She developed a construction process that turns knitted textiles into concrete buildings—saving money, carbon, and time
Mariana Popescu has developed a process and accompanying computational tools capable of turning knitted textiles into complex molds for concrete buildings. Her innovation makes it possible to build complex custom-made designs faster, with less waste and reduced carbon emissions.
“If you really want to make good structures that use less material, you end up having complicated geometries that are very often doubly curved or have other features that are difficult to mold,” says Popescu. Traditional construction that uses wood or foam supported by heavy scaffolding to create forms for pouring concrete takes months and limits what shapes are possible. All you have to do is look at a sweater, she says, to see that textile materials are perfect for making a wide array of holes, channels, and other complicated 3D shapes that are sought after in contemporary buildings.
So Popescu developed algorithms that automatically translate an architectural design into a textile-based mold that can be knitted by industrial machines in mere hours. The resulting mold is lightweight and flexible. Popescu, with the rest of her team, developed a system that uses steel cables to hold the mold in place while concrete is poured over it.
Popescu’s innovation is an efficient and ecologically conscious way of building complex structures with a minimal ecological footprint, in record time, and at low cost. It also has the potential to speed up construction of low-cost, sturdy, lightweight structures in settings like refugee camps, war zones, and sites of natural disasters.
She uses algorithms and AI to fight socioeconomic inequality
Rediet Abebe uses algorithms and AI to improve access to opportunity for historically marginalized communities. When Abebe moved from her native Ethiopia to the United States to attend Harvard College, she was struck by how vital resources often fail to reach the most vulnerable people, even in the world’s wealthiest nation. She now uses computational techniques to mitigate socioeconomic inequalities.
While she was an intern at Microsoft, Abebe formulated an AI project that analyzes search queries to shed light on the unmet health information needs of people in Africa. Her study revealed such information as which demographic groups are likely to show interest in natural cures for HIV and which countries’ residents are especially concerned about HIV/AIDS stigma and discrimination. This work is the first to use large web-based data to study health across all 54 African nations.
In an effort to inform health programming, Abebe is now taking these findings to health experts in ministries of health across the continent. She’s also working with the National Institutes of Health’s Advisory Committee to help reduce health disparities in the US.
To encourage growth in this area, she cofounded Mechanism Design for Social Good, a multi-institutional research initiative that uses algorithms to tackle problems ranging from allocating low-income housing to improving health outcomes.
Cesar de la Fuente
Digitizing evolution to make better antibiotics
Bacteria evolve faster than scientists can make new antibiotics to fight them. That’s why César de la Fuente has developed algorithms that follow Darwin’s laws of evolution to create optimized artificial antibiotics. An expert in engineering bits of protein called peptides to solve medical problems, he has also developed a method of turning toxic proteins, like one found in wasp venom, into antimicrobials. And he has mined huge existing databases of proteins in the human body to discover molecules that can kill harmful microbes.
“I wake up every day thinking about all the people that are dying in this country and around the world as a result of treatable infections, and try to come up with solutions,” says de la Fuente, who has always been fascinated by microbes’ knack for survival.
In addition to developing computer-made antibiotics, de la Fuente, an assistant professor at the University of Pennsylvania, hopes next to use the same engineering approach to find proteins implicated in psychiatric disorders like depression and anxiety and to modify them to affect brain function and behavior.
She found a better way to correct single-gene mutations
Nicole Gaudelli invented a way to potentially correct almost half of all genetic diseases caused by single-gene mutations.
Gene-editing tools such as CRISPR can fix some of the single-letter genetic “spelling” mistakes that can drive inherited diseases. But they don’t correct for having the nucleic acid adenine, or A, appear in a DNA strand where there should have been a guanine, or G. This misplaced A is involved in sickle cell disease, cystic fibrosis, Parkinson’s, Alzheimer’s, and many types of cancer.
So Gaudelli set out to make a new enzyme that can cleanly convert A-T base pairs into G-C base pairs with few undesired effects.
“It was a little bit of magic,” says Gaudelli, about getting her enzyme to work. She’s now a senior scientist at Beam Therapeutics, a biotech company based in Cambridge, Massachusetts, working to commercialize her approach.
