LAS VEGAS -- A tech company's most compelling product is the future it's creating. In practical terms, that means the products it's selling -- the data-crunching tools, the communications-management suites, the collaboration workspaces, the networking and security underpinnings. "Buy the product and the job becomes easier" is the implicit promise.
In slightly loftier terms, the future a tech company sells is the one where the products are secondary to the purpose you're using them -- a future where people who use the company's products are doing big things to improve the world.
The future IBM promised in its Research Science Slam on Monday afternoon here at IBM Think 2018 is one where you can trust your data to be secure, your AI free of data bias, your algae to be working hard to keep us all in clean air, and your computing to be quantum.
What is a Science Slam? A slam, as Senior Vice President Arvind Krishna explained, is a way to quickly convey ideas in a format for all audiences. "It gives the researchers a chance to present their work in a series of short, narrative stories."
In Monday's Science Slam, five presenters, picked from 3,000 scientists in 12 labs across six continents, talked about five technologies they think will change the world in the next five years.
Cryptoanchors to Fight Counterfeiting
Andreas Kind, who leads industry platform and blockchain at IBM Research Zurich in Switzerland, began by explaining the threat of counterfeit products. It goes beyond "this fake Louis Vuitton handbag dings a luxury conglomerate's quarterly profits." Counterfeiting is a public safety threat, affecting everything from medicine to cars. In fact, Kind said, 40 percent of aftermarket automotive parts are actually counterfeits.
"The root of the problem is that global supply chains have become very complex, with many participants," Kind said. As a result, it's easy for one link in the chain to be swapped out for a bogus one.
The problem with inventory-tracking systems right now is that what's digital stays digital -- say, a product ID number -- and what's in the physical world stays there, wholly separate. Even if a product is tagged with a physical tag, it's not hard to break the supply chain monitoring.
"The trust has to reach into the physical world," Kind said. "We need anchors to connect encrypted database entries with objects in the real world."
Enter cryptoanchors. These are embedded into the very nature of an item -- printed on a malaria test, for example -- and they are both physical marker and a collection of code that cannot be easily hacked. (That's where the crypto part comes in.) They correspond to an encrypted database entry. And if there's a mismatch -- well, there's your broken link in the supply chain.
Kind projects that cryptoanchors could cut the amount of counterfeits across the globe by 50 percent over five years.
Cecilia Boschini, a predoctoral researcher in lattice-based cryptography at IBM Research Zurich, posits that number theory will be what guarantees the safety and security of our data.
How? She began with the premise that "in math, once you know the fundamentals, the only limits are your imagination and the rules of logic."
The cryptography systems of today rely on very difficult math problems as the basis for their security. However, the rise of quantum computing means more computing power to throw at these math problems, and it will take much less time to solve them.
"We only need some quantum-resistant protocols and we're done," Boschini said. "We need problems that can somehow grow with us."
She then walked the audience through an extremely basic example of a lattice equation -- a way to determine the ordering of points on a two-dimensional grid -- and explained that a lattice equation on a two-dimensional grid is fairly easy to solve. But keep piling on the dimensions, and lattice equations become much more difficult to solve. They are the problems that grow with the increasing demands for security as quantum computing becomes more widely deployed.
Thomas Zimmerman, a research staff member and master inventor at the Almaden Research Center in San Jose, California, explained how his passion for pulling gadgets apart and repurposing them for entertainment inadvertently led to a passion for plankton and an appreciation of what they do for the ecosystem.
Ocean phytoplankton are responsible for approximately 70 percent of the world's oxygen production; they also make the planet habitable for us by "fixing" a lot of carbon dioxide, i.e., converting carbon dioxide into organic compounds through photosynthesis, thus leaching it out of the atmosphere so it doesn't help lock in heat to raise temperatures.
Zimmerman had been fooling around with the camera in a smartphone when he discovered he could repurpose it as a photo-snapping microscope. Eager to find subjects to view at large magnifications, he picked up some plankton-filled water at a pet store, then quickly became fascinated by the creatures and their role as an indicator of environmental health. Some scientists believe that phytoplankton production has dropped by 40 percent since the 1950s in response to steadily rising global temperatures.
Monitor the plankton and you can get a sense for how healthy the planet is, Zimmerman said. To do that, he's used his time at IBM to invent what he calls "AI microscopes" to monitor aquatic health. Using an image sensor and two lenses in a waterproof container, he's got a simple monitor. To handle the tremendous amount of data generated by the plankton observations, Zimmerman took the AI chips from smart cameras and retrained their neural networks to recognize plankton instead of faces.
"By using these AI microscopes, we can network them through the cloud and use them to alert scientists," Zimmerman said. And thus, with continuous data monitoring, we'll have real-time alerts about the health of the planet's oxygen generators.
Bias-Free Artificial Intelligence
Francesca Rossi, the AI ethics global leader for IBM, argued that it's in our best interests to ensure our artificial intelligence systems are as bias-free as possible -- and right now, a lot of artificial intelligence is biased because it's based on flawed and faulty data sets courtesy of, well … humans.
The example Rossi gave: teaching an AI system to recognize human faces. If you don't pick sufficiently diverse photos to train the system, it might not accurately recognize, for example, old people, nonwhite people, women or children.
"We would be doing ourselves a disservice if we didn't look critically at our systems to ensure they're as helpful as possible," Rossi said. Therefore, if you want your AI to be as useful and helpful as possible, you'll want the data system to be fair.
You'll also want the system to learn how to differentiate between faulty bias -- the exclusion or disadvantaging of one group -- and a bias rooted in expertise. For example, Rossi said, if an AI system were dealing with medical problem-solving, you might want it to be biased in favor of doctors' expertise.
"The more we learn about and understand AI bias, the more we understand our own biases," Rossi said. To do so, she recommends multidisciplinary AI teams, pulling people from psychology, philosophy, economics and topical disciplines.
"Most current AI systems are biased, but we believe in the next five years, the bias will be tamed and eliminated by companies," Rossi said. And those who don't tackle data bias head-on will find themselves out of business.
Quantum Computing as a Multidisciplinary Approach
Talia Gershon, a research staff member at the Thomas J. Watson Research Center in New York, began by asking how many members of the audience had been hearing about quantum computing. Nearly everyone's hands went up. Then she asked how many people understood quantum computing. Hands dropped.
In the next five minutes, she gave two pieces of advice: Don't apply classic linear logic to thinking about quantum computing, and understand that quantum computing doesn't exist in a vacuum.
To the second point, she said, "Quantum computing is not just a game for physicists." In order to bring the computers -- which require operating temperatures of 460 degrees Fahrenheit below zero -- into working state, materials science researchers, cryogenics researchers and computer scientists will all have to work together.
Quantum computing will also percolate out of its (chilly) computing center, with the IBM Q Network offering access to quantum computing for universities and corporations and the IBM Q Experience offering individual users the opportunity to run algorithms and experiments on IBM's quantum processor via IBM Cloud, using the Quantum Composer and QISKit software developer kit.
Gershon also pointed out that quantum computing and AI will help each other progress: As AI gets more complex, it will accelerate quantum computing, and as quantum computing brings more raw computing power online, it'll boost artificial intelligence capabilities.