Aligning LMs to human preferences can be resource-intensive and technically challenging.

We participate in CheckThat! Task 2 English and explore various methods of prompting and in-context learning, including few-shot prompting and fine-tuning with different LLM families, with the goal of extracting check-worthy claims from social media passages. Our best METEOR score is achieved by fine-tuning a FLAN-T5 model. However, we observe that higher-quality claims can sometimes be extracted using other methods, even when their METEOR scores are lower.

Social platforms such as Reddit have a network of communities of shared interests, with a prevalence of posts and comments from which one can infer users' Personal Information Identifiers (PIIs). While such self-disclosures can lead to rewarding social interactions, they pose privacy risks and the threat of online harms. Research into the identification and retrieval of such risky self-disclosures of PIIs is hampered by the lack of open-source labeled datasets. Important hindrances to sharing high-quality labelled data include high annotation costs and privacy risks associated with the release of datasets containing self-disclosive text, especially when users include vulnerable populations. To foster reproducible research into PII-revealing text detection, we develop a novel methodology to create synthetic equivalents of PII-revealing data that can be safely shared. Our contributions include creating a taxonomy of 19 PII-revealing categories for vulnerable populations and the creation and release of a synthetic PII-labeled multi-text span dataset generated from 3 text generation Large Language Models (LLMs), Llama2-7B, Llama3-8B, and zephyr-7b-beta, with sequential instruction prompting to resemble the original Reddit posts. The utility of our methodology to generate this synthetic dataset is evaluated with three metrics: First, we require reproducibility equivalence, i.e., results from training a model on the synthetic data should be comparable to those obtained by training the same models on the original posts. Second, we require that the synthetic data be unlinkable to the original users, through common mechanisms such as Google Search. Third, we wish to ensure that the synthetic data be indistinguishable from the original, i.e., trained humans should not be able to tell them apart.

When prompting a language model (LM), users often expect the model to adhere to a set of behavioral principles across diverse tasks, such as producing insightful content while avoiding harmful or biased language. Instilling such principles (i.e., a constitution) into a model is resource-intensive, technically challenging, and generally requires human preference labels or examples. We introduce SAMI, an iterative algorithm that finetunes a pretrained language model (without requiring preference labels or demonstrations) to increase the conditional mutual information between constitutions and self-generated responses given queries from a dataset. On single-turn dialogue and summarization, a SAMI-trained mistral-7b outperforms the initial pretrained model, with win rates between 66% and 77%. Strikingly, it also surpasses an instruction-finetuned baseline (mistral-7b-instruct) with win rates between 55% and 57% on single-turn dialogue. SAMI requires a model that writes the principles. To avoid dependence on strong models for writing principles, we align a strong pretrained model (mixtral-8x7b) using constitutions written by a weak instruction-finetuned model (mistral-7b-instruct), achieving a 65% win rate on summarization. Finally, we investigate whether SAMI generalizes to diverse summarization principles (e.g., "summaries should be scientific") and scales to stronger models (llama3-70b), finding that it achieves win rates of up to 68% for learned and 67% for held-out principles compared to the base model. Our results show that a pretrained LM can learn to follow constitutions without using preference labels, demonstrations, or human oversight.

Anyone who's been on the internet for a while will recognise the'distracted boyfriend' meme. And now AI is bringing this classic piece of internet history to life in a whole new way. X (formerly Twitter) users have shared their creepy animations of the iconic meme created using a variety of AI animation tools. The creators have even created different alternative endings for the meme, letting us imagine what might have happened after that moment. However, not everyone is happy with the innovation as some concerned commenters joke that'nobody asked for this'.

When writing, a person may need to anticipate questions from their audience, but different social groups may ask very different types of questions. If someone is writing about a problem they want to resolve, what kind of follow-up question will a domain expert ask, and could the writer better address the expert's information needs by rewriting their original post? In this paper, we explore the task of socially-aware question generation. We collect a data set of questions and posts from social media, including background information about the question-askers' social groups. We find that different social groups, such as experts and novices, consistently ask different types of questions. We train several text-generation models that incorporate social information, and we find that a discrete social-representation model outperforms the text-only model when different social groups ask highly different questions from one another. Our work provides a framework for developing text generation models that can help writers anticipate the information expectations of highly different social groups.

Below we summarize some of the significant developments employers should be on the lookout for in the new year. On August 12, 2019, Governor Andrew Cuomo of New York signed into law a bill that, as previously reported, significantly strengthened and expanded workplace anti-discrimination protections in New York State. For additional information regarding the developments already in effect, refer to our previous posts. In terms of changes still to come, contracts and other agreements entered into on or after January 1, 2020, that prevent the disclosure of information relating to any future claim of discrimination on the basis of any protected characteristic will be unenforceable, unless the provision notifies the individual that it does not prohibit them from speaking with law enforcement, the Equal Employment Opportunity Commission, the New York State Division of Human Rights ("NYSDHR"), a local commission on human rights, or an attorney retained by the individual. Likewise, effective February 8, 2020, the New York State Human Rights Law will be expanded to include all employers in the state, regardless of size.

Almost six months ago (May 28th 2018) I posted the "AI winter is well on its way" post that went viral. The post amassed nearly a quarter million views and got picked up in Bloomberg, Forbes, Politico, Venturebeat, BBC, Datascience Podcast and numerous other smaller media outlets and blogs [1, 2, 3, 4, ...], triggered violent debate on Hacker news and Reddit. I could not have anticipated this post to be so successful and hence I realized I touched on a very sensitive subject. One can agree with my claims or not, but the sheer popularity of the post almost itself serves as a proof that something is going on behind the scenes and people are actually curious and doubtful if there is anything solid behind the AI hype. Since the post made a prediction, that the AI hype is cracking (particularly in the space of autonomous vehicles) and as a result we will have another "AI winter" episode, I decided to periodically go over those claims, see what has changed and bring some new evidence.

Hitherto the present, there has been a post floating around the internet detailing multiple "types" of artificial intelligence, purportedly written by someone named "Yuli Ban". If you see this post, know that it wasn't written by me at all, absolutely not, I take no responsibility for the cringy contents of that post, and you are likely remembering something that never existed or perhaps was written by my evil twin, Tali. In all seriousness, I've been meaning to update that post for a while now thanks to some greater understanding of how AI works. I recall mentioning how it was a smorgasbord of buzzwords without much meaning, written by someone in 2016 with no experience in AI whatsoever. This one, I hope, provides greater usefulness. Artificial intelligence has a problem: no one can precisely tell you what it is supposed to be.