A new thing to come out of deep learning is something called generative design, which involves giving AI a set of constraints and then letting it “build” things. Autodesk’s DreamCatcher software can do this, and the results are often very organic-looking and in many cases use considerably less material than designs made by people. Some of these designs are terrific applications for additive manufacturing and 3D printing, because many of these objects cannot be produced using traditional methods.
Artificial intelligence may appear somewhat alien by human standards. In all likelihood, it could appear geeky, obsessive, and nerdy. This is perfectly fine if you are using AI to solve an engineering problem, but less desirable if in a personal digital assistant.
There is a big difference between comprehending language, and truly understanding what we are trying to express.
It is estimated that 7% of communication is based on words, 38% tone of voice, and 55% on nonverbal behavior (facial expressions and body language). Over time, personal digital assistants may be tuned into our emotional state, through a reading of our heartbeat, temperature, skin conductivity, and physical expression. They might even become more aware to our emotional state than our partners.
As such,
affective computing
is advancing rapidly and provides many real-world applications. Nemesysco specializes in emotional computing for stress analysis (useful in customer service) and fraud detection (transactions of any kind). Beyond Verbal is another emotion analytics company that identifies and categorizes feelings based on the human voice.
We’ve come a long way over the past few decades. When I was a teenager in the 1980’s, it was fun going to the arcade to frantically mash buttons on a machine. What happens when machines get good at pushing
our
buttons?
AI and Mental Health
AI and voice assistants have the potential to make mental health counseling relatively universal. We may find ourselves spending more time talking
to our smartphones
than we are talking to people on our smartphones. Think about that.
Whether is through meditation apps or always-on digital therapists monitoring our emotional state, we are going to have lots of help. We’ll be needing it.
AI and Decision-Making
Investing can be stressful, which is why the finance industry was among the first to embrace algorithmic decision-making. Old-school financial analysts have mostly been replaced by the new kids on the block, the “quants.” Many crowded trading floors have since gone silent, and the winners at gathering client assets over the past two decades have been robo-advisors and passively managed index funds.
Bridgewater, the largest hedge fund in the world, wants AI to make 75% of all investment decisions by 2022. More than half of the Exchange Traded funds (ETF’s) trade the market with a defined set of rules – few human managers needed.
There are still opportunities for active managers. I’ve noticed several instances over the past few years when AI’s struggled with markets when conditions deteriorated rapidly. In these times, AI-based strategies become perfectly optimized for a world that no longer exists.
Knowing what you don’t know can be an essential tool. We are years away from that level of self-awareness in AI.
AI's as Customers and Gatekeepers
In the past, we’ve thought about AI primarily on the side of
production
, but we should start thinking about AI as
a gatekeeper to the consumer.
As Kim Bates at Brain Reserve says, “The customer journey of the future will begin to emerge as Business to Robot to Consumer (B2R2C).” Virtual assistants will own the consumer relationships of the future.
Google, Amazon, and Facebook have done an amazing job of mining through our data to find out what we really want. Increasingly we may find ourselves marketing to algorithms.
You’ve already done it if you have ever tweaked a web site or blog post for search engine optimization (SEO).
As AI’s become a bigger part of our lives they will know more and more about what we like, want, and need. But they’ll get even better with two upcoming boosts in hardware technology…
Quantum Computing
The world around is an uncertain place. Subatomic particles are both particles and waves, yet we seldom notice when they blink in and out of existence. Similarly, we have difficulty predicting the exact location of an electron and can only probabilistically describe its orbit around a nucleus.
Back in 1981, Nobel prize-winning physicist Richard Feynman suggested that a quantum computer could solve difficult problems by better simulating the uncertainty of nature. With quantum computers this happens at warp-speed.
Eventually, we’ll go from a science based on slow “trial and error” in the physical world, to one in which experiments are fantastically accelerated by being performed in digital simulation
,
bringing a “golden age of discovery in new materials, new chemicals, and new drugs.”
Neuromorphic Chips
In an effort to simulate the brain, Intel, has built networks of artificial neurons into an experimental chip sets. They claim that the new breakthrough will enable computers to “think” at the level of small animal. Gartner predicts that by 2025, these chips could replace the current GPU’s used as the main hardware for AI.
Computer processing today excels at fast and precise application of rules and mathematics, while the brain performs better at tasks involving ambiguity, complexity, and creativity.
Perhaps by “re-wiring” the brains of our technology, we can make it more human.
Robotics
Moving past the brains of AI, let’s take look at their bodies.
A robot is a machine, usually one programmed by computer, that is capable of carrying out complex actions. The origins of the word “robot” comes from the Slavic root for “work”, and was first mentioned in the 1920 play R.U.R (Rossum’s Universal Robots), by Karel Capek.
In many regards, robots are the second wave of the Industrial revolution that was started in the late 1700’s. The differences between robotics and traditional machinery include flexibility, programmability, decision-making, and mobility.
While Capek originally envisioned robots as being humanoid in form, robots come in a broad range of specifications. This enables them to excel at many tasks.
