who draw inspiration from the proprioception of humans. In a new paper in Science Robotics, they describe the use of soft sensors distributed through a robotic finger at random. This placement is much like the constant adaptation of sensors in humans and animals, rather than relying on feedback from a limited number of positions.

The sensors allow the soft robot to react to touch and pressure in many different locations, forming a map of itself as it contorts into complicated positions. The machine-learning algorithm serves to interpret the signals from the randomly-distributed sensors: as the finger moves around, it’s observed by a motion capture system. After training the robot’s neural network, it can associate the feedback from the sensors with the position of the finger detected in the motion-capture system, which can then be discarded. The robot observes its own motions to understand the shapes that its soft body can take, and translate them into the language of these soft sensors.

“The advantages of our approach are the ability to predict complex motions and forces that the soft robot experiences (which is difficult with traditional methods) and the fact that it can be applied to multiple types of actuators and sensors,” said Michael Tolley of the University of California San Diego. “Our method also includes redundant sensors, which improves the overall robustness of our predictions.”

The use of machine learning lets the roboticists come up with a reliable model for this complex, non-linear system of motions for the actuators, something difficult to do by directly calculating the expected motion of the soft-bot. It also resembles the human system of proprioception, built on redundant sensors that change and shift in position as we age.

In Search of a Perfect Arm
Another approach to training robots in using their bodies comes from Robert Kwiatkowski and Hod Lipson of Columbia University in New York. In their paper “Task-agnostic self-modeling machines,” also recently published in Science Robotics, they describe a new type of robotic arm.

Robotic arms and hands are getting increasingly dexterous, but training them to grasp a large array of objects and perform many different tasks can be an arduous process. It’s also an extremely valuable skill to get right: Amazon is highly interested in the perfect robot arm. Google hooked together an array of over a dozen robot arms so that they could share information about grasping new objects, in part to cut down on training time.

Individually training a robot arm to perform every individual task takes time and reduces the adaptability of your robot: either you need an ML algorithm with a huge dataset of experiences, or, even worse, you need to hard-code thousands of different motions. Kwiatkowski and Lipson attempt to overcome this by developing a robotic system that has a “strong sense of self”: a model of its own size, shape, and motions.

They do this using deep machine learning. The robot begins with no prior knowledge of its own shape or the underlying physics of its motion. It then repeats a series of a thousand random trajectories, recording the motion of its arm. Kwiatkowski and Lipson compare this to a baby in the first year of life observing the motions of its own hands and limbs, fascinated by picking up and manipulating objects.

Again, once the robot has trained itself to interpret these signals and build up a robust model of its own body, it’s ready for the next stage. Using that deep-learning algorithm, the researchers then ask the robot to design strategies to accomplish simple pick-up and place and handwriting tasks. Rather than laboriously and narrowly training itself for each individual task, limiting its abilities to a very narrow set of circumstances, the robot can now strategize how to use its arm for a much wider range of situations, with no additional task-specific training.

Damage Control
In a further experiment, the researchers replaced part of the arm with a “deformed” component, intended to simulate what might happen if the robot was damaged. The robot can then detect that something’s up and “reconfigure” itself, reconstructing its self-model by going through the training exercises once again; it was then able to perform the same tasks with only a small reduction in accuracy.

Machine learning techniques are opening up the field of robotics in ways we’ve never seen before. Combining them with our understanding of how humans and other animals are able to sense and interact with the world around us is bringing robotics closer and closer to becoming truly flexible and adaptable, and, eventually, omnipresent.

But before they can get out and shape the world, as these studies show, they will need to understand themselves.

Image Credit: jumbojan / Shutterstock.com Continue reading

Posted in Human Robots

#434534 To Extend Our Longevity, First We Must ...

Healthcare today is reactive, retrospective, bureaucratic, and expensive. It’s sick care, not healthcare.

But that is radically changing at an exponential rate.

Through this multi-part blog series on longevity, I’ll take a deep dive into aging, longevity, and healthcare technologies that are working together to dramatically extend the human lifespan, disrupting the $3 trillion healthcare system in the process.

I’ll begin the series by explaining the nine hallmarks of aging, as explained in this journal article. Next, I’ll break down the emerging technologies and initiatives working to combat these nine hallmarks. Finally, I’ll explore the transformative implications of dramatically extending the human health span.

In this blog I’ll cover:

Why the healthcare system is broken
Why, despite this, we live in the healthiest time in human history
The nine mechanisms of aging

Let’s dive in.

