Atmospheric carbon gets turned back into coal, tobacco plants are repurposed as “green bioreactors,” and an artificial intelligence text generator becomes worrisomely good at its job. There’s all sorts of unexpected transformation — and more — in this week’s coolest scientific discoveries.
What is it? From the perspective of the Earth’s climate, the life cycle of coal is problematic, to say the least: Dead plant matter slowly carbonizes underground over millions of years before being dug up and burned for energy, thereby releasing carbon dioxide into the atmosphere — among other things — and helping warm the globe. But what if the cycle could be reversed — if atmospheric carbon could be turned back into coal? A new finding from researchers at RMIT University in Melbourne, Australia, points the way.
Why does it matter? Technologies for capturing carbon from the atmosphere and storing it in the ground could be crucial if humans are going to forestall the worst effects of climate change. The current technique is to convert the gaseous CO2 into a liquid, but this process has been “hampered by engineering challenges, issues around economic viability and environmental concerns about possible leaks from the storage sites,” according to RMIT. Converting the gas into solid form could be more practical. RMIT researcher Torben Daeneke said, “While we can’t literally turn back time, turning carbon dioxide back into coal and burying it back in the ground is a bit like rewinding the emissions clock.”
How does it work? The breakthrough involves the use of liquid metals as an electrocatalyst to spur the CO2 conversion. In the lab, researchers dissolved atmospheric carbon in a beaker with an electrolyte liquid and a little bit of liquid metal, to which they then applied an electrical current. The application of the current caused the CO2 to convert to solid flakes of carbon. Moreover, because the solid carbon can hold a charge, it could be used as an electrode. Dorna Esrafilzadeh, lead author on the paper published in Nature Communications, said, “A side benefit of the process is that the carbon can hold electrical charge, becoming a supercapacitor, so it could potentially be used as a component in future vehicles.”
What is it? Speaking of the land down under, Boeing Australia just introduced a new unmanned aircraft system for defense customers that centers around a concept demonstrator it calls the Loyal Wingman — a drone that uses artificial intelligence to provide support for manned fighter planes.
Why does it matter? Boeing is advertising the “teaming system” — its largest new unmanned program outside the United States — as a “force multiplier” for airborne fleets. Boeing Autonomous Systems vice president Kristin Robertson said, “The Boeing Airpower Teaming System will provide a disruptive advantage for allied forces’ manned/unmanned missions. With its ability to reconfigure quickly and perform different types of missions in tandem with other aircraft, our newest addition to Boeing’s portfolio will truly be a force multiplier as it protects and projects air power.”
How does it work? Its AI capabilities will help the Loyal Wingman fly “independently or in support of manned aircraft,” according to Boeing, while maintaining a safe distance. At 38 feet long, it’ll also have similar range to manned planes, able to fly more than 2,000 miles. The company expects the craft to fly for the first time next year. The U.S. will be “among the natural customers” for such a product, according to Reuters.
What is it? Researchers at Ontario’s Western University and Lawson Health Research Institute have found a way to use tobacco plants as “green bioreactors” to grow a particular protein that shows promise as a therapeutic agent.
Why does it matter? Interleukin 37, or IL-37, is an anti-inflammatory protein produced in small amounts in human kidneys that has potential to treat inflammatory and autoimmune conditions and diseases such as Type 2 diabetes, stroke, dementia and arthritis, according to a release from Western University. “The human kidney produces IL-37, but not nearly enough to get us out of an inflammation injury,” said Tony Jevnikar, a co-author of a new study on the protein published in Plant Cell Reports. Outside the body, the current tech for producing the protein is too expensive to churn it out at useful quantities — but Jevnikar’s team think they’ve discovered a better way.
How does it work? “Tobacco is high yield, and we can temporarily transform the plant so that we can begin making the protein of interest within two weeks,” said Jevnikar’s colleague Shengwu Ma. Now that researchers have demonstrated that tobacco can be used to grow the protein, they might be able to make green bioreactors out of other plants, such as potatoes.
What is it? Somewhat ironically given its name, the nonprofit research institute OpenAI created an artificial intelligence platform so good at generating text that it’s declining to publicly release the trained model, “due to our concerns about malicious applications of the technology.” Is JudiciouslyDisclosedAI taken?
Why does it matter? Text generated by artificial intelligence could have a multitude of beneficial uses: writing advertising copy, for instance, helping voice assistants like Siri speak more fluently. But the tech also raises the specter of “deepfakes” created for nefarious purposes such as fake news or social media profiles, as OpenAI writes in a blog post explaining its decision to play this one close: “Today, malicious actors — some of which are political in nature — have already begun to target the shared online commons, using things like ‘robotic tools, fake accounts and dedicated teams to troll individuals with hateful commentary or smears that make them afraid to speak, or difficult to be heard or believed.’” The group instead released a “much smaller model for researchers to experiment with.”
How does it work? The gist is to get the AI to decide what words should come next based on what words have come before. For their latest model, GPT-2, researchers trained an algorithm on a data set of 8.5 million web pages. They then tested the model by offering it prompts of one or two sentences and letting it write the rest, describing the AI as “chameleon-like” for its ability to produce prose resembling the source text: It was able to convincingly draft everything from celebrity news stories to “Lord of the Rings” fan fiction. There were some giveaways in terms of unnatural language — and the quality of the writing wasn’t quite Tolkienesque — it was enough to give the researchers pause. CNN has more.
What is it? Scientists at the University of Kansas and associated medical centers have produced a “lab-on-a-chip” that could detect signs of cancer from just one droplet of blood or plasma.
Why does it matter? As we’ve written here before, the ability to detect cancer early and in the blood would be a big step forward. Researcher Andrew Godwin said, “This area of study is especially important for cancers such as ovarian, given the vast majority of women are diagnosed at an advanced stage when, sadly, the disease is for the most part incurable.” The team published its findings in Nature Biomedical Engineering.
How does it work? The team zeroed in on exosomes, which a release from KU describes as “tiny parcels of biological information produced by tumor cells to stimulate tumor growth or metastasize,” and which are detected by the new diagnostic device. According to KU, “The new lab-on-a-chip’s key innovation is a 3D-nanoengineering method that mixes and senses biological elements based on a herringbone pattern commonly found in nature, pushing exosomes into contact with the chip’s sensing surface much more efficiently in a process called ‘mass transfer.’” They tested the device using samples from ovarian cancer patients, finding that it could identify the presence of cancer from just a tiny sample of plasma.