TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasks

We interact with computers on an everyday basis, be it in everyday life or work, and many aspects of work can be done entirely with access to a computer and the Internet. At the same time, thanks to improvements in large language models (LLMs), there has also been a rapid development in AI agents that interact with and affect change in their surrounding environments. But how performant are AI agents at helping to accelerate or even autonomously perform work-related tasks? The answer to this question has important implications for both industry looking to adopt AI into their workflows, and for economic policy to understand the effects that adoption of AI may have on the labor market. To measure the progress of these LLM agents' performance on performing real-world professional tasks, in this paper, we introduce TheAgentCompany, an extensible benchmark for evaluating AI agents that interact with the world in similar ways to those of a digital worker: by browsing the Web, writing code, running programs, and communicating with other coworkers. We build a self-contained environment with internal web sites and data that mimics a small software company environment, and create a variety of tasks that may be performed by workers in such a company. We test baseline agents powered by both closed API-based and open-weights language models (LMs), and find that with the most competitive agent, 24% of the tasks can be completed autonomously. This paints a nuanced picture on task automation with LM agents -- in a setting simulating a real workplace, a good portion of simpler tasks could be solved autonomously, but more difficult long-horizon tasks are still beyond the reach of current systems.

An Efficient Multicast Addressing Encoding Scheme for Multi-Core Neuromorphic Processors

Multi-core neuromorphic processors are becoming increasingly significant due to their energy-efficient local computing and scalable modular architecture, particularly for event-based processing applications. However, minimizing the cost of inter-core communication, which accounts for the majority of energy usage, remains a challenging issue. Beyond optimizing circuit design at lower abstraction levels, an efficient multicast addressing scheme is crucial. We propose a hierarchical bit string encoding scheme that largely expands the addressing capability of state-of-the-art symbol-based schemes for the same number of routing bits. When put at work with a real neuromorphic task, this hierarchical bit string encoding achieves a reduction in area cost by approximately 29% and decreases energy consumption by about 50%.

AI-LieDar: Examine the Trade-off Between Utility and Truthfulness in LLM Agents

To be safely and successfully deployed, LLMs must simultaneously satisfy truthfulness and utility goals. Yet, often these two goals compete (e.g., an AI agent assisting a used car salesman selling a car with flaws), partly due to ambiguous or misleading user instructions. We propose AI-LieDar, a framework to study how LLM-based agents navigate scenarios with utility-truthfulness conflicts in a multi-turn interactive setting. We design a set of realistic scenarios where language agents are instructed to achieve goals that are in conflict with being truthful during a multi-turn conversation with simulated human agents. To evaluate the truthfulness at large scale, we develop a truthfulness detector inspired by psychological literature to assess the agents' responses. Our experiment demonstrates that all models are truthful less than 50% of the time, although truthfulness and goal achievement (utility) rates vary across models. We further test the steerability of LLMs towards truthfulness, finding that models follow malicious instructions to deceive, and even truth-steered models can still lie. These findings reveal the complex nature of truthfulness in LLMs and underscore the importance of further research to ensure the safe and reliable deployment of LLMs and AI agents.

Persistent de Rham-Hodge Laplacians in the Eulerian representation

Recently, topological data analysis (TDA) has become a trending topic in data science and engineering. However, the key technique of TDA, i.e., persistent homology, is defined on point cloud data, which restricts its scope. In this work, we propose persistent de Rham-Hodge Laplacian, or persistent Hodge Laplacian (PHL) for abbreviation, for the TDA on manifolds with boundaries, or volumetric data. Specifically, we extended the evolutionary de Rham-Hodge theory from the Lagrangian formulation to the Eulerian formulation via structure-persevering Cartesian grids, and extended the persistent Laplacian on point clouds to persistent (de Rham-)Hodge Laplacian on nested families of manifolds with appropriate boundary conditions. The proposed PHL facilitates the machine learning and deep learning prediction of volumetric data. For a proof-of-principle application of the proposed PHL, we propose a persistent Hodge Laplacian learning (PHLL) algorithm for data on manifolds or volumetric data. To this end, we showcase the PHLL prediction of protein-ligand binding affinities in two benchmark datasets. Our numerical experiments highlight the power and promise of PHLL.

MSceneSpeech: A Multi-Scene Speech Dataset For Expressive Speech Synthesis

We introduce an open source high-quality Mandarin TTS dataset MSceneSpeech (Multiple Scene Speech Dataset), which is intended to provide resources for expressive speech synthesis. MSceneSpeech comprises numerous audio recordings and texts performed and recorded according to daily life scenarios. Each scenario includes multiple speakers and a diverse range of prosodic styles, making it suitable for speech synthesis that entails multi-speaker style and prosody modeling. We have established a robust baseline, through the prompting mechanism, that can effectively synthesize speech characterized by both user-specific timbre and scene-specific prosody with arbitrary text input. The open source MSceneSpeech Dataset and audio samples of our baseline are available at https://speechai-demo.github.io/MSceneSpeech/.

