The birth of ChatGPT, a cutting-edge language model-based chatbot developed by OpenAI, ushered in a new era in AI. However, due to potential pitfalls, its role in rigorous scientific research is not clear yet. This paper vividly showcases its innovative application within the field of drug discovery. Focused specifically on developing anti-cocaine addiction drugs, the study employs GPT-4 as a virtual guide, offering strategic and methodological insights to researchers working on generative models for drug candidates. The primary objective is to generate optimal drug-like molecules with desired properties. By leveraging the capabilities of ChatGPT, the study introduces a novel approach to the drug discovery process. This symbiotic partnership between AI and researchers transforms how drug development is approached. Chatbots become facilitators, steering researchers towards innovative methodologies and productive paths for creating effective drug candidates. This research sheds light on the collaborative synergy between human expertise and AI assistance, wherein ChatGPT's cognitive abilities enhance the design and development of potential pharmaceutical solutions. This paper not only explores the integration of advanced AI in drug discovery but also reimagines the landscape by advocating for AI-powered chatbots as trailblazers in revolutionizing therapeutic innovation.

We introduce Inference-Time Intervention (ITI), a technique designed to enhance the "truthfulness" of large language models (LLMs). ITI operates by shifting model activations during inference, following a set of directions across a limited number of attention heads. This intervention significantly improves the performance of LLaMA models on the TruthfulQA benchmark. On an instruction-finetuned LLaMA called Alpaca, ITI improves its truthfulness from 32.5% to 65.1%. We identify a trade-off between truthfulness and helpfulness and demonstrate how to balance it by tuning the intervention strength. ITI is minimally invasive and computationally inexpensive. Moreover, the technique is data efficient: while approaches like RLHF require extensive annotations, ITI locates truthful directions using only few hundred examples. Our findings suggest that LLMs may have an internal representation of the likelihood of something being true, even as they produce falsehoods on the surface.

Planning for goal-oriented dialogue often requires simulating future dialogue interactions and estimating task progress. Many approaches thus consider training neural networks to perform look-ahead search algorithms such as A* search and Monte Carlo Tree Search (MCTS). However, this training often requires abundant annotated data, which creates challenges when faced with noisy annotations or low-resource settings. We introduce GDP-Zero, an approach using Open-Loop MCTS to perform goal-oriented dialogue policy planning without any model training. GDP-Zero prompts a large language model to act as a policy prior, value function, user simulator, and system model during the tree search. We evaluate GDP-Zero on the goal-oriented task PersuasionForGood, and find that its responses are preferred over ChatGPT up to 59.32% of the time, and are rated more persuasive than ChatGPT during interactive evaluations.

This article was syndicated from the Bulletin of the Atomic Scientists, which has covered human-made threats to humanity since 1945. The main artery in Montreal's Little Italy is lined with cafés, wine bars, and pastry shops that spill into tree-lined, residential side streets. Generations of farmers, butchers, bakers, and fishmongers sell farm-to-table goods in the neighborhood's large, open-air market, the Marché Jean-Talon. But the quiet enclave also accommodates a modern, 90,000-square-foot global AI hub known as Mila–Quebec AI Institute. Mila claims to house the largest concentration of deep learning academic researchers in the world, including more than 1,000 researchers and more than 100 professors who work with more than 100 industry partners from around the globe.

This paper is devoted to the development of a localized Large Language Model (LLM) specifically for Arabic, a language imbued with unique cultural characteristics inadequately addressed by current mainstream models. Significant concerns emerge when addressing cultural sensitivity and local values. To address this, the paper proposes a comprehensive solution that includes further pre-training with Arabic texts, Supervised Fine-Tuning (SFT) utilizing native Arabic instructions, and GPT-4 responses in Arabic, alongside Reinforcement Learning with AI Feedback (RLAIF) employing a reward model attuned to local culture and values. The goal is to cultivate culturally cognizant and value-aligned Arabic LLMs capable of accommodating the diverse, application-specific needs of Arabic-speaking communities. Comprehensive evaluations reveal that the resulting model, dubbed 'AceGPT', sets the state-of-the-art standard for open Arabic LLMs across various benchmarks, including the instruction-following benchmark (i.e., Arabic Vicuna-80 and Arabic AlpacaEval), knowledge benchmark (i.e., Arabic MMLU and EXAMs), and the newly introduced Arabic Cultural and Value Alignment benchmark. Notably, AceGPT outperforms Turbo in the popular Vicuna-80 benchmark when evaluated with GPT-4, despite the benchmark's limited scale. Codes, data, and models are in https://github.com/FreedomIntelligence/AceGPT.

