Dimitri Coelho Mollo

Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao et al. · allen-ai, amazon-science

Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their potentially transformative impact, these new capabilities are as yet poorly characterized. In order to inform future research, prepare for disruptive new model capabilities, and ameliorate socially harmful effects, it is vital that we understand the present and near-future capabilities and limitations of language models. To address this challenge, we introduce the Beyond the Imitation Game benchmark (BIG-bench). BIG-bench currently consists of 204 tasks, contributed by 450 authors across 132 institutions. Task topics are diverse, drawing problems from linguistics, childhood development, math, common-sense reasoning, biology, physics, social bias, software development, and beyond. BIG-bench focuses on tasks that are believed to be beyond the capabilities of current language models. We evaluate the behavior of OpenAI's GPT models, Google-internal dense transformer architectures, and Switch-style sparse transformers on BIG-bench, across model sizes spanning millions to hundreds of billions of parameters. In addition, a team of human expert raters performed all tasks in order to provide a strong baseline. Findings include: model performance and calibration both improve with scale, but are poor in absolute terms (and when compared with rater performance); performance is remarkably similar across model classes, though with benefits from sparsity; tasks that improve gradually and predictably commonly involve a large knowledge or memorization component, whereas tasks that exhibit "breakthrough" behavior at a critical scale often involve multiple steps or components, or brittle metrics; social bias typically increases with scale in settings with ambiguous context, but this can be improved with prompting.

Andrea Aler Tubella, Dimitri Coelho Mollo, Adam Dahlgren Lindström et al.

Fairness is central to the ethical and responsible development and use of AI systems, with a large number of frameworks and formal notions of algorithmic fairness being available. However, many of the fairness solutions proposed revolve around technical considerations and not the needs of and consequences for the most impacted communities. We therefore want to take the focus away from definitions and allow for the inclusion of societal and relational aspects to represent how the effects of AI systems impact and are experienced by individuals and social groups. In this paper, we do this by means of proposing the ACROCPoLis framework to represent allocation processes with a modeling emphasis on fairness aspects. The framework provides a shared vocabulary in which the factors relevant to fairness assessments for different situations and procedures are made explicit, as well as their interrelationships. This enables us to compare analogous situations, to highlight the differences in dissimilar situations, and to capture differing interpretations of the same situation by different stakeholders.

Dimitri Coelho Mollo, Raphaël Millière

Large language models (LLMs) produce seemingly meaningful outputs, yet they are trained on text alone without direct interaction with the world. This leads to a modern variant of the classical symbol grounding problem in AI: can LLMs' internal states and outputs be about extra-linguistic reality, independently of the meaning human interpreters project onto them? We argue that they can. We first distinguish referential grounding -- the connection between a representation and its worldly referent -- from other forms of grounding and argue it is the only kind essential to solving the problem. We contend that referential grounding is achieved when a system's internal states satisfy two conditions derived from teleosemantic theories of representation: (1) they stand in appropriate causal-informational relations to the world, and (2) they have a history of selection that has endowed them with the function of carrying this information. We argue that LLMs can meet both conditions, even without multimodality or embodiment.

Jan-Willem van der Rijt, Dimitri Coelho Mollo, Bram Vaassen

This paper investigates how human interactions with AI-powered chatbots may offend human dignity. Current chatbots, driven by large language models (LLMs), mimic human linguistic behaviour but lack the moral and rational capacities essential for genuine interpersonal respect. Human beings are prone to anthropomorphise chatbots. Indeed, chatbots appear to be deliberately designed to elicit that response. As a result, human beings' behaviour toward chatbots often resembles behaviours typical of interaction between moral agents. Drawing on a second-personal, relational account of dignity, we argue that interacting with chatbots in this way is incompatible with the dignity of users. We show that, since second-personal respect is premised on reciprocal recognition of second-personal authority, behaving towards chatbots in ways that convey second-personal respect is bound to misfire in morally problematic ways, given the lack of reciprocity. Consequently, such chatbot interactions amount to subtle but significant violations of self-respect: the respect we are dutybound to show for our own dignity. We illustrate this by discussing four actual chatbot use cases (information retrieval, customer service, advising, and companionship), and propound that the increasing societal pressure to engage in such interactions with chatbots poses a hitherto underappreciated threat to human dignity.

Dimitri Coelho Mollo

Artificial Intelligence is a field that lives many lives, and the term has come to encompass a motley collection of scientific and commercial endeavours. In this paper, I articulate the contours of a rather neglected but central scientific role that AI has to play, which I dub `AI-as-exploration'.The basic thrust of AI-as-exploration is that of creating and studying systems that can reveal candidate building blocks of intelligence that may differ from the forms of human and animal intelligence we are familiar with. In other words, I suggest that AI is one of the best tools we have for exploring intelligence space, namely the space of possible intelligent systems. I illustrate the value of AI-as-exploration by focusing on a specific case study, i.e., recent work on the capacity to combine novel and invented concepts in humans and Large Language Models. I show that the latter, despite showing human-level accuracy in such a task, probably solve it in ways radically different, but no less relevant to intelligence research, to those hypothesised for humans.

Adam Dahlgren Lindström, Leila Methnani, Lea Krause et al.

This paper critically evaluates the attempts to align Artificial Intelligence (AI) systems, especially Large Language Models (LLMs), with human values and intentions through Reinforcement Learning from Feedback (RLxF) methods, involving either human feedback (RLHF) or AI feedback (RLAIF). Specifically, we show the shortcomings of the broadly pursued alignment goals of honesty, harmlessness, and helpfulness. Through a multidisciplinary sociotechnical critique, we examine both the theoretical underpinnings and practical implementations of RLxF techniques, revealing significant limitations in their approach to capturing the complexities of human ethics and contributing to AI safety. We highlight tensions and contradictions inherent in the goals of RLxF. In addition, we discuss ethically-relevant issues that tend to be neglected in discussions about alignment and RLxF, among which the trade-offs between user-friendliness and deception, flexibility and interpretability, and system safety. We conclude by urging researchers and practitioners alike to critically assess the sociotechnical ramifications of RLxF, advocating for a more nuanced and reflective approach to its application in AI development.