She’s using AI to help dream up a new generation of lighter, stronger materials
Grace Gu is using artificial intelligence to find ways to make better materials. Gu envisions materials that can be used for lighter and stronger body armors, 3D-printed and customizable medical implants, and tunable solar cell materials that push the boundaries of the renewable energy technology.
Gu’s work is inspired by natural materials such as seashells and bamboo, in which the structure of the base constituents results in strength and other desirable properties. Her team at UC Berkeley uses machine learning algorithms to discover new composite structures based on nature’s examples. This approach allows her to design materials that are superstrong and yet lightweight. These designs are then 3D-printed and tested to validate the algorithm, to make sure that the hypothetical materials work in the real world.
Thus far, Gu’s research has led to material designs with dramatically enhanced mechanical properties. And as the team continues its research, Gu hopes that bigger breakthroughs are around the corner.
Making the software that lets powerful AI programs run more smoothly
AlphaGo, the artificial intelligence that beat the best human player at Go in 2016, needed nearly 2,000 central processing units and 300 graphics processing units to function. As a consequence, its electricity bills were $3,000 per game. Song Han has designed software and hardware that enable powerful AI programs like AlphaGo to be deployed in low-power mobile devices.
The “deep compression” technique Han invented make it possible to run in real time on a smartphone AI algorithms that can recognize objects, generate imagery, and understand human language. Facebook, among other companies, uses Han’s software design to reduce the amount of computation an AI algorithm that can recognize objects needs. This allows people to use their smartphone camera to pinpoint objects in the real world and then add digital visual effects.
In 2016, based on his innovations, Han cofounded an AI chip company called DeePhi Tech, which Xilinx, an American semiconductor company, acquired last year.
In his new role as an assistant professor at MIT, Han is automating the design of AI algorithms. The goal is to “let any non-expert push a button and design compact neural networks,” he says, referring to the computing systems loosely modeled after the human brain that are central to how AI works.
Software developers without AI expertise, he says, would be able to use such neural networks to classify objects, improve the resolution of images, and analyze videos more efficiently.
His tiny robots can be programmed to treat infection
Jinxing Li pioneered the use of tiny robots—just a few micrometers across—to treat disease in a living animal.
Li designed rocket-like micromotors that run on gut fluids in a living animal and biodegrade after completing their mission.
The bots are made from polymer-coated balls of magnesium, which react with stomach acid to create hydrogen bubbles that propel them through the gut. Li and collaborators loaded one of the polymer layers with antibiotics, and the bots were administered to mice with stomach infections. On entering the stomach, they fired into the lining and stuck to the stomach wall before gradually dissolving to release their cargo over a long period to treat the infection.
Li recently showed that magnetically powered nanomotors cloaked in membranes from platelet cells could navigate efficiently through blood to remove toxins and pathogens without being cleared by the immune system or getting covered in sticky biomolecules, as foreign particles normally do.
The next step is to create “cyborg cells,” says Li, by taking the body’s immune cells, which hunt and destroy bacteria or cancer cells, and merging them with nanobots to navigate toward the disease site.
He taught a robot hand how to figure out things on its own
Wojciech Zaremba led a team that used machine learning to train a robot hand to teach itself to pick up a toy block in different environments. The robot was tasked with figuring out on its own how to accomplish the complex task of grasping a block and twisting it around with its robotic fingers in response to commands.
Zaremba powered the robot through a neural network, a computer program that mimics the type of networks our brains use.
Although reinforcement learning has been used before in robotics, it hasn’t worked on anything as complicated as a robotic hand, because the numerous tasks involved would require the equivalent of hundreds of years of experience. And robotic AIs trained in virtual worlds have typically failed to transfer successfully to reality, owing to the gap between simulated and real-world physics.
Zaremba, a cofounder of the AI research group OpenAI, hypothesized that varying the conditions in a virtual environment coiuld prepare a neural network for the messiness of reality.
He randomized 254 physical parameters—things like the mass of the block and the friction of fingertips—and found that the hand, after training, could manipulate the block the first time it was set loose in the real world.