In this next section, we’ll look the many industries in which robots may be deployed:
Agriculture:
At the turn of the 20
th
century, half of all Americans considered themselves “farmers.” That number has since fallen to less than two-percent, and may decrease even more.
Weeding, harvesting, planting are labor-intensive activities moving from immigrant labor to robotics. Walmart (WMT) recently filed a patent for robot bee pollinators to increase production of apple, peach, blueberries, and almond crops. Precision agriculture will be a big trend for the next decade.
Meanwhile, the protein business could
shift with the rising popularity of lab-grown meats and plant-based alternatives. Vertical farms will also greatly reduce the distance between farm-to-table.
Autonomous Vehicles:
Expect self-driving vehicles to become relatively commonplace by 2025.
Every
automotive maker has plans for autonomous vehicles – except for just two holdouts. Luxury cars made by Lamborghini and Ferrari will remain human-controlled for the foreseeable future.
This revolutionary technology will have many implications for society:
·
Car ownership may become less attractive
. If “transportation as a service” picks up, we may need fewer cars and parking spaces.
·
Without needing to buy large capacity SUV’s that do everything, we may have MUCH more variety in vehicles design
. We can pick-and-choose the right vehicle for every purpose. Take a two-seat commuter pod to work, or a car with a built-in desk. For date night, you may request a convertible or a sports car. Over the weekend, go to the hardware store in a pick-up truck.
·
Cities will need to find new streams of revenues
without traffic and parking violations. However, they may benefit from a building boom as parking lots and garages are repurposed.
·
There may be more cars on the roads
– including empty vehicles making trips for self-fueling and picking up new passengers.
·
“Sleeper cars” will become a viable alternative to air travel for many destinations
.
Hotels may feel some competition. Fitness clubs may see a resurgence as a place to freshen-up before meetings.
The biggest implication for autonomous vehicles may be decline of drivers needed for the for Ubers, taxis, trucking, and delivery.
Cybernetics:
Wearable exoskeletons may one day augment the physical capabilities of future workers. These come in a variety of configurations, including upper-body, lower-body and full-body models. Most of today’s models range in price from about $4,000 to $6,000. Ford, Boeing, Toyota, and others have all augmented a portion of their workforce with this technology.
In the military, exoskeletons are beings explored as a way for troops to carry heavy equipment for long periods of time.
“It’s not designed to give you superhuman strength; it’s designed to give you superhuman endurance,” says Zach Haldas of Ekso Bionics.
In the next step, active (powered) exoskeletons may allow for greater strength and mobility.
Drones:
One of the first robot stocks that I ever invested in was a company spun-out from M.I.T. called iRobot (IRBT). They specialized in building robots deployed for jobs that were “dirty, dull, and dangerous.”
Similar to General Electric, iRobot was an odd blend of consumer appliance company and military defense contractor. Some of their bots would scrub floors… others would remove explosives on the battlefield. Some would simply fascinate pets.
These days, the hottest drone applications are for food delivery and security.
Starship is a private company out of San Francisco that makes delivery bots that have been tested in over 100 cities. Their bots look like oversized beer coolers on wheels. They use machine learning to detect objects, and share the sidewalk with pedestrians.
Security cameras are free to roam and report. Knightscope, designs and builds drones for monitoring activity in malls, parking lots, neighborhoods and offices.
Manufacturing:
The big four of factory automation are ABB (ABB), Fanuc (FANUY), Kuka (KUKAY) and Yaskawa (YASKY)
Some industrial robots now cost about $23,000 – roughly the annual wage of an overseas factory worker. That makes for a rapid investment payoff. The are other advantages, too. Robots don’t go on sick leave, take vacation, or file complaints with HR.
The number of industrial robots is doubling every 7 years or so, and could well replace most industrial workers by 2040.
Because of high labor costs and large domestic markets, the biggest beneficiaries of automation will be S. Korea, Japan, Germany, and the U.S.
The outcome of industrial robotics will likely be higher productivity and wages for those who keep their jobs, and lower costs for consumer. Shareholders may be well-rewarded.
Medicine:
The first breakthrough robots for medicine came from Intuitive Surgical (ISRG) a decade ago. These “co-bots” (collaborative robots) assist in precision surgery. As a result, physicians are able to make smaller, more precise, incisions that heal more quickly are and less likely to become infected.
At Akara Robotics, “Violet” is an autonomous cleaning robot that rolls through rooms and hallways, sanitizing surfaces with ultraviolet light. It’s a hi-powered light bulb on wheels. “Stevie” is a social robot spending time in nursing homes, checking in on residents and making sure that they take their prescriptions.
Telepresence robots may bring in family and physicians to bedsides via videoconferencing. But, they could also enable patients to “visit” the world remotely.
In a few years, there will be an enormous demand for personal health aids that will assist the elderly in activities of daily living – eating, bathing, moving, etc.
Military:
The military has been an early-stage investor in robotics through companies such as iRobot, Lockheed Martin (LMT), and Raytheon (RTN).
In the future we’ll see bots on the ground… and in the skies.
