The benefits, risks and bounds of personalizing the alignment of large language models to individuals

The benefits, risks and bounds of personalizing the alignment of large language models to individuals

Large language models (LLMs) undergo ‘alignment’ so that they better reflect human values or preferences, and are safer or more useful. However, alignment is intrinsically difficult because the hundreds of millions of people who now interact with LLMs have different preferences for language and conv...

Disgrifiad llawn

Manylion Llyfryddiaeth
Prif Awduron:	Kirk, HR, Vidgen, B, Röttger, P, Hale, SA
Fformat:	Journal article
Iaith:	English
Cyhoeddwyd:	Springer Nature 2024

Eitemau Tebyg

Hatemoji: A test suite and adversarially-generated dataset for benchmarking and detecting emoji-based hate
gan: Kirk, HR, et al.
Cyhoeddwyd: (2022)

Is more data better? re-thinking the importance of efficiency in abusive language detection with transformers-based active learning
gan: Kirk, HR, et al.
Cyhoeddwyd: (2022)

Hatemoji: A test suite and adversarially-generated dataset for benchmarking and detecting emoji-based hate
gan: Kirk, H, et al.
Cyhoeddwyd: (2021)

Exploring large language models for ontology alignment
gan: He, Y, et al.
Cyhoeddwyd: (2023)

Survey on large language models alignment research
gan: LIU Kunlin, et al.
Cyhoeddwyd: (2024-06-01)

Survey on large language models alignment research
gan: LIU Kunlin, et al.
Cyhoeddwyd: (2024-06-01)

Two contrasting data annotation paradigms for subjective NLP tasks
gan: Röttger, P, et al.
Cyhoeddwyd: (2022)

Tackling racial bias in automated online hate detection: Towards fair and accurate detection of hateful users with geometric deep learning
gan: Ahmed, Z, et al.
Cyhoeddwyd: (2022)

Auditing large language models: a three-layered approach
gan: Mökander, J, et al.
Cyhoeddwyd: (2023)

Strong and weak alignment of large language models with human values
gan: Mehdi Khamassi, et al.
Cyhoeddwyd: (2024-08-01)

Aligning, autoencoding and prompting large language models for novel disease reporting
gan: Liu, F, et al.
Cyhoeddwyd: (2025)

HateCheck: functional tests for hate speech detection models
gan: Röttger, P, et al.
Cyhoeddwyd: (2021)

Deciphering implicit hate: evaluating automated detection algorithms for multimodal hate
gan: Botelho, A, et al.
Cyhoeddwyd: (2021)

Lower bounds on multiple sequence alignment using exact 3-way alignment
gan: Colbourn Charles J, et al.
Cyhoeddwyd: (2007-04-01)

Improvements in viral gene annotation using large language models and soft alignments
gan: William L. Harrigan, et al.
Cyhoeddwyd: (2024-04-01)

The Health Star Rating system – is its reductionist (nutrient) approach a benefit or risk for tackling dietary risk factors?
gan: Mark A Lawrence, et al.
Cyhoeddwyd: (2019-03-01)

Personality prediction based on large language models
gan: Wee, Jewel Xin Yu
Cyhoeddwyd: (2024)

Dyslexia application using axis align bounding boxes (AABB) /
gan: Awatif Baharuddin, 1993-, et al.
Cyhoeddwyd: (2016)

Evaluating the ability of large language models to emulate personality
gan: Yilei Wang, et al.
Cyhoeddwyd: (2025-01-01)

Directions in abusive language training data, a systematic review: Garbage in, garbage out.
gan: Bertie Vidgen, et al.
Cyhoeddwyd: (2020-01-01)

Individualizing the risks and benefits of postmenopausal hormone therapy.
gan: van Staa, T, et al.
Cyhoeddwyd: (2008)

Large Language Model Augmentation and Feature Alignment Method for Few-Shot Continual Relation Extraction
gan: LI Yifei, ZHANG Lingling, DONG Yuxuan, WANG Jiaxin, ZHONG Yujie, WEI Bifan
Cyhoeddwyd: (2024-09-01)

Reflection in Speech of the Individual-Typological Features of Language Personality
gan: Наталія Фоміна
Cyhoeddwyd: (2019-11-01)

Benefit-Risk Analysis of Buprenorphine for Pain Management
gan: Hale M, et al.
Cyhoeddwyd: (2021-05-01)

Personalized prediction of lifetime benefits with statin therapy for asymptomatic individuals: a modeling study.
gan: Bart S Ferket, et al.
Cyhoeddwyd: (2012-01-01)

Person re-identification based on large vision-language model
gan: Ding, Songyu
Cyhoeddwyd: (2024)

Aligning Large Language Models for Enhancing Psychiatric Interviews Through Symptom Delineation and Summarization: Pilot Study
gan: Jae-hee So, et al.
Cyhoeddwyd: (2024-10-01)

DrugReAlign: a multisource prompt framework for drug repurposing based on large language models
gan: Jinhang Wei, et al.
Cyhoeddwyd: (2024-10-01)

Benefits of spontaneous confidence alignment between dyad members
gan: Pescetelli, N, et al.
Cyhoeddwyd: (2022)

Hedgerow benefits align with food production and sustainability goals
gan: R Long, et al.
Cyhoeddwyd: (2017-09-01)

Hedgerow benefits align with food production and sustainability goals
gan: Rachael F. Long, et al.
Cyhoeddwyd: (2017-09-01)

Alignment-free sequence comparison: benefits, applications, and tools
gan: Andrzej Zielezinski, et al.
Cyhoeddwyd: (2017-10-01)

Tackling racial bias in automated online hate detection: Towards fair and accurate detection of hateful users with geometric deep learning
gan: Zo Ahmed, et al.
Cyhoeddwyd: (2022-02-01)

Cross-language Wikipedia editing of Okinawa, Japan
gan: Hale, SA
Cyhoeddwyd: (2015)

Constructive alignment in a graduate-level project management course: an innovative framework using large language models
gan: Estacio Pereira, et al.
Cyhoeddwyd: (2024-04-01)

Affine Matching with Bounded Sensor Error: A Study of Geometric Hashing and Alignment
gan: Grimson W. Eric L., et al.
Cyhoeddwyd: (2004)

An upper bound on the convergence rate of a second functional in optimal sequence alignment
gan: Hauser, R, et al.
Cyhoeddwyd: (2017)

Dyslexia application using axis align bounding boxes (AABB) [electronic resource] /
gan: Awatif Baharuddin, 1993-, et al.
Cyhoeddwyd: (2016)

Outward Bound as education for personal growth
gan: Katz, Richard, et al.
Cyhoeddwyd: (2009)

The Existential Truth: On Personal beyond Bounds
gan: K. S. Golikov
Cyhoeddwyd: (2020-11-01)