The Frontier - SiliconNoon (Page 22)

The Frontier

Anthropic's automated researcher beats humans at $4 an hour

Two research stories landed today on the same theme: machines doing work that used to be a human's job. In one corner, an Anthropic fellow shows that an LLM-driven research loop can already out-design its human handlers on alignment post-training. In the other, Meta is testing robots that pull the cables inside its data centers. Anthropic's Automated Alignment Researcher (AAR) is cheaper, faster, and — in a six-hour window — better than the human researchers it replaced. A paper from Anthropic

The Frontier

DeepMind's Co-Scientist now plans the experiment, runs the furnace, and writes the paper

Google DeepMind's Co-Scientist has crossed a line that the lab's earlier versions were careful to stay behind. The system, built on current Gemini models, no longer just proposes hypotheses — it now designs experiments, writes the code or machine-readable lab protocol that runs them, digests the results, and drafts a scientific manuscript. In a paper out today, the team describes a closed-loop research workflow that they validated across three disciplines: materials science, computer science, an

The Frontier

StreamPI teaches VLA models to think in time, not frames

Two robotics papers out of China this week are quietly rewriting what "embodied AI" actually has to know. A VLA model from Da Xiao and HKU learns to read continuous time, and a single-video pipeline from Niuwa, SJTU, and Shandong lets a robot push through a door it has only ever seen on tape. Both push past the single-frame ceiling — and both suggest the next wave of robot intelligence is less about bigger models and more about the dimensions they were ignoring. StreamPI gives VLA models a work

The Frontier

Tencent open-sources Hunyuan Hy4 preview — 770B params, 1M context

China's open-weight race just got its heaviest new contender, and Tencent is pitching it as a productivity workhorse rather than a benchmark-chaser. Tencent is open-sourcing Hy4 preview, a 770 billion-parameter Mixture-of-Experts model with a 1 million-token context window. The architecture activates only 49 billion parameters per token — a 78-layer stack where each layer beyond the first holds 256 routed experts plus one shared expert, with eight experts firing per token. It ships under the A

The Frontier

Ex-OpenAI VP: human AI researchers have 'two years' left in the driver's seat

The man who built OpenAI's o1 and o3 reasoning models now runs a lab whose stated goal is to make researchers like him obsolete — and he has put a number on how fast it will happen. Jerry Tworek, the former OpenAI research vice president who led the o-series reasoning models, said in a recent interview that human researchers have roughly two years left to play a meaningful role in AI research before the entire loop — from idea generation to validation — is automated. Tworek left OpenAI in April

The Frontier

Amap's ABot-Recon turns 12 frames into a 10,000-frame 3D scene

Two signals out of China this hour — one a new kind of spatial-intelligence model, the other proof that a frontier model can now run cheaply on domestic chips. Alibaba's Amap has released ABot-Recon, a streaming 3D-reconstruction model that rebuilds 10,000-frame scenes while looking at only 12 frames at a time. Announced August 28 and open-sourced with weights on Hugging Face and ModelScope, the model takes a single monocular RGB video stream — no depth sensor, no pre-known camera parameters —

The Frontier

Manycore's spatial-intelligence world model enters internal beta

The company behind Kujiale — one of China's biggest interior-design platforms — just made its clearest declaration yet that it wants to be infrastructure for the spatial-AI era, not just a design-software vendor. Its first interim results since listing show the AI business is now the only part of the company growing fast. Manycore Tech said its self-developed world model has entered internal beta, anchoring a pivot the company itself frames as becoming the "water seller" of spatial intelligence

The Frontier

Terminal-Bench-Science: the best AI agents solve only 30% of real research tasks

A new benchmark built from researchers' own lab work is drawing the first honest line between AI agents that can talk about science and agents that can actually do it. The answer is humbling: the best system in the world clears fewer than a third of the tasks. Terminal-Bench-Science 0.1 launches today with 70 tasks drawn from real scientific workflows across the life, physical, Earth, mathematical, and engineering sciences. Built by the Harbor framework team with Stanford's AI lab and domain ex

The Frontier

Google's Gemini Omni 1.1 Flash adds studio-grade video control

DeepMind rolled out Gemini Omni 1.1 Flash on Wednesday, a production-ready upgrade to its generative-video model that adds the editing controls professional creators have been asking for. It's a meaningful step toward treating AI video as something you direct rather than just generate. The headline additions are about control. Scene extension now lets the model analyze up to ten seconds of prior context — a leap over older versions, which only referenced the final second — so it can continue fo

The Frontier

Google pilots double-blind AI evals to stop benchmark cheating

Two moves this hour quietly attack the same weakness in how we trust AI: how to know a model is actually good — and how platform ages are confirmed. Google thinks cryptography can fix the first; Meta is leaning on Apple and Google to fix the second. Google DeepMind is piloting the world's first double-blind AI evaluations, keeping external test prompts sealed in a cryptographically protected "box" so a frontier model can't peek at questions before it sits the exam. The mechanism targets benchma

The Frontier

XPeng ships VLA 630: on-device driving model up 3.5×, 3× faster

Chinese electric-vehicle maker XPeng is pushing the self-driving model arms race from the datacenter down into the car, announcing a major upgrade to its second-generation visual-language-action (VLA) model that runs entirely on-vehicle. XPeng unveiled the first big update to its second-generation VLA, version 6.3.0 — dubbed the "630" — with the on-device model tripling its parameters and its end-to-end response getting three times faster. According to the company, the in-car model now packs 3.

The Frontier

AI 101 — What is continual learning?

Continual learning — also called lifelong learning — is the ability of an AI model to keep learning new information over time without losing the skills it already has. Instead of being frozen after training, or wiped and rebuilt whenever the world changes, a continually learning model gets smarter as it goes, accumulating knowledge the way a person does. Most AI models today do the opposite. Once a model is trained, it is frozen: it never gets better from use, and updating it means going back i

See all