Who has a different opinion here?
Science writer Charles Q. Choi identifies a number of limitations, including a, perhaps, surprising one: AIs are very bad at math.
Scientists at Stanford University and the University of North Carolina at Chapel Hill have created a 3D-printed vaccine patch that provides greater protection than a typical vaccine shot.
The trick is applying the vaccine patch directly to the skin, which is full of immune cells that vaccines target.
The resulting immune response from the vaccine patch was 10 times greater than vaccine delivered into an arm muscle with a needle jab, according to a study conducted in animals and published by the team of scientists in the Proceedings of the National Academy of Sciences.
Scientists at Stanford University and University of North Carolina at Chapel Hill create a vaccine patch with microneedles that dissolve into the skin.
Astronaut Garrett Reisman, who helped develop SpaceX’s Crew Dragon capsule, also has experience of working with NASA.
A research team led by IBM and the Skolkovo Institute of Science and Technology (Skoltech), Russia, has created an extremely energy-efficient optical switch. This could replace electronic transistors in a new generation of computers.
A tomato with higher levels of a nutrient linked to reduced stress can now be bought in Japan – it is the first CRISPR-edited food in the world to be launched commercially.
A subatomic particle has been found to switch between matter and antimatter, according to Oxford physicists analyzing data from the Large Hadron Collider. It turns out that an unfathomably tiny weight difference between two particles could have saved the universe from annihilation soon after it began.
Antimatter is kind of the “evil twin” of normal matter, but it’s surprisingly similar – in fact, the only real difference is that antimatter has the opposite charge. That means that if ever a matter and antimatter particle come into contact, they will annihilate each other in a burst of energy.
To complicate things, some particles, such as photons, are actually their own antiparticles. Others have even been seen to exist as a weird mixture of both states at the same time, thanks to the quantum quirk of superposition (illustrated most famously through the thought experiment of Schrödinger’s cat.) That means that these particles actually oscillate between being matter and antimatter.
The newly developed battery-free system runs on harvested energy and can help massively reduce the growing e-waste problem.
A sperm’s task may appear straightforward; after all, all it needs to do is swim to an egg and insert genetic material. However, in some cases, a healthy sperm’s inability to swim may result in infertility, which affects around 7 percent of all males.
This condition is called asthenozoospermia, and there is currently no cure. However, one study conducted in 2016 and published in the journal Nano Letters has set the example for what could be possible in the future: A team of researchers from the Institute for Integrative Nanosciences at IFW Dresden in Germany developed tiny motors that can make sperm swim better as they make their way to an egg, essentially acting as a taxi.
These so-called “spermbots” basically consist of a tiny micromotor, which is basically a spiraling piece of metal that wraps around the sperm’s tail. Serving as an “on-board power supply”, the motor navigates the sperm via a magnetic field, helping the sperm swim to the egg with ease. When the sperm makes contact with the egg for fertilization, the motor slips right off, and the magnetic field doesn’t harm any of the cells involved, making it ideal for usage on living tissue, according to the researchers.
As developers unlock new AI tools, the risk for perpetuating harmful biases becomes increasingly high — especially on the heels of a year like 2020, which reimagined many of our social and cultural norms upon which AI algorithms have long been trained.
A handful of foundational models are emerging that rely upon a magnitude of training data that makes them inherently powerful, but it’s not without risk of harmful biases — and we need to collectively acknowledge that fact.
Recognition in itself is easy. Understanding is much harder, as is mitigation against future risks. Which is to say that we must first take steps to ensure that we understand the roots of these biases in an effort to better understand the risks involved with developing AI models.
The challenge of getting high-quality real-world data.
Tesla is combining manual labeling, auto labeling, and simulation to create real-world datasets for fully self-driving cars.