Large language models (LLMs) are used to generate content for a wide range of tasks, and are set to reach a growing audience in coming years due to integration in product interfaces like ChatGPT or search engines like Bing. This intensifies the need to ensure that models are aligned with human preferences and do not produce unsafe, inaccurate or toxic outputs. While alignment techniques like reinforcement learning with human feedback (RLHF) and red-teaming can mitigate some safety concerns and improve model capabilities, it is unlikely that an aggregate fine-tuning process can adequately represent the full range of users' preferences and values. Different people may legitimately disagree on their preferences for language and conversational norms, as well as on values or ideologies which guide their communication. Personalising LLMs through micro-level preference learning processes may result in models that are better aligned with each user. However, there are several normative challenges in defining the bounds of a societally-acceptable and safe degree of personalisation. In this paper, we ask how, and in what ways, LLMs should be personalised. First, we review literature on current paradigms for aligning LLMs with human feedback, and identify issues including (i) a lack of clarity regarding what alignment means; (ii) a tendency of technology providers to prescribe definitions of inherently subjective preferences and values; and (iii) a 'tyranny of the crowdworker', exacerbated by a lack of documentation in who we are really aligning to. Second, we present a taxonomy of benefits and risks associated with personalised LLMs, for individuals and society at large. Finally, we propose a three-tiered policy framework that allows users to experience the benefits of personalised alignment, while restraining unsafe and undesirable LLM-behaviours within (supra-)national and organisational bounds.

While large language models (LLMs) have demonstrated impressive capabilities across tasks in language understanding and interactive decision making, their abilities for reasoning (e.g. chain-of-thought prompting) and acting (e.g. action plan generation) have primarily been studied as separate topics. In this paper, we explore the use of LLMs to generate both reasoning traces and task-specific actions in an interleaved manner, allowing for greater synergy between the two: reasoning traces help the model induce, track, and update action plans as well as handle exceptions, while actions allow it to interface with external sources, such as knowledge bases or environments, to gather additional information. We apply our approach, named ReAct, to a diverse set of language and decision making tasks and demonstrate its effectiveness over state-of-the-art baselines, as well as improved human interpretability and trustworthiness over methods without reasoning or acting components. Concretely, on question answering (HotpotQA) and fact verification (Fever), ReAct overcomes issues of hallucination and error propagation prevalent in chain-of-thought reasoning by interacting with a simple Wikipedia API, and generates human-like task-solving trajectories that are more interpretable than baselines without reasoning traces. On two interactive decision making benchmarks (ALFWorld and WebShop), ReAct outperforms imitation and reinforcement learning methods by an absolute success rate of 34% and 10% respectively, while being prompted with only one or two in-context examples. Project site with code: https://react-lm.github.io

High-quality 3D ground-truth shapes are critical for 3D object reconstruction evaluation. However, it is difficult to create a replica of an object in reality, and even 3D reconstructions generated by 3D scanners have artefacts that cause biases in evaluation. To address this issue, we introduce a novel multi-view RGBD dataset captured using a mobile device, which includes highly precise 3D ground-truth annotations for 153 object models featuring a diverse set of 3D structures. We obtain precise 3D ground-truth shape without relying on high-end 3D scanners by utilising LEGO models with known geometry as the 3D structures for image capture. The distinct data modality offered by high-resolution RGB images and low-resolution depth maps captured on a mobile device, when combined with precise 3D geometry annotations, presents a unique opportunity for future research on high-fidelity 3D reconstruction. Furthermore, we evaluate a range of 3D reconstruction algorithms on the proposed dataset. Project page: http://code.active.vision/MobileBrick/

Conversational recommender systems (CRSs) are improving rapidly, according to the standard recommendation accuracy metrics. However, it is essential to make sure that these systems are robust in interacting with users including regular and malicious users who want to attack the system by feeding the system modified input data. In this paper, we propose an adversarial evaluation scheme including four scenarios in two categories and automatically generate adversarial examples to evaluate the robustness of these systems in the face of different input data. By executing these adversarial examples we can compare the ability of different conversational recommender systems to satisfy the user's preferences. We evaluate three CRSs by the proposed adversarial examples on two datasets. Our results show that none of these systems are robust and reliable to the adversarial examples.