The System is Broken—Here’s the Data:

Doctors spend $210 billion per year on procedures that aren’t based on patient need, but fear of liability.
Americans spend, on average, $8,915 per person on healthcare—more than any other country on Earth.
Prescription drugs cost around 50 percent more in the US than in other industrialized countries.
At current rates, by 2025, nearly 25 percent of the US GDP will be spent on healthcare.
It takes 12 years and $359 million, on average, to take a new drug from the lab to a patient.
Only 5 in 5,000 of these new drugs proceed to human testing. From there, only 1 of those 5 is actually approved for human use.

And Yet, We Live in the Healthiest Time in Human History
Consider these insights, which I adapted from Max Roser’s excellent database Our World in Data:

Right now, the countries with the lowest life expectancy in the world still have higher life expectancies than the countries with the highest life expectancy did in 1800.
In 1841, a 5-year-old had a life expectancy of 55 years. Today, a 5-year-old can expect to live 82 years—an increase of 27 years.
We’re seeing a dramatic increase in healthspan. In 1845, a newborn would expect to live to 40 years old. For a 70-year-old, that number became 79. Now, people of all ages can expect to live to be 81 to 86 years old.
100 years ago, 1 of 3 children would die before the age of 5. As of 2015, the child mortality rate fell to just 4.3 percent.
The cancer mortality rate has declined 27 percent over the past 25 years.

Figure: Around the globe, life expectancy has doubled since the 1800s. | Image from Life Expectancy by Max Roser – Our World in Data / CC BY SA
Figure: A dramatic reduction in child mortality in 1800 vs. in 2015. | Image from Child Mortality by Max Roser – Our World in Data / CC BY SA
The 9 Mechanisms of Aging
*This section was adapted from CB INSIGHTS: The Future Of Aging.

Longevity, healthcare, and aging are intimately linked.

With better healthcare, we can better treat some of the leading causes of death, impacting how long we live.

By investigating how to treat diseases, we’ll inevitably better understand what causes these diseases in the first place, which directly correlates to why we age.

Following are the nine hallmarks of aging. I’ll share examples of health and longevity technologies addressing each of these later in this blog series.

Genomic instability: As we age, the environment and normal cellular processes cause damage to our genes. Activities like flying at high altitude, for example, expose us to increased radiation or free radicals. This damage compounds over the course of life and is known to accelerate aging.
Telomere attrition: Each strand of DNA in the body (known as chromosomes) is capped by telomeres. These short snippets of DNA repeated thousands of times are designed to protect the bulk of the chromosome. Telomeres shorten as our DNA replicates; if a telomere reaches a certain critical shortness, a cell will stop dividing, resulting in increased incidence of disease.
Epigenetic alterations: Over time, environmental factors will change how genes are expressed, i.e., how certain sequences of DNA are read and the instruction set implemented.
Loss of proteostasis: Over time, different proteins in our body will no longer fold and function as they are supposed to, resulting in diseases ranging from cancer to neurological disorders.
Deregulated nutrient-sensing: Nutrient levels in the body can influence various metabolic pathways. Among the affected parts of these pathways are proteins like IGF-1, mTOR, sirtuins, and AMPK. Changing levels of these proteins’ pathways has implications on longevity.
Mitochondrial dysfunction: Mitochondria (our cellular power plants) begin to decline in performance as we age. Decreased performance results in excess fatigue and other symptoms of chronic illnesses associated with aging.
Cellular senescence: As cells age, they stop dividing and cannot be removed from the body. They build up and typically cause increased inflammation.
Stem cell exhaustion: As we age, our supply of stem cells begins to diminish as much as 100 to 10,000-fold in different tissues and organs. In addition, stem cells undergo genetic mutations, which reduce their quality and effectiveness at renovating and repairing the body.
Altered intercellular communication: The communication mechanisms that cells use are disrupted as cells age, resulting in decreased ability to transmit information between cells.

Conclusion
Over the past 200 years, we have seen an abundance of healthcare technologies enable a massive lifespan boom.

Now, exponential technologies like artificial intelligence, 3D printing and sensors, as well as tremendous advancements in genomics, stem cell research, chemistry, and many other fields, are beginning to tackle the fundamental issues of why we age.

In the next blog in this series, we will dive into how genome sequencing and editing, along with new classes of drugs, are augmenting our biology to further extend our healthy lives.