Bots can fly faster and handle higher levels of G-force than human pilots. Within the next three years the U.S. Airforce Skyborg drone will replace the F-16 combat fighter.
Restaurants:
Fast-food kitchens now have the Burger Bot, (which can serve a fresh-made hamburger every 10 seconds), the burrito bot, and a pizza bot. The University of Texas employs a robotic barista (the “Briggo”) that serves coffee to 10,000 students a day.
We are starting to see completely automated restaurants, where food orders are taken via touchscreen, with table service performed by drone. No tipping required….
Retail and Distribution:
When Amazon (AMZN) was first started as an online bookseller, many of their orders were filled by guys skating around in warehouses on rollerblades. Workers are still on wheels, but things have changed.
In 2012 Amazon paid $775 million for an automation company called Kiva systems. Kiva makes robots for inventory management; these bots roam around the floor and move warehouse shelves without human intervention. Forty-five thousand Kiva robots delivered three-hundred items
per second
during Christmas season.
Amazon also made a $15m early stage investment in ReThink Robotics. This company makes a robot which costs about $30,000, has visual recognition, and the ability to pick and pack objects from the shelves delivered by Kiva robots.
Amazon is also pushing towards drone-based delivery, and now has developed more than 20 generations of flying drones.
You can easily find the pattern here… Amazon is its way to becoming a pioneer in completely automated order fulfillment, from shopping online to delivering a package at your door.
At warehouses, Amazon now uses 3,000 robots per every 10,000 employees – roughly twice the automation density of the U.S. auto industry.
The company is continuing to pioneer the live shopping experience, too. In a prototype Amazon Go store in Seattle, customers scan their Prime membership codes at the entrance. A combination of cameras, weight sensors, and RFID sensors track purchases. Check-out simply involves walking out the front door with a basket full of stuff, which is charged automatically to the customer’s account. At this point, the future of shopping starts to resemble shoplifting.
Robotic Process Automation:
This isn’t so much about physical robotics as it is about using artificial intelligence to do standard office work.
RPA’s are used by human resource departments, banks to manage credit limits. Insurance companies are using RPA to manage claims. Airlines can automate refund requests. Bill paying can also be easily automated.
So far, RPA technology is still in the early stages. It can handle tedious tasks to make existing employees more productive. Things won’t get serious until RPA’s start making their own decisions.
The Internet of Things (IoT)
While the internet connected computer and people, the Internet of Things will connect
stuff
. Cars, televisions refrigerators, traffic lights, trash cans, and even some frying pans now have a connection to the internet. A few implications:
·
Everything is trackable.
Few objects get lost or stolen once they are on the internet.
·
Everything is chatty, too.
For example, your car reminding you that it needs its filters changed and oil replaced.
·
Everything gets a lot smarter.
The frying pan knows how long it takes to poach an egg, even if you don’t
·
Everything wants to buy stuff for you.
Your empty refrigerator might remember to order milk and eggs, even if you forget.
By 2030, there will be five times as more objects connected to the internet than people. And these objects will all need sources of power, sensors, and microchips.
3D Printing
Most people are familiar with the basic technology… think about “printing” entire objects one drop of ink at time.
While 3D printers would once only produce objects using plastic, they can now create things with an amazing range of materials, including metal, glass, concrete, chocolate, and even organic materials.
It opens up all sorts of possibilities for customization and efficiency. In additive manufacturing, you only print what you needed. Not a drop is wasted. Design systems will let you “stress-test” and edit objects digitally before you print them into existence.
While current 3-D printers are slow, they have to potential to disrupt multiple industries.
Consider housing. Mighty Buildings has been able to satisfy U.S. construction codes while printing single-family homes at 1/10 the labor costs, and 1/3 of the total cost.
Using a 3D-printer that extrudes fast-drying concrete, Chinese company Win-Sun 3D printed ten single-family homes in less than 24-hours., at a cost of less than $5,000 each. They have since moved on to printing a 5-story apartment building over the course of a weekend.
With AI and robotics, we get cheaper housing, fresher food, less paperwork, easy access to transportation, and more affordable health care. This is all sounding pretty good, right? We need to talk about just one more thing…
Job Displacement
Kai-Fu Lee, an former top executive for Google in China, worries that AI advances will may be more disruptive to workers than the internet. He estimates that 40 percent of the world’s jobs will be lost to automation in the next 15 years. Lee also says that “AI will make phenomenal companies and tycoons faster, and it will also displace jobs faster, than computers and the internet…. It’s going to be a serious matter for social stability.”
Research is consistently finding that the effects of automation on the labor market are more highly concentrated on lower-paid, lower-skilled, and less-educated workers.
This, in turn has the potential of magnifying economic inequality. If not handled correctly, automation could be the cause of significant social unrest.
The first wave of industrialization hit assembly-line work and farming. More recently, it has accelerated substantially in the post-COVID world, by replacing human labor with “clean” automation, particularly in the retail and fast food industries.
McDonalds and Taco Bell have made the shift towards touchscreen kiosks for ordering. Checkout lines are automated, and now even the kitchen is being staffed by robots.