Maintaining a consistent persona is a key quality for any open domain dialogue system. Current state-of-the-art systems do this by training agents with supervised learning or online reinforcement learning (RL). However, systems trained with supervised learning often lack consistency as they are never punished for uttering contradictions. Additional training with RL can alleviate some of these issues, however the training process is expensive. Instead, we propose an offline RL framework to improve the persona consistency of dialogue systems. Our framework allows us to combine the advantages of previous methods as we can inexpensively train our model on existing data as in supervised learning, while punishing and rewarding specific utterances as in RL. We also introduce a simple importance sampling method to reduce the variance of importance weights in offline RL training which we call Variance-Reducing MLE-Initialized (VaRMI) importance sampling. Our automatic and human evaluations show that our framework improves both the persona consistency and dialogue quality of a state-of-the-art social chatbot.

Both standalone language models (LMs) as well as LMs within downstream-task systems have been shown to generate statements which are factually untrue. This problem is especially severe for low-resource languages, where training data is scarce and of worse quality than for high-resource languages. In this opinion piece, we argue that LMs in their current state will never be fully trustworthy in critical settings and suggest a possible novel strategy to handle this issue: by building LMs such that can cite their sources - i.e., point a user to the parts of their training data that back up their outputs. We first discuss which current NLP tasks would or would not benefit from such models. We then highlight the expected benefits such models would bring, e.g., quick verifiability of statements. We end by outlining the individual tasks that would need to be solved on the way to developing LMs with the ability to cite. We hope to start a discussion about the field's current approach to building LMs, especially for low-resource languages, and the role of the training data in explaining model generations.

Recommender systems are the cornerstone of today's information dissemination, yet a disconnect between offline metrics and online performance greatly hinders their development. Addressing this challenge, we envision a recommendation simulator, capitalizing on recent breakthroughs in human-level intelligence exhibited by Large Language Models (LLMs). We propose Agent4Rec, a novel movie recommendation simulator, leveraging LLM-empowered generative agents equipped with user profile, memory, and actions modules specifically tailored for the recommender system. In particular, these agents' profile modules are initialized using the MovieLens dataset, capturing users' unique tastes and social traits; memory modules log both factual and emotional memories and are integrated with an emotion-driven reflection mechanism; action modules support a wide variety of behaviors, spanning both taste-driven and emotion-driven actions. Each agent interacts with personalized movie recommendations in a page-by-page manner, relying on a pre-implemented collaborative filtering-based recommendation algorithm. We delve into both the capabilities and limitations of Agent4Rec, aiming to explore an essential research question: to what extent can LLM-empowered generative agents faithfully simulate the behavior of real, autonomous humans in recommender systems? Extensive and multi-faceted evaluations of Agent4Rec highlight both the alignment and deviation between agents and user-personalized preferences. Beyond mere performance comparison, we explore insightful experiments, such as emulating the filter bubble effect and discovering the underlying causal relationships in recommendation tasks.

Language models (LMs) with less than 100B parameters are known to perform poorly on chain-of-thought (CoT) reasoning in contrast to large LMs when solving unseen tasks. In this work, we aim to equip smaller LMs with the step-by-step reasoning capability by instruction tuning with CoT rationales. In order to achieve this goal, we first introduce a new instruction-tuning dataset called the CoT Collection, which augments the existing Flan Collection (including only 9 CoT tasks) with additional 1.84 million rationales across 1,060 tasks. We show that CoT fine-tuning Flan-T5 (3B & 11B) with CoT Collection enables smaller LMs to have better CoT capabilities on unseen tasks. On the BIG-Bench-Hard (BBH) benchmark, we report an average improvement of +4.34% (Flan-T5 3B) and +2.60% (Flan-T5 11B), in terms of zero-shot task accuracy. Furthermore, we show that instruction tuning with CoT Collection allows LMs to possess stronger few-shot learning capabilities on 4 domain-specific tasks, resulting in an improvement of +2.24% (Flan-T5 3B) and +2.37% (Flan-T5 11B), even outperforming ChatGPT utilizing demonstrations until the max length by a +13.98% margin. Our code, the CoT Collection data, and model checkpoints are publicly available.

Conversational systems based on Large Language Models (LLMs), such as ChatGPT, show exceptional proficiency in context understanding and response generation. However, despite their impressive capabilities, they still possess limitations, such as providing randomly-guessed answers to ambiguous queries or failing to refuse users' requests, both of which are considered aspects of a conversational agent's proactivity. This raises the question of whether LLM-based conversational systems are equipped to handle proactive dialogue problems. In this work, we conduct a comprehensive analysis of LLM-based conversational systems, specifically focusing on three aspects of proactive dialogue systems: clarification, target-guided, and non-collaborative dialogues. To trigger the proactivity of LLMs, we propose the Proactive Chain-of-Thought prompting scheme, which augments LLMs with the goal planning capability over descriptive reasoning chains. Empirical findings are discussed to promote future studies on LLM-based proactive dialogue systems.