What will you be able to achieve with an extra 30 to 50 healthy years (or longer) in your lifespan? Personally, I’m excited for a near-infinite lifespan to take on moonshots.

Join Me
Abundance-Digital Online Community: I’ve created a Digital/Online community of bold, abundance-minded entrepreneurs called Abundance-Digital. Abundance-Digital is my ‘onramp’ for exponential entrepreneurs – those who want to get involved and play at a higher level. Click here to learn more.

Image Credit: David Carbo / Shutterstock.com Continue reading

Posted in Human Robots

#434336 These Smart Seafaring Robots Have a ...

Drones. Self-driving cars. Flying robo taxis. If the headlines of the last few years are to be believed, terrestrial transportation in the future will someday be filled with robotic conveyances and contraptions that will require little input from a human other than to download an app.

But what about the other 70 percent of the planet’s surface—the part that’s made up of water?

Sure, there are underwater drones that can capture 4K video for the next BBC documentary. Remotely operated vehicles (ROVs) are capable of diving down thousands of meters to investigate ocean vents or repair industrial infrastructure.

Yet most of the robots on or below the water today still lean heavily on the human element to operate. That’s not surprising given the unstructured environment of the seas and the poor communication capabilities for anything moving below the waves. Autonomous underwater vehicles (AUVs) are probably the closest thing today to smart cars in the ocean, but they generally follow pre-programmed instructions.

A new generation of seafaring robots—leveraging artificial intelligence, machine vision, and advanced sensors, among other technologies—are beginning to plunge into the ocean depths. Here are some of the latest and most exciting ones.

The Transformer of the Sea
Nic Radford, chief technology officer of Houston Mechatronics Inc. (HMI), is hesitant about throwing around the word “autonomy” when talking about his startup’s star creation, Aquanaut. He prefers the term “shared control.”

Whatever you want to call it, Aquanaut seems like something out of the script of a Transformers movie. The underwater robot begins each mission in a submarine-like shape, capable of autonomously traveling up to 200 kilometers on battery power, depending on the assignment.

When Aquanaut reaches its destination—oil and gas is the primary industry HMI hopes to disrupt to start—its four specially-designed and built linear actuators go to work. Aquanaut then unfolds into a robot with a head, upper torso, and two manipulator arms, all while maintaining proper buoyancy to get its job done.

The lightbulb moment of how to engineer this transformation from submarine to robot came one day while Aquanaut’s engineers were watching the office’s stand-up desks bob up and down. The answer to the engineering challenge of the hull suddenly seemed obvious.

“We’re just gonna build a big, gigantic, underwater stand-up desk,” Radford told Singularity Hub.

Hardware wasn’t the only problem the team, comprised of veteran NASA roboticists like Radford, had to solve. In order to ditch the expensive support vessels and large teams of humans required to operate traditional ROVs, Aquanaut would have to be able to sense its environment in great detail and relay that information back to headquarters using an underwater acoustics communications system that harkens back to the days of dial-up internet connections.

To tackle that problem of low bandwidth, HMI equipped Aquanaut with a machine vision system comprised of acoustic, optical, and laser-based sensors. All of that dense data is compressed using in-house designed technology and transmitted to a single human operator who controls Aquanaut with a few clicks of a mouse. In other words, no joystick required.

“I don’t know of anyone trying to do this level of autonomy as it relates to interacting with the environment,” Radford said.

HMI got $20 million earlier this year in Series B funding co-led by Transocean, one of the world’s largest offshore drilling contractors. That should be enough money to finish the Aquanaut prototype, which Radford said is about 99.8 percent complete. Some “high-profile” demonstrations are planned for early next year, with commercial deployments as early as 2020.

“What just gives us an incredible advantage here is that we have been born and bred on doing robotic systems for remote locations,” Radford noted. “This is my life, and I’ve bet the farm on it, and it takes this kind of fortitude and passion to see these things through, because these are not easy problems to solve.”

On Cruise Control
Meanwhile, a Boston-based startup is trying to solve the problem of making ships at sea autonomous. Sea Machines is backed by about $12.5 million in capital venture funding, with Toyota AI joining the list of investors in a $10 million Series A earlier this month.

Sea Machines is looking to the self-driving industry for inspiration, developing what it calls “vessel intelligence” systems that can be retrofitted on existing commercial vessels or installed on newly-built working ships.

