Agent-Based Models (ABMs) are powerful tools for studying emergent properties in complex systems. In ABMs, agent behaviors are governed by local interactions and stochastic rules. However, these rules are, in general, non-differentiable, limiting the use of gradient-based methods for optimization, and thus integration with real-world data. We propose a novel framework to learn a differentiable surrogate of any ABM by observing its generated data. Our method combines diffusion models to capture behavioral stochasticity and graph neural networks to model agent interactions. Distinct from prior surrogate approaches, our method introduces a fundamental shift: rather than approximating system-level outputs, it models individual agent behavior directly, preserving the decentralized, bottom-up dynamics that define ABMs. We validate our approach on two ABMs (Schelling's segregation model and a Predator-Prey ecosystem) showing that it replicates individual-level patterns and accurately forecasts emergent dynamics beyond training. Our results demonstrate the potential of combining diffusion models and graph learning for data-driven ABM simulation.

Reasoning about actions and change (RAC) has historically driven the development of many early AI challenges, such as the frame problem, and many AI disciplines, including non-monotonic and commonsense reasoning. The role of RAC remains important even now, particularly for tasks involving dynamic environments, interactive scenarios, and commonsense reasoning. Despite the progress of Large Language Models (LLMs) in various AI domains, their performance on RAC is underexplored. To address this gap, we introduce a new benchmark, ActionReasoningBench, encompassing 13 domains and rigorously evaluating LLMs across eight different areas of RAC. These include - Object Tracking, Fluent Tracking, State Tracking, Action Executability, Effects of Actions, Numerical RAC, Hallucination Detection, and Composite Questions. Furthermore, we also investigate the indirect effect of actions due to ramification constraints for every domain. Finally, we evaluate our benchmark using open-sourced and commercial state-of-the-art LLMs, including GPT-4o, Gemini-1.0-Pro, Llama2-7b-chat, Llama2-13b-chat, Llama3-8b-instruct, Gemma-2b-instruct, and Gemma-7b-instruct. Our findings indicate that these models face significant challenges across all categories included in our benchmark.

It might sound like a scene ripped straight from The Matrix series, but the ramifications of Artificial Intelligence are troubling to Keanu Reeves. Speaking to Wired, the Canadian star aired his grievances with recent developments in technology, including ChatGPT and the Metaverse. At one point, Reeves asked his interviewer, Angela Watercutter, if she thought a bot could take her place and be conducting celebrity interviews in the future. When Watercutter said that she didn't think such a thing would happen in her lifetime, Reeves gave an ominous response. Looking his interviewer'dead in the eye', Reeves said: "Oh no, you should be worried about that happening next month."

As evidenced by the example of ChatGPT, artificial intelligence is advancing in unprecedented directions to solve exciting new problems. But, as AI is being pointed toward critical cybersecurity operations, do the gains outweigh the potential risks and concerns? "You should be worried," said Andy Thurai (pictured), vice president and principal analyst at Constellation Research Inc. "The problem people don't realize is that ChatGPT, being a new, shiny object, it's all the craze that's about. But the problem is that most of the content that's produced either by ChatGPT or others are assets with no warranties, accountability or whatsoever. If it is content, it's OK. But if it is something like code that you use, then it's mostly not."

Around 60 years ago, the U.S. Department of Justice Civil Rights Division was established for prohibiting discrimination based on protected attributes. Over these 60 years, they established a set of policies and guidelines to identify and penalize those who discriminate1. The widespread use of machine learning (ML) models in routine life has prompted researchers to begin studying the extent to which these models are discriminatory. However, some researcher are unaware that the legal system already has well established procedures for describing and proving discrimination in law. In this series of blog posts, we'll try to bridge this gap.

In this work, we study the ramifications of human label uncertainty (HLU). Our evaluation of existing uncertainty estimation algorithms, with the presence of HLU, indicates the limitations of existing uncertainty metrics and algorithms themselves in response to HLU. Meanwhile, we observe undue effects in predictive uncertainty and generalizability. To mitigate the undue effects, we introduce a novel natural scene statistics (NSS) based label dilution training scheme without requiring massive human labels. Specifically, we first select a subset of samples with low perceptual quality ranked by statistical regularities of images. We then assign separate labels to each sample in this subset to obtain a training set with diluted labels. Our experiments and analysis demonstrate that training with NSS-based label dilution alleviates the undue effects caused by HLU.

A chief AI officer is a senior executive in an organization whose responsibilities include overseeing or managing the development of artificial intelligence. Strategically, they play a crucial role in directing the AI strategy as a whole and ensuring that the organization's algorithms efficiently achieve their goals. In addition to this supervision and control, they should have in-depth expertise in artificial intelligence and its potential implications for business operations. This is a leadership position since it demands an in-depth understanding of artificial intelligence and its possible organizational implications. It is often held by someone with a degree in artificial intelligence (AI) or a related discipline, as well as senior-level management experience.

In this series of Tales from the Dark Architecture articles I will be discussing some of the more extreme deep cognitive Artificial Intelligence designs that we are exploring on the pathway to Superintelligence. "Reality is a perception of trust fabricated by the human mind" We have approached a threshold in the design of Superintelligence. The issue before us is the nature of reality. Currently we are building AI machines to reflect our own human reality but what if they perceive far more than humans? Does human reality limit a Superintelligence and should a Superintelligence be free to experience a reality we humans can only contemplate but never experience? The truth is that advanced Cognitive AI systems not only perceive more of the natural world but they have the capacity to render more of the cognitively perceptive world than we humans can.

What if I told you that companies that don't embrace Ethical AI are leaving significant amounts of "Money on the Table"; that they are not only missing out on potentially profitable customers, but that over time they are eroding their Total Addressable Market (TAM)? Do I have your attention now? After I published the blog "The Ethical AI Application Pyramid", a question from Karrie Sullivan coupled with a mentoring session with the startup unfog.ai "If your AI model doesn't take into consideration the ultimate outcomes of the AI model's False Negatives, then confirmation bias in the AI model could set in and eventually the company's Total Addressable Market (TAM) could shrink to a point where the business might no longer be viable." Yea, not only is Ethical AI the right thing to do from a cultural and society perspective, but there are direct bottom-line financial ramifications if your AI models are not learning and adapting from the AI model's False Negatives.

A neuron is a thing of beauty. Ever since Santiago Ramón y Cajal stained them with silver nitrate to make them visible under the microscopes of the 1880s (see drawing above), their ramifications have fired the scientific imagination. Ramón y Cajal called them the butterflies of the soul. Your browser does not support the audio element. Those ramifications--dendrites by the dozen to collect incoming signals, called action potentials, from other neurons, and a single axon to pass on the summed wisdom of those signals in the form of another action potential, turn neurons into parts of far bigger structures known as neural networks.