Summary: Researchers developed a new machine learning technique to improve red-teaming, a process used to test AI models for safety by identifying prompts that trigger toxic responses. By employing a curiosity-driven exploration method, their approach encourages a red-team model to generate diverse and novel prompts that reveal potential weaknesses in AI systems. This method has …
Computer-science students are being shielded from the liberal arts. That may be a problem. Illustration by The Atlantic. Source: Max Whittaker / The New York Times / Redux March 22, 2024, 1:17 PM ET This is Atlantic Intelligence, a limited-run series in which our writers help you wrap your mind around artificial intelligence and a …
This illustration shows electric current being pumped into platinum (the bottom slab), which results in the creation of an electron spin current that switches the magnetic state of the 2D ferromagnet on top. The colored spheres represent the atoms in the 2D material. Credit: Courtesy of the researchers An MIT team precisely controlled an ultrathin …
Lightmatter, a company founded by MIT alumni, is pioneering the use of light for data processing and transfer to address the limitations of traditional computing methods. (Artist’s concept.) Credit: SciTechDaily.com Lightmatter, founded by three MIT alumni, is using photonic technologies to reinvent how chips communicate and calculate. Our ability to cram ever-smaller transistors onto a …
The fractional quantum Hall effect has generally been seen under very high magnetic fields, but MIT physicists have now observed it in simple graphene. In a five-layer graphene/hexagonal boron nitride (hBN) moire superlattice, electrons (blue ball) interact with each other strongly and behave as if they are broken into fractional charges. Credit: Sampson Wilcox, RLE …
Only 23% of workers’ wages could be cost-effectively replaced by AI, study said. Artificial Intelligence (AI) might not be coming for your jobs just yet. A new study from the Massachusetts Institute of Technology (MIT) has found that AI might still be expensive for companies to replace human workers. The research looked into the practicalities …
Researchers have developed a novel synaptic transistor that mimics the human brain’s integrated processing and memory capabilities. This device operates at room temperature, is energy-efficient, and can perform complex cognitive tasks such as associative learning, making it a significant advancement in the field of artificial intelligence. Credit: Xiaodong Yan/Northwestern University A transistor conducts energy-efficient associative …
Image Credits: Kirillm / Getty Images Will AI automate human jobs, and — if so — which jobs and when? That’s the trio of questions a new research study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), out this morning, tries to answer. There’s been many attempts to extrapolate out and project how the …
Researchers at , which focuses on computer engineering and , built two machine learning algorithms that can detect pancreatic cancer at a higher threshold than current diagnostic standards. The two models together formed to create the “PRISM” neural network. It is designed to specifically detect pancreatic ductal adenocarcinoma (PDAC), the most prevalent form of pancreatic …
Who do you trust on the Internet? A better question might be: is trust even possible on the Internet? And by extension: is trust possible anywhere in media? With these questions in mind, let’s look at some of the work being done around building trust online. Here’s another question: what does all of this have …