For instance, the startup announced a deal earlier this year with Maersk, the world’s largest container shipping company, to deploy a system of artificial intelligence, computer vision, and LiDAR on the Danish company’s new ice-class container ship. The technology works similar to advanced driver-assistance systems found in automobiles to avoid hazards. The proof of concept will lay the foundation for a future autonomous collision avoidance system.

It’s not just startups making a splash in autonomous shipping. Radford noted that Rolls Royce—yes, that Rolls Royce—is leading the way in the development of autonomous ships. Its Intelligence Awareness system pulls in nearly every type of hyped technology on the market today: neural networks, augmented reality, virtual reality, and LiDAR.

In augmented reality mode, for example, a live feed video from the ship’s sensors can detect both static and moving objects, overlaying the scene with details about the types of vessels in the area, as well as their distance, heading, and other pertinent data.

While safety is a primary motivation for vessel automation—more than 1,100 ships have been lost over the past decade—these new technologies could make ships more efficient and less expensive to operate, according to a story in Wired about the Rolls Royce Intelligence Awareness system.

Sea Hunt Meets Science
As Singularity Hub noted in a previous article, ocean robots can also play a critical role in saving the seas from environmental threats. One poster child that has emerged—or, invaded—is the spindly lionfish.

A venomous critter endemic to the Indo-Pacific region, the lionfish is now found up and down the east coast of North America and beyond. And it is voracious, eating up to 30 times its own stomach volume and reducing juvenile reef fish populations by nearly 90 percent in as little as five weeks, according to the Ocean Support Foundation.

That has made the colorful but deadly fish Public Enemy No. 1 for many marine conservationists. Both researchers and startups are developing autonomous robots to hunt down the invasive predator.

At the Worcester Polytechnic Institute, for example, students are building a spear-carrying robot that uses machine learning and computer vision to distinguish lionfish from other aquatic species. The students trained the algorithms on thousands of different images of lionfish. The result: a lionfish-killing machine that boasts an accuracy of greater than 95 percent.

Meanwhile, a small startup called the American Marine Research Corporation out of Pensacola, Florida is applying similar technology to seek and destroy lionfish. Rather than spearfishing, the AMRC drone would stun and capture the lionfish, turning a profit by selling the creatures to local seafood restaurants.

Lionfish: It’s what’s for dinner.

Water Bots
A new wave of smart, independent robots are diving, swimming, and cruising across the ocean and its deepest depths. These autonomous systems aren’t necessarily designed to replace humans, but to venture where we can’t go or to improve safety at sea. And, perhaps, these latest innovations may inspire the robots that will someday plumb the depths of watery planets far from Earth.

Image Credit: Houston Mechatronics, Inc. Continue reading

Posted in Human Robots

#434246 How AR and VR Will Shape the Future of ...

How we work and play is about to transform.

After a prolonged technology “winter”—or what I like to call the ‘deceptive growth’ phase of any exponential technology—the hardware and software that power virtual (VR) and augmented reality (AR) applications are accelerating at an extraordinary rate.

Unprecedented new applications in almost every industry are exploding onto the scene.

Both VR and AR, combined with artificial intelligence, will significantly disrupt the “middleman” and make our lives “auto-magical.” The implications will touch every aspect of our lives, from education and real estate to healthcare and manufacturing.

The Future of Work
How and where we work is already changing, thanks to exponential technologies like artificial intelligence and robotics.

But virtual and augmented reality are taking the future workplace to an entirely new level.

Virtual Reality Case Study: eXp Realty

I recently interviewed Glenn Sanford, who founded eXp Realty in 2008 (imagine: a real estate company on the heels of the housing market collapse) and is the CEO of eXp World Holdings.

Ten years later, eXp Realty has an army of 14,000 agents across all 50 US states, three Canadian provinces, and 400 MLS market areas… all without a single traditional staffed office.

In a bid to transition from 2D interfaces to immersive, 3D work experiences, virtual platform VirBELA built out the company’s office space in VR, unlocking indefinite scaling potential and an extraordinary new precedent.

Real estate agents, managers, and even clients gather in a unique virtual campus, replete with a sports field, library, and lobby. It’s all accessible via head-mounted displays, but most agents join with a computer browser. Surprisingly, the campus-style setup enables the same type of water-cooler conversations I see every day at the XPRIZE headquarters.

With this centralized VR campus, eXp Realty has essentially thrown out overhead costs and entered a lucrative market without the same constraints of brick-and-mortar businesses.

