In the year since OpenAI released ChatGPT, high school teacher Vicki Davis has been rethinking every single assignment she gives her students. Davis, a computer science teacher at Sherwood Christian Academy in Georgia, was well-positioned to be an early adopter of the technology. She's also the IT director at the school and helped put together an AI policy in March: the school opted to allow the use of AI tools for specific projects so long as students discussed it with their teachers and cited the tool. In Davis' mind, there were good and bad uses of AI, and ignoring its growing popularity was not going to help students unlock the productive uses or understand its dangers. "It's actually changed how I design my projects because there are some times I want my students to use AI, and then there are times I don't want them to," Davis said.

The UK's AI safety summit opens at Bletchley Park this week, and is the passion project of Rishi Sunak: a prime minister desperate for a good news story as his government looks down the barrel of a crushing election defeat. Sunak appears to want progress on AI to become his lasting legacy. Last week, he delivered a speech about the risks of AI if weaponised by terrorists and cybercriminals, and published a series of documents on "frontier AI", an industry term for generative AI tools such as ChatGPT and DALL-E. He even unveiled a UK AI safety institute. The slick – albeit very behind in the polls – Stanford MBA grad who likes to holiday in California had, to use a favoured phrase of his, "got to grips" with the problem.

Model pruning is a popular approach to enable the deployment of large deep learning models on edge devices with restricted computational or storage capacities. Although sparse models achieve performance comparable to that of their dense counterparts at the level of the entire dataset, they exhibit high accuracy drops for some data sub-groups. Existing methods to mitigate this disparate impact induced by pruning (i) rely on surrogate metrics that address the problem indirectly and have limited interpretability; or (ii) scale poorly with the number of protected sub-groups in terms of computational cost. We propose a constrained optimization approach that directly addresses the disparate impact of pruning: our formulation bounds the accuracy change between the dense and sparse models, for each subgroup. This choice of constraints provides an interpretable success criterion to determine if a pruned model achieves acceptable disparity levels. Experimental results demonstrate that our technique scales reliably to problems involving large models and hundreds of protected sub-groups. Current deep learning practice displays a trend towards larger architectures (Bommasani et al., 2021), as exemplified by popular models such as GPT-4 (OpenAI, 2023), Llama 2 (Touvron et al., 2023) and DALL-E 2 (Ramesh et al., 2022). Model compression techniques such as pruning (Gale et al., 2019), knowledge distillation (Hinton et al., 2015), or quantization (Gholami et al., 2021) are crucial towards enabling the deployment of large models across a wide range of platforms, including resource-constrained edge devices like smartphones. Despite achieving comparable performance at an aggregate level over the entire dataset, pruned models often exhibit significant accuracy reduction for some data sub-groups (Hooker et al., 2019; 2020; Paganini, 2020). In particular, under-represented groups can suffer high performance degradation while the overall performance remains unaffected, thus exacerbating systemic biases in machine learning models. Tran et al. (2022) refer to this phenomenon as the disparate impact of pruning. Existing mitigation methods face challenges in terms of interpretability and scalability to a large number of sub-groups. Tran et al. (2022) introduce constraints aiming to equalize the loss of the sparse model across sub-groups. However, their approach does not account for the unequal grouplevel performance of the dense model. Moreover, while the loss can be a useful surrogate for training, this method addresses the disparate impact issue indirectly as it focuses on controlling the loss, rather than group-level changes in accuracy. Alternatively, Lin et al. (2022) compute per-group importance scores for every model parameter to determine the weights to be pruned. This approach becomes prohibitively expensive when the model or the number of sub-groups is large.

The recent wave of generative AI has sparked unprecedented global attention, with both excitement and concern over potentially superhuman levels of artificial intelligence: models now take only seconds to produce outputs that would challenge or exceed the capabilities even of expert humans. At the same time, models still show basic errors in understanding that would not be expected even in non-expert humans. This presents us with an apparent paradox: how do we reconcile seemingly superhuman capabilities with the persistence of errors that few humans would make? In this work, we posit that this tension reflects a divergence in the configuration of intelligence in today's generative models relative to intelligence in humans. Specifically, we propose and test the Generative AI Paradox hypothesis: generative models, having been trained directly to reproduce expert-like outputs, acquire generative capabilities that are not contingent upon -- and can therefore exceed -- their ability to understand those same types of outputs. This contrasts with humans, for whom basic understanding almost always precedes the ability to generate expert-level outputs. We test this hypothesis through controlled experiments analyzing generation vs. understanding in generative models, across both language and image modalities. Our results show that although models can outperform humans in generation, they consistently fall short of human capabilities in measures of understanding, as well as weaker correlation between generation and understanding performance, and more brittleness to adversarial inputs. Our findings support the hypothesis that models' generative capability may not be contingent upon understanding capability, and call for caution in interpreting artificial intelligence by analogy to human intelligence.

