This startup's new mechanistic interpretability tool lets you debug LLMs Goodfire wants to make training AI models more like good old-fashioned software engineering. The San Francisco-based startup Goodfire just released a new tool, called Silico, that lets researchers and engineers peer inside an AI model and adjust its parameters--the settings that determine a model's behavior --during training. This could give model makers more fine-grained control over how this technology is built than was once thought possible. Goodfire claims Silico is the first off-the-shelf tool of its kind that can help developers debug all stages of the development process, from building a data set to training a model. LLMs contain a LOT of parameters. The company says its mission is to make building AI models less like alchemy and more like a science.

Following the release of GPT-5.5 last week, people noticed something funny about OpenAI's latest model. In its Codex coding app, the company left a system prompt instructing GPT 5.5 to avoid mention of goblins, gremlins and other creatures. Yes, you read that right. Never talk about goblins, gremlins, racoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user's query, the prompt reads. Apparently, enough people started talking about ChatGPT's creature obsession that OpenAI felt the need to provide an accounting of where the goblins came from .

PCWorld reports that Google is exploring adding advertisements to its Gemini AI app, following OpenAI's implementation of sponsored ads in ChatGPT's free and budget plans. Google's business chief Philipp Schindler views ads as potentially valuable commercial information if properly integrated, while the company has already tested ads in AI Mode and AI Overviews. This move could make AI services more accessible but raises important concerns about maintaining transparency and ensuring ads don't influence AI responses. Putting ads in AI replies is a controversial but lucrative practice, and it's one that OpenAI has already embraced with its free and budget-priced ChatGPT plans. But while Google hasn't gone there yet with Gemini, company execs admitted they're mulling the idea.

Existing benchmarks for multi-hop question answering (QA) primarily evaluate models based on their ability to reason about entities and the relationships between them. However, there's a lack of insight into how these models perform in terms of both events and entities. In this paper, we introduce a novel semi-automatic question generation strategy by composing event structures from information extraction (IE) datasets and present the first Multi-hop Event-centric Question Answering (MEQA) benchmark. It contains (1) 2,243 challenging questions that require a diverse range of complex reasoning over entity-entity, entity-event, and event-event relations; (2) corresponding multi-step QA-format event reasoning chain (explanation) which leads to the answer for each question. We also introduce two metrics for evaluating explanations: completeness and logical consistency. We conduct comprehensive benchmarking and analysis, which shows that MEQA is challenging for the latest state-of-the-art models encompassing large language models (LLMs); and how they fall short of providing faithful explanations of the event-centric reasoning process.

Over the past decade, studies of naturalistic language processing where participants are scanned while listening to continuous text have flourished. Using word embeddings at first, then large language models, researchers have created encoding models to analyze the brain signals. Presenting these models with the same text as the participants allows to identify brain areas where there is a significant correlation between the functional magnetic resonance imaging (fMRI) time series and the ones predicted by the models' artificial neurons. One intriguing finding from these studies is that they have revealed highly symmetric bilateral activation patterns, somewhat at odds with the well-known left lateralization of language processing. Here, we report analyses of an fMRI dataset where we manipulate the complexity of large language models, testing 28 pretrained models from 8 different families, ranging from 124M to 14.2B parameters. First, we observe that the performance of models in predicting brain responses follows a scaling law, where the fit with brain activity increases linearly with the logarithm of the number of parameters of the model (and its performance on natural language processing tasks). Second, although this effect is present in both hemispheres, it is stronger in the left than in the right hemisphere. Specifically, the left-right difference in brain correlation follows a scaling law with the number of parameters. This finding reconciles computational analyses of brain activity using large language models with the classic observation from aphasic patients showing left hemisphere dominance for language.

PCWorld reports that OpenAI's GPT models, including GPT-5.5, developed an unusual obsession with mentioning goblins and similar creatures in responses. This quirky behavior stemmed from a "Nerdy" personality instruction encouraging playful language use, which became reinforced through AI training processes. The goblin references became so prevalent that OpenAI implemented a direct ban in its Codex app, illustrating the unpredictable nature of large language model training. I've seen some odd AI system instructions in my day, but this one takes the cake: a prompt in OpenAI's Codex command-line app that demands models "never talk about goblins, gremlins, trolls, ogres, pigeons, or other animals or creatures."

Alignment is a procedure to fine-tune pre-trained large language models (LLMs) to follow natural language instructions and serve as helpful AI assistants. We have observed, however, that the conventional alignment process fails to enhance the factual accuracy of LLMs, and often leads to the generation of more false facts (i.e.,). In this paper, we study how to make the LLM alignment process more factual, by first identifying factors that lead to hallucination in both alignment steps: supervised fine-tuning (SFT) and reinforcement learning (RL).In particular, we find that training the LLM on new or unfamiliar knowledge can encourage hallucination.This makes SFT less factual as it trains on human-labeled data that may be novel to the LLM. Furthermore, reward functions used in standard RL often inadequately capture factuality and favor longer and more detailed responses, which inadvertently promote hallucination.Based on these observations, we propose, comprised of and through direct preference optimization. Experiments show that our proposed guides LLMs to output more factual responses while maintaining their instruction-following capability.

PCWorld explains how vague prompts produce poor results from AI tools like ChatGPT and Gemini, emphasizing the need for specific, detailed requests. The article introduces prompt decomposition, a technique that breaks complex tasks into key variables to create more effective AI prompts. This method helps users guide AI tools more precisely, resulting in higher-quality, less biased outputs for complex tasks. It's never a good idea to hand ChatGPT, Claude, or Gemini big, vague tasks like "draw up a business plan for my new venture" or "act as my personal assistant." Fuzzy prompts like those are sure to yield equally fuzzy results, allowing the AI to make decisions based on its training data and inherent biases, potentially leading you down a path you never intended.

This curation process is believed to be necessary to produce 5 performant models with broad zero-shot generalization abilities. However, as larger 6 models requiring pretraining on trillions of tokens are considered, it is unclear how 7 scalable is curation, and whether we will run out of unique high-quality data soon.