Delocalize with VR, and you can now hire anyone with internet access (right next door or on the other side of the planet), redesign your corporate office every month, throw in an ocean-view office or impromptu conference room for client meetings, and forget about guzzled-up hours in traffic.

As a leader, what happens when you can scalably expand and connect your workforce, not to mention your customer base, without the excess overhead of office space and furniture? Your organization can run faster and farther than your competition.

But beyond the indefinite scalability achieved through digitizing your workplace, VR’s implications extend to the lives of your employees and even the future of urban planning:

Home Prices: As virtual headquarters and office branches take hold of the 21st-century workplace, those who work on campuses like eXp Realty’s won’t need to commute to work. As a result, VR has the potential to dramatically influence real estate prices—after all, if you don’t need to drive to an office, your home search isn’t limited to a specific set of neighborhoods anymore.

Transportation: In major cities like Los Angeles and San Francisco, the implications are tremendous. Analysts have revealed that it’s already cheaper to use ride-sharing services like Uber and Lyft than to own a car in many major cities. And once autonomous “Car-as-a-Service” platforms proliferate, associated transportation costs like parking fees, fuel, and auto repairs will no longer fall on the individual, if not entirely disappear.

Augmented Reality: Annotate and Interact with Your Workplace

As I discussed in a recent Spatial Web blog, not only will Web 3.0 and VR advancements allow us to build out virtual worlds, but we’ll soon be able to digitally map our real-world physical offices or entire commercial high-rises.

Enter a professional world electrified by augmented reality.

Our workplaces are practically littered with information. File cabinets abound with archival data and relevant documents, and company databases continue to grow at a breakneck pace. And, as all of us are increasingly aware, cybersecurity and robust data permission systems remain a major concern for CEOs and national security officials alike.

What if we could link that information to specific locations, people, time frames, and even moving objects?

As data gets added and linked to any given employee’s office, conference room, or security system, we might then access online-merge-offline environments and information through augmented reality.

Imagine showing up at your building’s concierge and your AR glasses automatically check you into the building, authenticating your identity and pulling up any reminders you’ve linked to that specific location.

You stop by a friend’s office, and his smart security system lets you know he’ll arrive in an hour. Need to book a public conference room that’s already been scheduled by another firm’s marketing team? Offer to pay them a fee and, once accepted, a smart transaction will automatically deliver a payment to their company account.

With blockchain-verified digital identities, spatially logged data, and virtually manifest information, business logistics take a fraction of the time, operations grow seamless, and corporate data will be safer than ever.

Or better yet, imagine precise and high-dexterity work environments populated with interactive annotations that guide an artisan, surgeon, or engineer through meticulous handiwork.

Take, for instance, AR service 3D4Medical, which annotates virtual anatomy in midair. And as augmented reality hardware continues to advance, we might envision a future wherein surgeons perform operations on annotated organs and magnified incision sites, or one in which quantum computer engineers can magnify and annotate mechanical parts, speeding up reaction times and vastly improving precision.

The Future of Free Time and Play
In Abundance, I wrote about today’s rapidly demonetizing cost of living. In 2011, almost 75 percent of the average American’s income was spent on housing, transportation, food, personal insurance, health, and entertainment. What the headlines don’t mention: this is a dramatic improvement over the last 50 years. We’re spending less on basic necessities and working fewer hours than previous generations.

Chart depicts the average weekly work hours for full-time production employees in non-agricultural activities. Source: Diamandis.com data
Technology continues to change this, continues to take care of us and do our work for us. One phrase that describes this is “technological socialism,” where it’s technology, not the government, that takes care of us.

Extrapolating from the data, I believe we are heading towards a post-scarcity economy. Perhaps we won’t need to work at all, because we’ll own and operate our own fleet of robots or AI systems that do our work for us.

As living expenses demonetize and workplace automation increases, what will we do with this abundance of time? How will our children and grandchildren connect and find their purpose if they don’t have to work for a living?

As I write this on a Saturday afternoon and watch my two seven-year-old boys immersed in Minecraft, building and exploring worlds of their own creation, I can’t help but imagine that this future is about to enter its disruptive phase.

Exponential technologies are enabling a new wave of highly immersive games, virtual worlds, and online communities. We’ve likely all heard of the Oasis from Ready Player One. But far beyond what we know today as ‘gaming,’ VR is fast becoming a home to immersive storytelling, interactive films, and virtual world creation.