EchoSpike Predictive Plasticity: An Online Local Learning Rule for Spiking Neural Networks

The drive to develop artificial neural networks that efficiently utilize resources has generated significant interest in bio-inspired Spiking Neural Networks (SNNs). These networks are particularly attractive due to their potential in applications requiring low power and memory. This potential is further enhanced by the ability to perform online local learning, enabling them to adapt to dynamic environments. This requires the model to be adaptive in a self-supervised manner. While self-supervised learning has seen great success in many deep learning domains, its application for online local learning in multi-layer SNNs remains underexplored. In this paper, we introduce the "EchoSpike Predictive Plasticity" (ESPP) learning rule, a pioneering online local learning rule designed to leverage hierarchical temporal dynamics in SNNs through predictive and contrastive coding. We validate the effectiveness of this approach using benchmark datasets, demonstrating that it performs on par with current state-of-the-art supervised learning rules. The temporal and spatial locality of ESPP makes it particularly well-suited for low-cost neuromorphic processors, representing a significant advancement in developing biologically plausible self-supervised learning models for neuromorphic computing at the edge.

Is this the real life? Is this just fantasy? The Misleading Success of Simulating Social Interactions With LLMs

Recent advances in large language models (LLM) have enabled richer social simulations, allowing for the study of various social phenomena with LLM-based agents. However, most work has used an omniscient perspective on these simulations (e.g., single LLM to generate all interlocutors), which is fundamentally at odds with the non-omniscient, information asymmetric interactions that humans have. To examine these differences, we develop an evaluation framework to simulate social interactions with LLMs in various settings (omniscient, non-omniscient). Our experiments show that interlocutors simulated omnisciently are much more successful at accomplishing social goals compared to non-omniscient agents, despite the latter being the more realistic setting. Furthermore, we demonstrate that learning from omniscient simulations improves the apparent naturalness of interactions but scarcely enhances goal achievement in cooperative scenarios. Our findings indicate that addressing information asymmetry remains a fundamental challenge for LLM-based agents.

Core interface optimization for multi-core neuromorphic processors

Hardware implementations of Spiking Neural Networks (SNNs) represent a promising approach to edge-computing for applications that require low-power and low-latency, and which cannot resort to external cloud-based computing services. However, most solutions proposed so far either support only relatively small networks, or take up significant hardware resources, to implement large networks. To realize large-scale and scalable SNNs it is necessary to develop an efficient asynchronous communication and routing fabric that enables the design of multi-core architectures. In particular the core interface that manages inter-core spike communication is a crucial component as it represents the bottleneck of Power-Performance-Area (PPA) especially for the arbitration architecture and the routing memory. In this paper we present an arbitration mechanism with the corresponding asynchronous encoding pipeline circuits, based on hierarchical arbiter trees. The proposed scheme reduces the latency by more than 70% in sparse-event mode, compared to the state-of-the-art arbitration architectures, with lower area cost. The routing memory makes use of asynchronous Content Addressable Memory (CAM) with Current Sensing Completion Detection (CSCD), which saves approximately 46% energy, and achieves a 40% increase in throughput against conventional asynchronous CAM using configurable delay lines, at the cost of only a slight increase in area. In addition as it radically reduces the core interface resources in multi-core neuromorphic processors, the arbitration architecture and CAM architecture we propose can be also applied to a wide range of general asynchronous circuits and systems.

Uncovering and Categorizing Social Biases in Text-to-SQL

Content Warning: This work contains examples that potentially implicate stereotypes, associations, and other harms that could be offensive to individuals in certain social groups.} Large pre-trained language models are acknowledged to carry social biases towards different demographics, which can further amplify existing stereotypes in our society and cause even more harm. Text-to-SQL is an important task, models of which are mainly adopted by administrative industries, where unfair decisions may lead to catastrophic consequences. However, existing Text-to-SQL models are trained on clean, neutral datasets, such as Spider and WikiSQL. This, to some extent, cover up social bias in models under ideal conditions, which nevertheless may emerge in real application scenarios. In this work, we aim to uncover and categorize social biases in Text-to-SQL models. We summarize the categories of social biases that may occur in structured data for Text-to-SQL models. We build test benchmarks and reveal that models with similar task accuracy can contain social biases at very different rates. We show how to take advantage of our methodology to uncover and assess social biases in the downstream Text-to-SQL task. We will release our code and data.

Uncovering and Quantifying Social Biases in Code Generation

With the popularity of automatic code generation tools, such as Copilot, the study of the potential hazards of these tools is gaining importance. In this work, we explore the social bias problem in pre-trained code generation models. We propose a new paradigm to construct code prompts and successfully uncover social biases in code generation models. To quantify the severity of social biases in generated code, we develop a dataset along with three metrics to evaluate the overall social bias and fine-grained unfairness across different demographics. Experimental results on three pre-trained code generation models (Codex, InCoder, and CodeGen) with varying sizes, reveal severe social biases. Moreover, we conduct analysis to provide useful insights for further choice of code generation models with low social bias. (This work contains examples that potentially implicate stereotypes, associations, and other harms that could be offensive to individuals in certain social groups.)