Three Future-Defining MIT Breakthrough Technologies, Explained

Originally published on June 5, 2017.
No matter what role you find yourself playing in the tech industry, it seems that every hobbyist, hacker, or startup entrepreneur in the business all end up asking themselves the same two questions as they develop their biggest plans: what is the future going to look like, and how can I get there first?
Forward-thinkers and futurists exist in every industry, but in tech they seem to be the rule rather than its exception — and with that many brilliant folks all trying to solve the same set of problems, it’s easy to feel like you might be missing out on the Next Big Thing even as you’re reading this.
Well, worry no longer — we sent our team of interns into the field to MIT’s annual Breakthrough Technologies conference in Cambridge to break down some bleeding-edge innovations in a way that everyone can understand.
#1 — Practical Quantum Computers:
For keynote speaker and Technology Review Senior Editor Antonio Regalado, quantum computing has long been exciting, but never practical enough to be considered seriously as a world-changing innovation. “Every year,” he stated, “we discuss quantum computing as a candidate for this list, and every year we decide: well, maybe next year.”
Well, that year is finally here. And don’t worry, if you need a refresher on what quantum computing means, we’ve got you covered.
Essentially, through quantum computing, computers can complete memory and processing tasks much faster than ever before. This is accomplished by using the power of subatomic particles, which have the unique ability to be in more than one state at any given time.
Two states at the same time? Thanks to the power of quantum mechanics, it’s possible: meet the quantum bit, or “qubit”. A qubit is a unit of data comparable to the classical bits (1s and 0s) that you and I use in contemporary computers, but qubits have more power than classical bits for two key reasons. Firstly, qubits are capable of representing a superposition of both 1 and 0 at the same time in addition to either 1 or 0. And secondly, positions of individuals qubits can affect other qubits.
What all this means is that quantum computers are able to “hack” the way data typically moves, enabling us to take shortcuts and achieve answers in certain types of calculations contemporary technologies cannot. In other words, quantum computers get to perform much faster than other computers because they aren’t built on the numerical rulebook.
The reason quantum computing finally made the jump to MIT’s list this year is because companies like Google, IBM, Intel, and Microsoft are finally investing in the project, allowing key research and development to become a reality. To date, scientists have built fully programmable five-qubit computers and even a handful of 10- to 20-qubit systems. This actually isn’t much, considering top supercomputers already can perform the same feats as these test quantum computers.
However, in a few years, somewhere between 30 and 100 qubit computers will become commercially available, and we’ll be able to harness the ability to rewrite encryption, materials science, pharmaceutical research, and artificial intelligence. There’s even talk that 100,000-qubit systems will come to fruition and will have the potential to disrupt the materials, chemistry, and drug industries by producing completely accurate molecular models that would lead to the discovery of new materials and drugs.
While quantum computing of that magnitude is years down the line, we can all get started experimenting with the next generation of data when commercially available quantum computing units become accessible within 4–5 years.
#2 — The Cell Atlas:
You might think a full map of all 37.2 trillion cells that make up the human body is a project so massive and so fragile that completing it would take decades of labor. But a team of top scientists from across the globe tasked with this effort intend to catalog every cell in the human body into a robust atlas within the next five years.
As difficult as this sounds, scientists are leveraging three innovative technologies along the way that will allow them to look into each cell and view active molecules with high degrees of both accuracy and speed.
The first is called “cellular microfluidics”, which allows cells to be separated and tagged on the molecular level through a process of propelling individual cells down artificial capillaries. Scientists are able to view active genes in each individual cell thanks to this method.
But working with individual cells would take way too long — that’s why their second tool allows scientists to process 10,000 cells a day! Sequencing machines give them the ability to classify active genes at scale, no longer needing to propel cells one-by-one.
Finally, their third technology employs highly advanced labeling techniques that allow cells to be located based on their gene activity within any given organ or human tissue.
Top tech executives have already helped to back this effort to create this game-changing atlas of our bodies, and once it becomes available to healthcare providers, the world of medicine is likely to change for good.
#3 — Reinforcement Learning:
It seems like everyone is talking about self-driving cars these days, but how are unmanned vehicles going to learn to merge at 75 miles per hour on the freeway, or navigate a rotary correctly? That’s exactly what reinforcement learning aims to solve.
Essentially, scientists have found a way for computers to assign value to each right or wrong move they make during the learning process. These values are stored and the computer continuously updates the values while it learns, making the learning process much more robust and fluid.
Combined with deep learning, which works within large neural networks to pick up on data patterns, reinforcement learning stands to allow computers to improve their performance in complex contexts better than ever before.
We’ve seen a few successful events using these technologies over the past decades, but none as successful as AlphaGo. The AlphaGo software, trained with reinforcement learning, beat the world’s best human player of an abstract strategy game called Go for the first time this year. This was something researchers thought would take years to happen, especially in a game as complex as Go where from time to time players struggle to explain the motives behind their moves.
As researchers aim to use reinforcement learning more frequently going forward, they’re attempting to find ways for computers to better handle complex situations that have more than one outcome. With this technology being only a year or two away from wider use, the possibility of pairing reinforcement learning with the quantum computer holds outrageous potential for machine learning that could change technology forever.
—
These technologies certainly hold a lot of promise, but they make up only a part of the groundbreaking products discussed at MIT last week. In the next edition of our newsletter, we’ll be breaking down even more of these game-changing technologies, so be sure to subscribe if you haven’t already.
For more information about MIT Technology Review, their website is available here.