Within the virtual world space, let’s take one of today’s greatest precursors, the aforementioned game Minecraft.

For reference, Minecraft is over eight times the size of planet Earth. And in their free time, my kids would rather build in Minecraft than almost any other activity. I think of it as their primary passion: to create worlds, explore worlds, and be challenged in worlds.

And in the near future, we’re all going to become creators of or participants in virtual worlds, each populated with assets and storylines interoperable with other virtual environments.

But while the technological methods are new, this concept has been alive and well for generations. Whether you got lost in the world of Heidi or Harry Potter, grew up reading comic books or watching television, we’ve all been playing in imaginary worlds, with characters and story arcs populating our minds. That’s the nature of childhood.

In the past, however, your ability to edit was limited, especially if a given story came in some form of 2D media. I couldn’t edit where Tom Sawyer was going or change what Iron Man was doing. But as a slew of new software advancements underlying VR and AR allow us to interact with characters and gain (albeit limited) agency (for now), both new and legacy stories will become subjects of our creation and playgrounds for virtual interaction.

Take VR/AR storytelling startup Fable Studio’s Wolves in the Walls film. Debuting at the 2018 Sundance Film Festival, Fable’s immersive story is adapted from Neil Gaiman’s book and tracks the protagonist, Lucy, whose programming allows her to respond differently based on what her viewers do.

And while Lucy can merely hand virtual cameras to her viewers among other limited tasks, Fable Studio’s founder Edward Saatchi sees this project as just the beginning.

Imagine a virtual character—either in augmented or virtual reality—geared with AI capabilities, that now can not only participate in a fictional storyline but interact and dialogue directly with you in a host of virtual and digitally overlayed environments.

Or imagine engaging with a less-structured environment, like the Star Wars cantina, populated with strangers and friends to provide an entirely novel social media experience.

Already, we’ve seen characters like that of Pokémon brought into the real world with Pokémon Go, populating cities and real spaces with holograms and tasks. And just as augmented reality has the power to turn our physical environments into digital gaming platforms, advanced AR could bring on a new era of in-home entertainment.

Imagine transforming your home into a narrative environment for your kids or overlaying your office interior design with Picasso paintings and gothic architecture. As computer vision rapidly grows capable of identifying objects and mapping virtual overlays atop them, we might also one day be able to project home theaters or live sports within our homes, broadcasting full holograms that allow us to zoom into the action and place ourselves within it.

Increasingly honed and commercialized, augmented and virtual reality are on the cusp of revolutionizing the way we play, tell stories, create worlds, and interact with both fictional characters and each other.

Join Me
Abundance-Digital Online Community: I’ve created a Digital/Online community of bold, abundance-minded entrepreneurs called Abundance-Digital. Abundance-Digital is my ‘onramp’ for exponential entrepreneurs – those who want to get involved and play at a higher level. Click here to learn more.

Image Credit: nmedia / Shutterstock.com Continue reading

Posted in Human Robots

#434151 Life-or-Death Algorithms: The Black Box ...

When it comes to applications for machine learning, few can be more widely hyped than medicine. This is hardly surprising: it’s a huge industry that generates a phenomenal amount of data and revenue, where technological advances can improve or save the lives of millions of people. Hardly a week passes without a study that suggests algorithms will soon be better than experts at detecting pneumonia, or Alzheimer’s—diseases in complex organs ranging from the eye to the heart.

The problems of overcrowded hospitals and overworked medical staff plague public healthcare systems like Britain’s NHS and lead to rising costs for private healthcare systems. Here, again, algorithms offer a tantalizing solution. How many of those doctor’s visits really need to happen? How many could be replaced by an interaction with an intelligent chatbot—especially if it can be combined with portable diagnostic tests, utilizing the latest in biotechnology? That way, unnecessary visits could be reduced, and patients could be diagnosed and referred to specialists more quickly without waiting for an initial consultation.

As ever with artificial intelligence algorithms, the aim is not to replace doctors, but to give them tools to reduce the mundane or repetitive parts of the job. With an AI that can examine thousands of scans in a minute, the “dull drudgery” is left to machines, and the doctors are freed to concentrate on the parts of the job that require more complex, subtle, experience-based judgement of the best treatments and the needs of the patient.

