Yelp has announced new AI-powered call answering features for restaurants and services as part of its Spring Product Release. With the service, currently under development, the company hopes that "businesses never have to miss a call again." "In this next step of our product transformation, we're continuing to harness AI to unlock the potential of Yelp's rich data in ways that build trust and simplify decision-making -- whether users are hiring a pro or booking a reservation," Yelp's chief product officer, Craig Saldanha, said in a statement. "By grounding our AI in real consumer behavior and business data, we're creating intuitive, transparent features that improve the experience for everyone on Yelp." The AI-powered system "will be fully integrated into Yelp's platform with customizable features and the ability to answer general questions, filter spam, transfer calls when needed, and capture messages."

The crowd-sourced review site Yelp unveiled a new feature that uses AI and customer reviews to rate common facets of nightlife and food-related business. The new Review Insights feature is available now on the iOS version of the Yelp app, according to the company's official blog. Review Insights aggregates customer reviews and feeds them into a large language model (LLM), which will assign specific aspects -- like the vibe or service time -- a rating out of 100. Supposedly it will be able to infer customer sentiment about these parts of a business "even when a review doesn't explicitly mention one of the topics." Yelp will also be adding an AI-powered homepage in the coming weeks.

Prompt tuning is a modular and efficient solution for training large language models (LLMs). One of its main advantages is task modularity, making it suitable for multi-task problems. However, current soft-prompt-based methods often sacrifice multi-task modularity, requiring the training process to be fully or partially repeated for each newly added task. While recent work on task vectors applied arithmetic operations on full model weights to achieve the desired multi-task performance, a similar approach for soft-prompts is still missing. To this end, we introduce Task Prompt Vectors, created by element-wise difference between weights of tuned soft-prompts and their random initialization. Experimental results on 12 NLU datasets show that task prompt vectors can be used in low-resource settings to effectively initialize prompt tuning on similar tasks. In addition, we show that task prompt vectors are independent of the random initialization of prompt tuning. This allows prompt arithmetics with the pre-trained vectors from different tasks. In this way, by arithmetic addition of task prompt vectors from multiple tasks, we are able to outperform a state-of-the-art baseline in some cases.

Compositional generalization, representing the model's ability to generate text with new attribute combinations obtained by recombining single attributes from the training data, is a crucial property for multi-aspect controllable text generation (MCTG) methods. Nonetheless, a comprehensive compositional generalization evaluation benchmark of MCTG is still lacking. We propose CompMCTG, a benchmark encompassing diverse multi-aspect labeled datasets and a crafted three-dimensional evaluation protocol, to holistically evaluate the compositional generalization of MCTG approaches. We observe that existing MCTG works generally confront a noticeable performance drop in compositional testing. To mitigate this issue, we introduce Meta-MCTG, a training framework incorporating meta-learning, where we enable models to learn how to generalize by simulating compositional generalization scenarios in the training phase. We demonstrate the effectiveness of Meta-MCTG through achieving obvious improvement (by at most 3.64%) for compositional testing performance in 94.4% cases.

We present the first domain-adapted and fully-trained large language model, RecGPT-7B, and its instruction-following variant, RecGPT-7B-Instruct, for text-based recommendation. Experimental results on rating prediction and sequential recommendation tasks show that our model, RecGPT-7B-Instruct, outperforms previous strong baselines. We are releasing our RecGPT models as well as their pre-training and fine-tuning datasets to facilitate future research and downstream applications in text-based recommendation. Public "huggingface" links to our RecGPT models and datasets are available at: https://github.com/VinAIResearch/RecGPT

Generative AI has ushered in the ability to generate content that closely mimics human contributions, introducing an unprecedented threat: Deployed en masse, these models can be used to manipulate public opinion and distort perceptions, resulting in a decline in trust towards digital platforms. This study contributes to marketing literature and practice in three ways. First, it demonstrates the proficiency of AI in fabricating disinformative user-generated content (UGC) that mimics the form of authentic content. Second, it quantifies the disruptive impact of such UGC on marketing research, highlighting the susceptibility of analytics frameworks to even minimal levels of disinformation. Third, it proposes and evaluates advanced detection frameworks, revealing that standard techniques are insufficient for filtering out AI-generated disinformation. We advocate for a comprehensive approach to safeguarding marketing research that integrates advanced algorithmic solutions, enhanced human oversight, and a reevaluation of regulatory and ethical frameworks. Our study seeks to serve as a catalyst, providing a foundation for future research and policy-making aimed at navigating the intricate challenges at the nexus of technology, ethics, and marketing.

Yelp just released a substantial app update with more than 20 new features, and several of these tools are packed with, wait for it, AI. The biggest news for regular users is the addition of summaries of business automatically written by AI, which Yelp says will help people find the perfect restaurant or service to meet their needs. There's also new visuals for the home feed and revamped search experience, which the company says will also help users find that perfect dinner spot. This home feed incorporates AI to provide more relevant content to users and will also display images from nearby restaurants that match previous user queries, in addition to videos posted by local businesses. The AI tomfoolery extends to business users.

Online reviews in the form of user-generated content (UGC) significantly impact consumer decision-making. However, the pervasive issue of not only human fake content but also machine-generated content challenges UGC's reliability. Recent advances in Large Language Models (LLMs) may pave the way to fabricate indistinguishable fake generated content at a much lower cost. Leveraging OpenAI's GPT-4-Turbo and DALL-E-2 models, we craft AiGen-FoodReview, a multi-modal dataset of 20,144 restaurant review-image pairs divided into authentic and machine-generated. We explore unimodal and multimodal detection models, achieving 99.80% multimodal accuracy with FLAVA. We use attributes from readability and photographic theories to score reviews and images, respectively, demonstrating their utility as hand-crafted features in scalable and interpretable detection models, with comparable performance. The paper contributes by open-sourcing the dataset and releasing fake review detectors, recommending its use in unimodal and multimodal fake review detection tasks, and evaluating linguistic and visual features in synthetic versus authentic data.

Numerous studies have highlighted the privacy risks associated with pretrained large language models. In contrast, our research offers a unique perspective by demonstrating that pretrained large language models can effectively contribute to privacy preservation. We propose a locally differentially private mechanism called DP-Prompt, which leverages the power of pretrained large language models and zero-shot prompting to counter author de-anonymization attacks while minimizing the impact on downstream utility. When DP-Prompt is used with a powerful language model like ChatGPT (gpt-3.5), we observe a notable reduction in the success rate of de-anonymization attacks, showing that it surpasses existing approaches by a considerable margin despite its simpler design. For instance, in the case of the IMDB dataset, DP-Prompt (with ChatGPT) perfectly recovers the clean sentiment F1 score while achieving a 46\% reduction in author identification F1 score against static attackers and a 26\% reduction against adaptive attackers. We conduct extensive experiments across six open-source large language models, ranging up to 7 billion parameters, to analyze various effects of the privacy-utility tradeoff.

On the internet, every app can be a dating app. Just ask Strava user Courtney Hollingsworth. She has long used the exercise-tracking app Strava to log her runs and workouts, and leave kudos and comments for fellow athletes. Two years ago, she also used it to let a fellow runner know she was single and interested.