Microsoft’s CEO of AI said that content on the open web can be copied and used to create new content.
Grief-laden vitriol directed at AI fails to help us understand paths to better futures that are neither utopian nor dystopian, but open to radically weird possibilities.
Generative AI models have changed the way we create and consume content, particularly images and art. Diffusion models such as MidJourney and Stable Diffusion have been trained on large datasets of scraped images from online, many of which are copyrighted, private, or sensitive in subject matter. Many artists have discovered significant numbers of their art pieces in training data such as LAION-5B, without their knowledge, consent, credit or compensation.
To make it worse, many of these models are now used to copy individual artists, through a process called style mimicry. Home users can take art work from human artists, perform “fine-tuning” or LoRA on models like stable diffusion, and end up with a model that is capable of producing arbitrary images in the “style” of the target artist, when evoked with their name as a prompt. Popular independent artists find low quality facsimilies of their artwork online, often with their names still embedded in the metadata from model prompts.
Tech companies, including Amazon Web Services, are striking deals with U.S. nuclear power plants to secure electricity for their data centers, driven by the skyrocketing demands of artificial intelligence. This move promises 24/7 carbon-free power but stirs controversy, as it could divert existing energy supplies, raise prices, and increase reliance on natural gas. These nuclear-powered data centers might accelerate the AI race, but they also spark debates over economic development, grid reliability, and climate goals. Could this be the future of tech or a risky gamble with unforeseen consequences?
As reported by WSJ, tech businesses searching the country for electrical supplies have focused on one important target: America’s nuclear power facilities.
The owners of about one-third of the United States’ nuclear power reactors are in negotiations with technology companies about providing electricity to new data centers needed to satisfy the needs of an artificial intelligence boom.
Now, you can have almost ‘any text on your phone’ read to you in the voices of legendary actors. But many are disturbed by the idea.
Chinese researchers have developed an open-source “brain-on-chip” interface system, which is the first of its kind in the world. The system can instruct a robot to avoid obstacles, track, and grasp through “mind control,” the Science and Technology Daily reported on Wednesday.
The interface system was co-developed by research teams from Tianjin University and Southern University of Science and Technology.
The system uses an artificial brain cultivated in vitro – such as a “brain-like organ,” which can interact with external information through encoding, decoding and stimulus-feedback when coupled with electrode chips, according to the report.
Terahertz technology has the potential to address the growing need for faster data transfer rates, but converting terahertz signals to various lower frequencies remains a challenge. Recently, Japanese researchers have devised a novel approach to both up-and down-convert terahertz signals within a waveguide. This is achieved by dynamically altering the waveguide’s conductivity using light, thereby creating a temporal boundary. Their breakthrough could lead to advancements in optoelectronics and improved telecommunications efficiency.
As we plunge deeper into the Information Age, the demand for faster data transmission keeps soaring, accentuated by fast progress in fields like deep learning and robotics. Against this backdrop, more and more scientists are exploring the potential of using terahertz waves to develop high-speed telecommunication technologies.
However, to use the terahertz band efficiently, we need frequency division multiplexing (FDM) techniques to transmit multiple signals simultaneously. Of course, being able to up-convert or down-convert the frequency of a terahertz signal to another arbitrary frequency is a logical prerequisite to FDM. This has unfortunately proven quite difficult with current technologies. The main issue is that terahertz waves are extremely high-frequency waves from the viewpoint of conventional electronics and very low-energy light in the context of optics, exceeding the capabilities of most devices and configurations across both fields. Therefore, a radically different approach will be needed to overcome current limitations.
The brain is the most complex organ ever created. Its functions are supported by a network of tens of billions of densely packed neurons, with trillions of connections exchanging information and performing calculations. Trying to understand the complexity of the brain can be dizzying. Nevertheless, if we ever hope to understand how the brain works, we need to be able to map neurons and study how they are wired.
Now, publishing in Nature Communications, researchers from Kyushu University have developed a new AI tool, which they call QDyeFinder, that can automatically identify and reconstruct individual neurons from images of the mouse brain. The process involves tagging neurons with a super-multicolor labeling protocol, and then letting the AI automatically identify the neuron’s structure by matching similar color combinations.
Researchers find instances of systems double-crossing opponents, bluffing, pretending to be human and modifying behaviour in tests.
Driven by the war with Russia, many Ukrainian companies are working on a major leap forward in the weaponization of consumer technology.
