00:00:00 – Explaining model jaggedness
00:09:39 — Emotions and value functions
00:18:49 – What are we scaling?
00:25:13 – Why humans generalize better than models
00:35:45 – Straight-shotting superintelligence
00:46:47 – SSI’s model will learn from deployment
A paper published in Biology Methods and Protocols, finds that it is now possible to distinguish wild from farmed salmon using deep learning, potentially greatly improving strategies for environmental protection. The paper is titled “Identifying escaped farmed salmon from fish scales using deep learning.”
Norway is home to the largest remaining wild populations of wild salmon and is also one of the largest producers of farmed salmon. Atlantic salmon abundance in Norway has declined by over 50% since the 1980s and is now at historically low levels. Escaped farmed salmon are an important reason for this decline.
Norway produces over 1.5 million metric tons of farmed Atlantic salmon annually. Each year, however, approximately 300,000 farmed salmon escape into the wild.
Machine learning models called convolutional neural networks (CNNs) power technologies like image recognition and language translation. A quantum counterpart—known as a quantum convolutional neural network (QCNN)—could process information more efficiently by using quantum states instead of classical bits.
Photons are fast, stable, and easy to manipulate on chips, making photonic systems a promising platform for QCNNs. However, photonic circuits typically behave linearly, limiting the flexible operations that neural networks need.
A warm hand is enough to drive motion in tiny Salmonella-inspired robots that harness molecular-level dynamic bonding.
A team of researchers from China and the U.S. came together to design soft robots with a coordination-motorized oscillator (CoMO) that can make self-sustained micromovements by harvesting small amounts of energy from sunlight or body heat. At the heart of this innovation is a new supramolecular polydimethylsiloxane (PDMS)-based elastic polymer dynamically crosslinked by Eu3+ at the center.
The findings are published in Angewandte Chemie.
If you’ve spent any time with ChatGPT or another AI chatbot, you’ve probably noticed they are intensely, almost overbearingly, agreeable. They apologize, flatter and constantly change their “opinions” to fit yours.
It’s such common behavior that there’s even a term for it: AI sycophancy.
However, new research from Northeastern University reveals that AI sycophancy is not just a quirk of these systems; it can actually make large language models more error-prone. The research is published on the arXiv preprint server.
“Campaign leverages fake adult websites (xHamster, PornHub clones) as its phishing mechanism, likely distributed via malvertising,” Acronis said in a new report shared with The Hacker News. “The adult theme, and possible connection to shady websites, adds to the victim’s psychological pressure to comply with sudden ‘security update’ installation.”
ClickFix-style attacks have surged over the past year, typically tricking users into running malicious commands on their own machines using prompts for technical fixes or completing CAPTCHA verification checks. According to data from Microsoft, ClickFix has become the most common initial access method, accounting for 47% of attacks.
The latest campaign displays highly convincing fake Windows update screens in an attempt to get the victim to run malicious code, indicating that attackers are moving away from the traditional robot-check lures. The activity has been codenamed JackFix by the Singapore-based cybersecurity company.
AI chip arena gears up for epic battle as Meta plans to ditch NVIDIA for Google TPUs.
(https://interestingengineering.com/ai-robotics/meta-google-t…ta-centers)
Zuckerberg-led Meta could shift to Google TPUs for AI workloads by 2027, challenging Nvidia’s data center dominance.
New Warner–Suno deal lets creators legally use artist voices and likeness in AI tracks.
It was only last year when Hollywood unions battled AI use in creative industries. Today, Warner Music Group is handing artificial intelligence a front-row seat in the music business.
The company has signed a licensing deal with Suno, the fast-growing AI music platform, marking one of the biggest shifts yet in how the industry approaches AI-generated music.
Under the agreement, Warner Music Group will allow users to generate songs using the voices, likenesses, names, images, and compositions of artists who voluntarily opt in.