Large language models are able to exploit in-context learning to access external knowledge beyond their training data through retrieval-augmentation. While promising, its inner workings remain unclear. In this work, we shed light on the mechanism of in-context retrieval augmentation for question answering by viewing a prompt as a composition of informational components. We propose an attribution-based method to identify specialized attention heads, revealing in-context heads that comprehend instructions and retrieve relevant contextual information, and parametric heads that store entities' relational knowledge. To better understand their roles, we extract function vectors and modify their attention weights to show how they can influence the answer generation process. Finally, we leverage the gained insights to trace the sources of knowledge used during inference, paving the way towards more safe and transparent language models.

Manual fruit harvesting is common in agriculture, but the amount of time that pickers spend on nonproductive activities can make it very inefficient. Accurately identifying picking vs. non-picking activity is crucial for estimating picker efficiency and optimizing labor management and the harvest process. In this study, a practical system was developed to calculate the efficiency of pickers in commercial strawberry harvesting. Instrumented picking carts were used to record in real-time the harvested fruit weight, geo-location, and cart movement. A fleet of these carts was deployed during the commercial strawberry harvest season in Santa Maria, CA. The collected data was then used to train a CNN-LSTM-based deep neural network to classify a picker's activity into ``Pick" and ``NoPick" classes. Experimental evaluations showed that the CNN-LSTM model showed promising activity recognition performance with an F1 score accuracy of up to 0.974. The classification results were then used to compute two worker efficiency metrics: the percentage of time spent actively picking, and the time required to fill a tray. Analysis of the season-long harvest data showed that the pickers spent an average of 73.56% of their total harvest time actively picking strawberries, with an average tray fill time of 6.22 minutes. The mean accuracies of these metrics were 96.29% and 95.42%, respectively. When integrated on a commercial scale, the proposed technology could aid growers in automated worker activity monitoring and harvest optimization, ultimately helping to reduce non-productive time and enhance overall harvest efficiency.

People are increasingly using technologies equipped with large language models (LLM) to write texts for formal communication, which raises two important questions at the intersection of technology and society: Who do LLMs write like (model alignment); and can LLMs be prompted to change who they write like (model steerability). We investigate these questions in the high-stakes context of undergraduate admissions at a selective university by comparing lexical and sentence variation between essays written by 30,000 applicants to two types of LLM-generated essays: one prompted with only the essay question used by the human applicants; and another with additional demographic information about each applicant. We consistently find that both types of LLM-generated essays are linguistically distinct from human-authored essays, regardless of the specific model and analytical approach. Further, prompting a specific sociodemographic identity is remarkably ineffective in aligning the model with the linguistic patterns observed in human writing from this identity group. This holds along the key dimensions of sex, race, first-generation status, and geographic location. The demographically prompted and unprompted synthetic texts were also more similar to each other than to the human text, meaning that prompting did not alleviate homogenization. These issues of model alignment and steerability in current LLMs raise concerns about the use of LLMs in high-stakes contexts.

We introduce Moshi, a speech-text foundation model and full-duplex spoken dialogue framework. Current systems for spoken dialogue rely on pipelines of independent components, namely voice activity detection, speech recognition, textual dialogue and text-to-speech. Such frameworks cannot emulate the experience of real conversations. First, their complexity induces a latency of several seconds between interactions. Second, text being the intermediate modality for dialogue, non-linguistic information that modifies meaning -- such as emotion or non-speech sounds -- is lost in the interaction. Finally, they rely on a segmentation into speaker turns, which does not take into account overlapping speech, interruptions and interjections. Moshi solves these independent issues altogether by casting spoken dialogue as speech-to-speech generation. Starting from a text language model backbone, Moshi generates speech as tokens from the residual quantizer of a neural audio codec, while modeling separately its own speech and that of the user into parallel streams. This allows for the removal of explicit speaker turns, and the modeling of arbitrary conversational dynamics. We moreover extend the hierarchical semantic-to-acoustic token generation of previous work to first predict time-aligned text tokens as a prefix to audio tokens. Not only this "Inner Monologue" method significantly improves the linguistic quality of generated speech, but we also illustrate how it can provide streaming speech recognition and text-to-speech. Our resulting model is the first real-time full-duplex spoken large language model, with a theoretical latency of 160ms, 200ms in practice, and is available at https://github.com/kyutai-labs/moshi.

The Stout AgTech autonomous cultivator uses machine vision and AI to identify weeds. The smart implement then uses articulated "blades" to cut the undesired plants from their root systems. Today, farmers have a number of options for cultivating their crops. Herbicides and genetically modified crops are one cultivation method that has increasingly gone out of fashion due primarily to market pressure from consumers for organically grown crops. For organic-certified farms, the only option for cultivation is to use a mechanical means of removing weeds.

Fox News Flash top headlines for Nov. 24 are here. Check out what's clicking on FoxNews.com The prosthetic legs of a double amputee and soon-to-be high school wrestling captain were stolen from a gym closet in California last week, putting his dreams of winning a state championship or even wrestling this season in doubt. Brett Winters, a senior at Pacific High School in San Bernardino, California, was born without tibia bones in his legs. As a baby, his mother was told by doctors that Winters could either spend life in a wheelchair or amputate his legs.

Anaya Bussey didn't know much about "artificial intelligence" when she arrived at a camp at Princeton University earlier this summer other than that "it was definitely blowing up." But after just three weeks here she and other students--all incoming high school juniors--teamed up to use the technology to help diagnose melanoma by looking at skin lesions. Bussey, 15, who is from the Bronx borough in New York City, has been interested in computer science since she was in elementary school. But there have been times when she's been one of only a handful of girls--or black students--in a computer class or program. That wasn't the case at the Princeton summer camp, run by AI4ALL, a two-year-old nonprofit that seeks to increase diversity and inclusion in AI education, research, and policy.

A robotic strawberry picker built by AgroBot, a Spanish company. It's being tested in California as hiring laborers becomes increasingly difficult. But for one small corner, agricultural technology, it represents an opportunity. Farmers have been facing an increasingly tight labor market for years. The immigrant workforce that has long picked and packed the nation's fruits and vegetables move to better jobs as soon as they can, replaced by new immigrants.

On the first day of "camp," two dozen rising high school sophomores arrive at the Stanford Artificial Intelligence Laboratory's Outreach Summer program (SAILORS) giddy and ready to get started. The rigorous, two-week program is designed to encourage young women from underrepresented populations to get more involved in the field of science. High school students Ishla Zareef-Mustafa and Genaro Pamatz participate in an anatomy lab as part of the Stanford Medical Youth Science Program. On the last day of the summer residential Stanford Medical Youth Science Program (SMYSP), 24 high school students, surrounded by family members, friends and mentors, present the research they have been working on during the five-week summer program.These programs, which fall under the umbrella of Stanford Pre-Collegiate Studies, are designed to provide teenagers from underrepresented populations with an opportunity to explore careers in science, but also to build new relationships, while taking what they've learned back to their home communities. The SAILORS curriculum includes lectures, hands-on research projects and mentoring activities that are intended to educate and excite young women about artificial intelligence.

The cultivation and domestication of plants and animals first began around 12,000 years ago, making agriculture the oldest of all enterprises. It is still among the largest. Despite this heritage, the timeless uncertainties of weather, land, and demand are driving the industry to adopt artificial intelligence (AI) technologies – at least in the developed world. One example is the Western Growers Center for Innovation & Technology (WGCIT) in Salinas, California. The WGCIT was created to discover new technologies, set up testing, facilitate industry feedback, and communicate progress to produce farmers in California, Arizona, and Colorado.