Large Language Models (LLMs) have become a pivotal research area, potentially making beneficial contributions in fields like healthcare where they can streamline automated billing and decision support. However, the frequent use of specialized coded languages like ICD-10, which are regularly updated and deviate from natural language formats, presents potential challenges for LLMs in creating accurate and meaningful latent representations. This raises concerns among healthcare professionals about potential inaccuracies or ``hallucinations" that could result in the direct impact of a patient. Therefore, this study evaluates whether large language models (LLMs) are aware of medical code ontologies and can accurately generate names from these codes. We assess the capabilities and limitations of both general and biomedical-specific generative models, such as GPT, LLaMA-2, and Meditron, focusing on their proficiency with domain-specific terminologies. While the results indicate that LLMs struggle with coded language, we offer insights on how to adapt these models to reason more effectively.

Eliciting reasoning capabilities from language models (LMs) is a critical direction on the path towards building intelligent systems. Most recent studies dedicated to reasoning focus on out-of-distribution performance on procedurally-generated synthetic benchmarks, bespoke-built to evaluate specific skills only. This trend makes results hard to transfer across publications, slowing down progress. Three years ago, a similar issue was identified and rectified in the field of neural algorithmic reasoning, with the advent of the CLRS benchmark. CLRS is a dataset generator comprising graph execution traces of classical algorithms from the Introduction to Algorithms textbook. Inspired by this, we propose CLRS-Text -- a textual version of these algorithmic traces. Out of the box, CLRS-Text is capable of procedurally generating trace data for thirty diverse, challenging algorithmic tasks across any desirable input distribution, while offering a standard pipeline in which any additional algorithmic tasks may be created in the benchmark. We fine-tune and evaluate various LMs as generalist executors on this benchmark, validating prior work and revealing a novel, interesting challenge for the LM reasoning community. Our code is available at https://github.com/google-deepmind/clrs/tree/master/clrs/_src/clrs_text.

Masked (or absorbing) diffusion is actively explored as an alternative to autoregressive models for generative modeling of discrete data. However, existing work in this area has been hindered by unnecessarily complex model formulations and unclear relationships between different perspectives, leading to suboptimal parameterization, training objectives, and ad hoc adjustments to counteract these issues. In this work, we aim to provide a simple and general framework that unlocks the full potential of masked diffusion models. We show that the continuous-time variational objective of masked diffusion models is a simple weighted integral of cross-entropy losses. Our framework also enables training generalized masked diffusion models with state-dependent masking schedules. When evaluated by perplexity, our models trained on OpenWebText surpass prior diffusion language models at GPT-2 scale and demonstrate superior performance on 4 out of 5 zero-shot language modeling tasks. Furthermore, our models vastly outperform previous discrete diffusion models on pixel-level image modeling, achieving 2.78~(CIFAR-10) and 3.42 (ImageNet 64$\times$64) bits per dimension that are comparable or better than autoregressive models of similar sizes.

Investors led by Microsoft pressured OpenAI to reinstate Altman, which it did within days, alongside vague promises to be more responsible. Then, last month, the company disbanded the internal group tasked with safety research, known as the "superalignment team." Some of the team's most prominent members publicly resigned, including its head, Jan Leike, who posted on X that "over the past years, safety culture and processes have taken a backseat to shiny products." Fortune reported that OpenAI did not provide anywhere near the resources it had initially, publicly promised for safety research. Saunders, who also worked on superalignment, said he resigned when he "lost hope a few months before Jan did."

Nvidia's market value has surged past 3tn ( 2.3tn), lifting the chipmaker ahead of Apple to become the second most valuable publicly listed company in the world. The firm's share price rose more than 5% on Wednesday, to more than 1,224. It extended a breathtakingly rapid climb that started last year, powered by bets that the firm is positioned to be a major winner from a wave of investment in artificial intelligence (AI). Its market value now sits just behind Microsoft, another key player in the industry thanks to its investments in Chat GPT-maker OpenAI. Valued at "just" 2tn as recently as February, Nvidia sparked a new wave of share purchases after it announced plans last month for a so-called stock split.

This could be a costly interview for William Saunders. The former safety researcher resigned from OpenAI in February, and--like many other departing employees--signed a non-disparagement agreement in order to keep the right to sell his equity in the company. Although he says OpenAI has since told him that it does not intend to enforce the agreement, and has made similar public commitments, he is still taking a risk by speaking out. "By speaking to you I might never be able to access vested equity worth millions of dollars," he tells TIME. "But I think it's more important to have a public dialogue about what is happening at these AGI companies."

Designed and manufactured in France, the Thalamus is a groundbreaking autonomous security robot that is redefining the way we think about surveillance and protection.

Zoom users in the not-too-distant future could send AI avatars to attend meetings in their absence, the company's chief executive has suggested, delegating the drudge-work of corporate life to a system trained on their own content. Such a system would be "five or six years" away, Eric Yuan told The Verge magazine, but he added that the company was working on nearer-term technologies that could bring it closer to reality. "Let's assume, fast-forward five or six years, that AI is ready," Yuan said. "AI probably can help for maybe 90% of the work, but in terms of real-time interaction, today, you and I are talking online. So, I can send my digital version, you can send your digital version."

A group of OpenAI insiders is blowing the whistle on what they say is a culture of recklessness and secrecy at the San Francisco artificial intelligence company, which is racing to build the most powerful AI systems ever created. The group, which includes nine current and former OpenAI employees, has rallied in recent days around shared concerns that the company has not done enough to prevent its AI systems from becoming dangerous. The members say OpenAI, which started as a nonprofit research lab and burst into public view with the 2022 release of ChatGPT, is putting a priority on profits and growth as it tries to build artificial general intelligence, or AGI, the industry term for a computer program capable of doing anything a human can.

How to accurately and efficiently assess AI-generated images (AIGIs) remains a critical challenge for generative models. Given the high costs and extensive time commitments required for user studies, many researchers have turned towards employing large multi-modal models (LMMs) as AIGI evaluators, the precision and validity of which are still questionable. Furthermore, traditional benchmarks often utilize mostly natural-captured content rather than AIGIs to test the abilities of LMMs, leading to a noticeable gap for AIGIs. Therefore, we introduce A-Bench in this paper, a benchmark designed to diagnose whether LMMs are masters at evaluating AIGIs. Specifically, A-Bench is organized under two key principles: 1) Emphasizing both high-level semantic understanding and low-level visual quality perception to address the intricate demands of AIGIs. 2) Various generative models are utilized for AIGI creation, and various LMMs are employed for evaluation, which ensures a comprehensive validation scope. Ultimately, 2,864 AIGIs from 16 text-to-image models are sampled, each paired with question-answers annotated by human experts, and tested across 18 leading LMMs. We hope that A-Bench will significantly enhance the evaluation process and promote the generation quality for AIGIs. The benchmark is available at https://github.com/Q-Future/A-Bench.