High Stakes
But, as ever with AI algorithms, there are risks involved with relying on them—even for tasks that are considered mundane. The problems of black-box algorithms that make inexplicable decisions are bad enough when you’re trying to understand why that automated hiring chatbot was unimpressed by your job interview performance. In a healthcare context, where the decisions made could mean life or death, the consequences of algorithmic failure could be grave.

A new paper in Science Translational Medicine, by Nicholson Price, explores some of the promises and pitfalls of using these algorithms in the data-rich medical environment.

Neural networks excel at churning through vast quantities of training data and making connections, absorbing the underlying patterns or logic for the system in hidden layers of linear algebra; whether it’s detecting skin cancer from photographs or learning to write in pseudo-Shakespearean script. They are terrible, however, at explaining the underlying logic behind the relationships that they’ve found: there is often little more than a string of numbers, the statistical “weights” between the layers. They struggle to distinguish between correlation and causation.

This raises interesting dilemmas for healthcare providers. The dream of big data in medicine is to feed a neural network on “huge troves of health data, finding complex, implicit relationships and making individualized assessments for patients.” What if, inevitably, such an algorithm proves to be unreasonably effective at diagnosing a medical condition or prescribing a treatment, but you have no scientific understanding of how this link actually works?

Too Many Threads to Unravel?
The statistical models that underlie such neural networks often assume that variables are independent of each other, but in a complex, interacting system like the human body, this is not always the case.

In some ways, this is a familiar concept in medical science—there are many phenomena and links which have been observed for decades but are still poorly understood on a biological level. Paracetamol is one of the most commonly-prescribed painkillers, but there’s still robust debate about how it actually works. Medical practitioners may be keen to deploy whatever tool is most effective, regardless of whether it’s based on a deeper scientific understanding. Fans of the Copenhagen interpretation of quantum mechanics might spin this as “Shut up and medicate!”

But as in that field, there’s a debate to be had about whether this approach risks losing sight of a deeper understanding that will ultimately prove more fruitful—for example, for drug discovery.

Away from the philosophical weeds, there are more practical problems: if you don’t understand how a black-box medical algorithm is operating, how should you approach the issues of clinical trials and regulation?

Price points out that, in the US, the “21st-Century Cures Act” allows the FDA to regulate any algorithm that analyzes images, or doesn’t allow a provider to review the basis for its conclusions: this could completely exclude “black-box” algorithms of the kind described above from use.

Transparency about how the algorithm functions—the data it looks at, and the thresholds for drawing conclusions or providing medical advice—may be required, but could also conflict with the profit motive and the desire for secrecy in healthcare startups.

One solution might be to screen algorithms that can’t explain themselves, or don’t rely on well-understood medical science, from use before they enter the healthcare market. But this could prevent people from reaping the benefits that they can provide.

Evaluating Algorithms
New healthcare algorithms will be unable to do what physicists did with quantum mechanics, and point to a track record of success, because they will not have been deployed in the field. And, as Price notes, many algorithms will improve as they’re deployed in the field for a greater amount of time, and can harvest and learn from the performance data that’s actually used. So how can we choose between the most promising approaches?

Creating a standardized clinical trial and validation system that’s equally valid across algorithms that function in different ways, or use different input or training data, will be a difficult task. Clinical trials that rely on small sample sizes, such as for algorithms that attempt to personalize treatment to individuals, will also prove difficult. With a small sample size and little scientific understanding, it’s hard to tell whether the algorithm succeeded or failed because it’s bad at its job or by chance.

Add learning into the mix and the picture gets more complex. “Perhaps more importantly, to the extent that an ideal black-box algorithm is plastic and frequently updated, the clinical trial validation model breaks down further, because the model depends on a static product subject to stable validation.” As Price describes, the current system for testing and validation of medical products needs some adaptation to deal with this new software before it can successfully test and validate the new algorithms.

Striking a Balance
The story in healthcare reflects the AI story in so many other fields, and the complexities involved perhaps illustrate why even an illustrious company like IBM appears to be struggling to turn its famed Watson AI into a viable product in the healthcare space.

A balance must be struck, both in our rush to exploit big data and the eerie power of neural networks, and to automate thinking. We must be aware of the biases and flaws of this approach to problem-solving: to realize that it is not a foolproof panacea.

But we also need to embrace these technologies where they can be a useful complement to the skills, insights, and deeper understanding that humans can provide. Much like a neural network, our industries need to train themselves to enhance this cooperation in the future.

Image Credit: Connect world / Shutterstock.com Continue reading

Posted in Human Robots