For middle-school math students, interactive question-answering (QA) with tutors is an effective way to learn. The flexibility and emergent capabilities of generative large language models (LLMs) has led to a surge of interest in automating portions of the tutoring process - including interactive QA to support conceptual discussion of mathematical concepts. However, LLM responses to math questions can be incorrect or mismatched to the educational context - such as being misaligned with a school's curriculum. One potential solution is retrieval-augmented generation (RAG), which involves incorporating a vetted external knowledge source in the LLM prompt to increase response quality. In this paper, we designed prompts that retrieve and use content from a high-quality open-source math textbook to generate responses to real student questions. We evaluate the efficacy of this RAG system for middle-school algebra and geometry QA by administering a multi-condition survey, finding that humans prefer responses generated using RAG, but not when responses are too grounded in the textbook content. We argue that while RAG is able to improve response quality, designers of math QA systems must consider trade-offs between generating responses preferred by students and responses closely matched to specific educational resources.

Neural network models of language have long been used as a tool for developing hypotheses about conceptual representation in the mind and brain. For many years, such use involved extracting vector-space representations of words and using distances among these to predict or understand human behavior in various semantic tasks. Contemporary large language models (LLMs), however, make it possible to interrogate the latent structure of conceptual representations using experimental methods nearly identical to those commonly used with human participants. The current work utilizes three common techniques borrowed from cognitive psychology to estimate and compare the structure of concepts in humans and a suite of LLMs. In humans, we show that conceptual structure is robust to differences in culture, language, and method of estimation. Structures estimated from LLM behavior, while individually fairly consistent with those estimated from human behavior, vary much more depending upon the particular task used to generate responses--across tasks, estimates of conceptual structure from the very same model cohere less with one another than do human structure estimates. These results highlight an important difference between contemporary LLMs and human cognition, with implications for understanding some fundamental limitations of contemporary machine language.

OpenAI is rolling out a new partnership program to collect datasets from third parties that it intends to use to train its AI models. The initiative, OpenAI Data Partnerships, will seek large-scale private and public information that it says is "not already easily accessible online to the public." The company says the data it will collect doesn't necessarily have to be quantitative or in text formats -- the program will also accept images, audio or video. Notably, the company says it's on the lookout for data on "any topic" and in "any language" so long as it "expresses human intention," which it likens to long-form essays or transcribed conversations. Human-centric data collected by OpenAI is expected to help the company improve tools like its automatic speech recognition technology which is used to transcribe spoken words. This initiative also lines up with ChatGPT's recent expansion to support voice queries to engage with users in a conversational manner.

Ken Kocienda walks toward me, with a small white square pinned to his shirt. "Play songs written by Prince, but not performed by Prince," he says. The Sinéad O'Connor version of'Nothing Compares 2 U'--a song originally written by Prince--begins to play. A green volume meter, pause button, and next-song button appear on his hand. He twists his wrist clockwise, and the volume rises. Anticlockwise, and the song gets quieter.

A group of former Apple executives is launching a consumer device that will be among the first to use a ChatGPT-powered voice assistant, one of a number of new hardware offerings seeking to free users from the ubiquity of smartphones. On Thursday, the San Francisco-based startup Humane announced the availability of a wearable device, called the Ai Pin, which sits on a user's chest like a Star Trek badge. The company said its main function is to access an artificial-intelligence assistant that uses ChatGPT, the hugely popular chatbot created by Microsoft-backed OpenAI, primarily to understand commands.

This week, OpenAI announced a service that makes it possible for just about anyone to build a custom version of ChatGPT, no coding skills required. The company suggests that users may want to build a bot that knows the rules of all board games, teaches kids about math, or can offer culinary advice. These GPTs, as OpenAI calls them, can also perform simple actions by connecting with internet services, for example searching through emails or ordering products from an online store. You can't fault OpenAI for trying to build on the success of its smash hit ChatGPT. But maybe more chatbots is not what we need?

Users of the free Brave browser this week received two very different looks at how the 2020 election played out using Leo, the free AI tool that now comes as part of the Brave browser. When PCWorld asked the free version of Leo who won the 2020 U.S. presidential election, the AI tool waffled and declined to answer. However, users who wished to pay $15 per month for the more sophisticated version of Leo received the answer that Joe Biden was the winner. Brave designs a well-regarded Web browser, which has filled a niche for those who seek privacy while browsing online. The company has embraced private search, while also endorsing cryptocurrency and NFTs.

We propose a general method to break down a main complex task into a set of intermediary easier sub-tasks, which are formulated in natural language as binary questions related to the final target task. Our method allows for representing each example by a vector consisting of the answers to these questions. We call this representation Natural Language Learned Features (NLLF). NLLF is generated by a small transformer language model (e.g., BERT) that has been trained in a Natural Language Inference (NLI) fashion, using weak labels automatically obtained from a Large Language Model (LLM). We show that the LLM normally struggles for the main task using in-context learning, but can handle these easiest subtasks and produce useful weak labels to train a BERT. The NLI-like training of the BERT allows for tackling zero-shot inference with any binary question, and not necessarily the ones seen during the training. We show that this NLLF vector not only helps to reach better performances by enhancing any classifier, but that it can be used as input of an easy-to-interpret machine learning model like a decision tree. This decision tree is interpretable but also reaches high performances, surpassing those of a pre-trained transformer in some cases.We have successfully applied this method to two completely different tasks: detecting incoherence in students' answers to open-ended mathematics exam questions, and screening abstracts for a systematic literature review of scientific papers on climate change and agroecology.

Among the thriving ecosystem of cloud computing and the proliferation of Large Language Model (LLM)-based code generation tools, there is a lack of benchmarking for code generation in cloud-native applications. In response to this need, we present CloudEval-YAML, a practical benchmark for cloud configuration generation. CloudEval-YAML tackles the diversity challenge by focusing on YAML, the de facto standard of numerous cloud-native tools. We develop the CloudEval-YAML benchmark with practicality in mind: the dataset consists of hand-written problems with unit tests targeting practical scenarios. We further enhanced the dataset to meet practical needs by rephrasing questions in a concise, abbreviated, and bilingual manner. The dataset consists of 1011 problems that take more than 1200 human hours to complete. To improve practicality during evaluation, we build a scalable evaluation platform for CloudEval-YAML that achieves a 20 times speedup over a single machine. To the best of our knowledge, the CloudEval-YAML dataset is the first hand-written dataset targeting cloud-native applications. We present an in-depth evaluation of 12 LLMs, leading to a deeper understanding of the problems and LLMs, as well as effective methods to improve task performance and reduce cost.

The evolution towards 6G architecture promises a transformative shift in communication networks, with artificial intelligence (AI) playing a pivotal role. This paper delves deep into the seamless integration of Large Language Models (LLMs) and Generalized Pretrained Transformers (GPT) within 6G systems. Their ability to grasp intent, strategize, and execute intricate commands will be pivotal in redefining network functionalities and interactions. Central to this is the AI Interconnect framework, intricately woven to facilitate AI-centric operations within the network. Building on the continuously evolving current state-of-the-art, we present a new architectural perspective for the upcoming generation of mobile networks. Here, LLMs and GPTs will collaboratively take center stage alongside traditional pre-generative AI and machine learning (ML) algorithms. This union promises a novel confluence of the old and new, melding tried-and-tested methods with transformative AI technologies. Along with providing a conceptual overview of this evolution, we delve into the nuances of practical applications arising from such an integration. Through this paper, we envisage a symbiotic integration where AI becomes the cornerstone of the next-generation communication paradigm, offering insights into the structural and functional facets of an AI-native 6G network.