Recent progress in text-to-image (TTI) systems, such as StableDiffusion, Imagen, and DALL-E 2, have made it possible to create realistic images with simple text prompts. It is tempting to use these systems to eliminate the manual task of obtaining natural images for training a new machine learning classifier. However, in all of the experiments performed to date, classifiers trained solely with synthetic images perform poorly at inference, despite the images used for training appearing realistic. Examining this apparent incongruity in detail gives insight into the limitations of the underlying image generation processes. Through the lens of diversity in image creation vs.accuracy of what is created, we dissect the differences in semantic mismatches in what is modeled in synthetic vs. natural images. This will elucidate the roles of the image-languag emodel, CLIP, and the image generation model, diffusion. We find four issues that limit the usefulness of TTI systems for this task: ambiguity, adherence to prompt, lack of diversity, and inability to represent the underlying concept. We further present surprising insights into the geometry of CLIP embeddings.

Generative AI systems produce a range of risks. To ensure the safety of generative AI systems, these risks must be evaluated. In this paper, we make two main contributions toward establishing such evaluations. First, we propose a three-layered framework that takes a structured, sociotechnical approach to evaluating these risks. This framework encompasses capability evaluations, which are the main current approach to safety evaluation. It then reaches further by building on system safety principles, particularly the insight that context determines whether a given capability may cause harm. To account for relevant context, our framework adds human interaction and systemic impacts as additional layers of evaluation. Second, we survey the current state of safety evaluation of generative AI systems and create a repository of existing evaluations. Three salient evaluation gaps emerge from this analysis. We propose ways forward to closing these gaps, outlining practical steps as well as roles and responsibilities for different actors. Sociotechnical safety evaluation is a tractable approach to the robust and comprehensive safety evaluation of generative AI systems.

Earlier today, President Joe Biden signed the most sweeping set of regulatory principles on artificial intelligence in America to date: a lengthy executive order that directs all types of government agencies to make sure America is leading the way in developing the technology while also addressing the many dangers that it poses. The order explicitly pushes agencies to establish rules and guidelines, write reports, and create funding and research initiatives for AI--"the most consequential technology of our time," in the president's own words. The scope of the order is impressive, especially given that the generative-AI boom began just about a year ago. But the document's many parts--and there are many--are at times in tension, revealing a broader confusion over what, exactly, America's primary attitude toward AI should be: Is it a threat to national security, or a just society? Is it a geopolitical weapon?

This article is from Big Technology, a newsletter by Alex Kantrowitz. Sam Altman sat comfortably between Satya Nadella and Sundar Pichai at a White House gathering of top A.I. CEOs in May--with one noticeable gap in the guest list. With Alphabet, Microsoft, and OpenAI in attendance, it was impossible to miss Mark Zuckerberg's absence. And that appeared to be no accident. The meeting, one administration official said, "was focused on companies currently leading in the space."

In the weeks since Hamas launched its October 7 surprise attack on Israel, the ensuing conflict has generated an unprecedented wave of disinformation, an "algorithmically driven fog of war" that has tripped up major new organizations and left social media companies floundering. Yet, amid all of the deceptive images and video moving around on social media, the content generated by artificial intelligence tools has remained relatively peripheral. Even as some wondered if the Israel-Hamas war would be the first conflict dominated by false generative AI images, the technology has had a more complex and subtle impact. "There are definitely AI images circulating but not to the degree where I think it's playing a central role in the spread of information," says Layla Mashkoor, an associate editor at the Atlantic Council's Digital Forensic Research Lab, which studies online disinformation. Primarily, Mashkoor says, AI-generated disinformation is being used by activists to solicit support--or give the impression of wider support--for a particular side.

Transformer neural networks show promising capabilities, in particular for uses in materials analysis, design and manufacturing, including their capacity to work effectively with both human language, symbols, code, and numerical data. Here we explore the use of large language models (LLMs) as a tool that can support engineering analysis of materials, applied to retrieving key information about subject areas, developing research hypotheses, discovery of mechanistic relationships across disparate areas of knowledge, and writing and executing simulation codes for active knowledge generation based on physical ground truths. When used as sets of AI agents with specific features, capabilities, and instructions, LLMs can provide powerful problem solution strategies for applications in analysis and design problems. Our experiments focus on using a fine-tuned model, MechGPT, developed based on training data in the mechanics of materials domain. We first affirm how finetuning endows LLMs with reasonable understanding of domain knowledge. However, when queried outside the context of learned matter, LLMs can have difficulty to recall correct information. We show how this can be addressed using retrieval-augmented Ontological Knowledge Graph strategies that discern how the model understands what concepts are important and how they are related. Illustrated for a use case of relating distinct areas of knowledge - here, music and proteins - such strategies can also provide an interpretable graph structure with rich information at the node, edge and subgraph level. We discuss nonlinear sampling strategies and agent-based modeling applied to complex question answering, code generation and execution in the context of automated force field development from actively learned Density Functional Theory (DFT) modeling, and data analysis.