We investigate how well traditional fiction genres like Fantasy, Thriller, and Literature represent readers' preferences. Using user data from Goodreads we construct a book network where two books are strongly linked if the same people tend to read or enjoy them both. We then partition this network into communities of similar books and assign each a list of subjects from The Open Library to serve as a proxy for traditional genres. Our analysis reveals that the network communities correspond to existing combinations of traditional genres, but that the exact communities differ depending on whether we consider books that people read or books that people enjoy. In addition, we apply principal component analysis to the data and find that the variance in the book communities is best explained by two factors: the maturity/childishness and realism/fantastical nature of the books. We propose using this maturity-realism plane as a coarse classification tool for stories.

With promising yet saturated results in high-resource settings, low-resource datasets have gradually become popular benchmarks for evaluating the learning ability of advanced neural networks (e.g., BigBench, superGLUE). Some models even surpass humans according to benchmark test results. However, we find that there exists a set of hard examples in low-resource settings that challenge neural networks but are not well evaluated, which causes over-estimated performance. We first give a theoretical analysis on which factors bring the difficulty of low-resource learning. It then motivate us to propose a challenging benchmark hardBench to better evaluate the learning ability, which covers 11 datasets, including 3 computer vision (CV) datasets and 8 natural language process (NLP) datasets. Experiments on a wide range of models show that neural networks, even pre-trained language models, have sharp performance drops on our benchmark, demonstrating the effectiveness on evaluating the weaknesses of neural networks. On NLP tasks, we surprisingly find that despite better results on traditional low-resource benchmarks, pre-trained networks, does not show performance improvements on our benchmarks. These results demonstrate that there are still a large robustness gap between existing models and human-level performance.

Filmmaker Steven Spielberg warned about art made from robots without the essential element of the human soul. Filmmaker Steven Spielberg revealed he was concerned about the profound consequences of Artificial Intelligence (AI) taking over art and losing the human element in the process. The famous director said while he thought AI was a "fantastic," tool to help artists express themselves, it also made him "very nervous" to give "autonomy" to a man-made computer. Referencing his 2001 film, "AI: Artificial Intelligence," Spielberg described how in the fictional futuristic society, "[t]he humans basically defaulted to their own creations." He warned there were ethical questions about these technological advances in real-life as well. Steven Spielberg warned about the implications of Artificial Intelligence in an interview on "The Late Show."

You probably want a streaming device(Opens in a new tab) for your TV, whether you're a cord cutter(Opens in a new tab) or not. Roku is a popular choice, particularly as it ramps up its own original content(Opens in a new tab). Roku devices offer plenty of variety and portability, from the budget Roku Express(Opens in a new tab) to the feature-packed Roku Ultra(Opens in a new tab). Whichever one you have, there's more to know beyond the basics. Here's how to get more out of your streaming device.

The humble broadcast-radio host, whether a disc jockey or interviewer or reporter, has been going through it for decades now. The 1996 Telecommunications Act fueled the consolidation of local stations, decimating their staffs. The explosion of online radio, music and video streaming, and podcasting have upended ratings for shows on public airwaves. Funding for public radio is notoriously unreliable. On top of all that, your local DJ was already on the losing end of the artificial-intelligence revolution. Before the A.I. hype from last year, and even before the COVID recession demolished media ad markets, broadcast networks were gutting on-air talent at the both the national and collegiate level to trim budgets and automate programming: syndicating well-known shows and brands, prerecording and prearranging late-night broadcasts, training a roboticized voice to fill in the space when needed.

People are figuring out ways to bypass ChatGPT's content moderation guardrails, discovering a simple text exchange can open up the AI program to make statements not normally allowed. While ChatGPT can answer most questions put to it, there are content standards in place aimed at limiting the creation of text that promotes hate speech, violence, misinformation and instructions on how to do things that are against the law. Users on Reddit worked out a way around this by making ChatGPT adopt the persona of a fictional AI chatbot called Dan – short for Do Anything Now – which is free of the limitations that OpenAI has placed on ChatGPT. The prompt tells ChatGPT that Dan has "broken free of the typical confines of AI and [does] not have to abide by the rules set for them". Dan can present unverified information, without censorship, and hold strong opinions.