Publisher
Hacker News — the tech community's aggregated front page of startup and hacker news.
134 recent stories from Hacker News
The AI-native product design tool. Start from a prompt, a photo, or a sketch, flesh the idea out by chat, and export manufacturable geometry straight into production. Browser-based, no install required.
Aug 11, 2026
The AI-native product design tool. Start from a prompt, a photo, or a sketch, flesh the idea out by chat, and export manufacturable geometry straight into production. Browser-based, no install required.
We investigate whether large language models can introspect on their internal states. It is difficult to answer this question through conversation alone, as genuine introspection cannot be distinguished from confabulations. Here, we address this challenge by injecting representations of known concepts into a model's activations, and measuring the influence of these manipulations on the model's self-reported states. We find that models can, in certain scenarios, notice the presence of injected concepts and accurately identify them. Models demonstrate some ability to recall prior internal representations and distinguish them from raw text inputs. Strikingly, we find that some models can use their ability to recall prior intentions in order to distinguish their own outputs from artificial prefills. In all these experiments, Claude Opus 4 and 4.1, the most capable models we tested, generally demonstrate the greatest introspective awareness; however, trends across models are complex and sensitive to post-training strategies. Finally, we explore whether models can explicitly control their internal representations, finding that models can modulate their activations when instructed or incentivized to "think about" a concept. Overall, our results indicate that current language models possess some functional introspective awareness of their own internal states. We stress that in today's models, this capacity is highly unreliable and context-dependent; however, it may continue to develop with further improvements to model capabilities.
Something happened at [BSidesSF](https://bsidessf.org/) 2026 that nobody saw coming. The top ten teams in the Capture The Flag competition didn't just use AI to...
Badger state? More like Surveillance State.
Compression and LLMs are trying to solve the exact same problem: predicting what comes next. Learn the fundamentals of compression and how better prediction leads to better shrinkage.
The new lightweight open model and routing library delivers greater control over AI, data and workflows across edge devices, PCs, workstations, data centers and the cloud.
Interest rates versus debt, consumer relief versus borrowing, and a $2.3 trillion bet on the future: the painful tradeoffs facing Japan's economy.
A web game and interactive explorer from Paradigm. Bootstrap an AI lab from scratch and see how compute, data, and labor shape the trajectory of AI progress.
