10 Breakthrough Technologies of 2018 According to MIT
Journalists from MIT Technology Review have published their annual list of technologies that, in their opinion, will rapidly develop in the coming years. Some of them already exist, while others are still in development, but all are worth watching now. Here’s a look at these technologies.
1) 3D Metal Printing
Although 3D printing has existed for several decades, it was long limited to experimenters and designers creating one-off prototypes. Recently, however, the situation has changed: more and more companies are developing affordable 3D printers designed for manufacturing large metal parts. For example, in 2017, the American startup Markforged released its first metal 3D printer costing less than $100,000. Desktop Metal developed software that can independently design part models and speed up production by 100 times. That same year, General Electric, which has been testing 3D printing for aviation manufacturing, unveiled a prototype 3D printer capable of printing parts up to 1 meter in diameter. The company plans to start selling the printer in 2018 and is also developing a printer for even larger parts.
2) Artificial Embryos
This year, researchers at the University of Cambridge grew an embryo from stem cells for the first time—without using an egg or sperm. The experiment was conducted on mice, but theoretically, it could be repeated with human cells. Scientists used embryonic stem cells and extra-embryonic trophoblast stem cells (which form the placenta), as well as an extracellular matrix—a 3D scaffold that allows cells to develop and form tissues outside the body. After 96 hours, the cells formed a structure with the same characteristics as a mouse embryo at 96 hours post-fertilization. “Both embryonic and extra-embryonic cells interact and organize into a structure that looks and behaves like an embryo. It has anatomically correct sections that develop in the right place at the right time,” explains lead researcher Magdalena Zernicka-Goetz.
After 96 hours, development stopped because the embryo lacked a source of oxygen and nutrients. In future experiments, scientists plan to add yolk sac stem cells to the 3D scaffold, which may extend embryo development to the beginning of organ formation.
3) The “Sensing City”
In October, Sidewalk Labs, a subsidiary of Alphabet, announced plans to build a high-tech neighborhood called Quayside on Toronto’s waterfront—a “sensing city” covered with a network of sensors to collect detailed data on city operations, from air quality to noise levels. Based on this data, Sidewalk Labs plans to develop software to optimize city functions, such as traffic management. All software created for Quayside will be open source, allowing other developers to use it in their own smart city projects.
4) “Artificial Intelligence for Everyone”
Until recently, AI development was the privilege of large tech companies like Amazon, Baidu, Google, and Microsoft. However, the rise of cloud technologies is making AI development accessible even to startups without significant funding. Microsoft, with its Azure cloud platform, partnered with Amazon to create Gluon, an open-source deep learning library. Gluon aims to make building neural networks as easy as creating standard smartphone apps. Google also announced the launch of Cloud AutoML, a cloud-based AI development system.
According to MIT, the availability of cloud platforms will accelerate the adoption of AI technologies in medicine, energy, and industry.
5) Generative Adversarial Networks (GANs)
In 2017, neural networks learned not only to recognize images with high accuracy but also to create new ones that are sometimes indistinguishable from real photos. Generative adversarial networks (GANs), used for this purpose, consist of two parts: a generator and a discriminator. They are trained on the same dataset—for example, in Nvidia’s experiment, on photos of Hollywood celebrities. The generator learns to create new images based on these, while the discriminator checks their “realism.” “The main idea is to gradually increase the size of the generator and discriminator, starting with low-resolution images and adding new layers for higher-resolution details as training progresses,” explain Nvidia representatives.
That year, Nvidia also used this technology to create training sets for self-driving car systems. According to developers, by using GANs and unsupervised learning, they developed artificial “imagination”—the ability for algorithms to envision how an object would look under different conditions, such as different weather. This means that in the future, algorithms may not need large, manually labeled datasets for training; a single source image could suffice, with the generator processing it under various conditions.
For example, the neural network could turn a winter road into a summer one, or a dry road into a rain-soaked one.
6) Real-Time Translation Earbuds
In Douglas Adams’ science fiction novel “The Hitchhiker’s Guide to the Galaxy,” characters used a Babel fish in their ears for instant translation. This year, Google released a similar gadget—Pixel Buds, $159 earbuds that provide near-simultaneous translation in 40 languages. Currently, they only work with Pixel smartphones. To converse with a foreign language speaker, the user wears the earbuds and keeps their phone close to the other person. Pixel Buds are available in the US, UK, Australia, and Germany.
The device runs on Google Translate, which has become much smarter in the past two years. According to Google, the new Machine Translation Neural version makes 55-85% fewer errors than the previous version.
7) Clean Natural Gas
In the US, Net Power has built an experimental gas power plant that, thanks to a new fuel processing technology, will not emit CO2 into the atmosphere. The gas is burned in a turbine ten times smaller than those in conventional plants. It fits in a 60-square-foot room, and the entire plant occupies less space than a football field.
Most of the carbon dioxide released from natural gas under high pressure and temperature is used to keep the turbine running, while the rest is compressed into a dense state. It can be buried underground or sold to other companies—for example, for making cement, plastics, and other carbon-based materials.
The experimental plant will be connected to the grid in 2018, producing 50 megawatts of energy—enough to power 40,000 homes. Its efficiency is expected to be 80%, compared to the US average of 60% for conventional power plants.
8) True Online Privacy
True privacy online may finally become possible thanks to a new cryptographic protocol called zero-knowledge proof. With this protocol, a user can, for example, prove their solvency without revealing how much money is in their account. Zero-knowledge proof is already used by American banks JP Morgan Chase and ING.
Although scientists have worked on this technology for about a decade, it became widely known only last year with the launch of the cryptocurrency Zcash. Unlike Bitcoin and other popular cryptocurrencies, Zcash offers maximum anonymity: all transactions are confirmed in a distributed ledger, but it’s impossible to trace the participants’ addresses. This uses the zk-SNARK cryptography method, developed by Israeli scientists in collaboration with MIT. All transactions are confirmed on the network, but participant addresses cannot be tracked.
9) Genetic Fate Prediction
Thanks to a series of genetic studies in recent years, it is now theoretically possible to determine a person’s predisposition to most diseases by analyzing their DNA. For example, in California, doctors are testing a program that can predict a patient’s risk of developing coronary heart disease based on their DNA. This year, Myriad Genetics launched an algorithm that can calculate the risk of breast cancer for any woman of European descent. These developments are based on polygenic analysis—studying thousands of a patient’s genes rather than just one, which significantly increases accuracy.
Scientists predict that within three years, genetic prediction technologies will be implemented in healthcare, allowing for personalized vaccination and screening schedules for each patient. Pharmaceutical companies will also be able to use genetic predictions when developing and testing drugs. Knowing which diseases most people are vulnerable to, they can produce the most in-demand medications and test them on volunteers with the highest risk of illness.
10) A Quantum Leap in Materials
In 2017, researchers began exploring the potential of creating new molecules using quantum computers—devices that use quantum superposition and entanglement to process and transmit data, enabling calculations 100 million times faster than conventional computers.
IBM scientists have already created a simple three-atom molecule using a 7-qubit quantum computer. MIT predicts that in the near future, scientists will be able to create more complex and interesting molecules, as quantum computers with more qubits and advanced algorithms are being developed. This year, Microsoft released a free beta version of a software development kit for future quantum computers. The kit includes a new programming language, Q# (Q Sharp), designed specifically for quantum computing, and a 40-qubit quantum computer simulator, allowing developers to test their programs on a regular computer or via the Azure cloud platform. Microsoft assures that all programs created with this kit will eventually run on a real quantum computer, which the company is currently developing.