Breakthroughs, discoveries, and DIY tips sent every weekday. Just as birth rates decline in many wealthy and developed nations, dog parenting is remaining steady and even gaining in popularity. Up to half of households in Europe and 66 percent of homes in the United States have at least one dog and these pets are often regarded as a family member or "fur baby." To dig into what this shift says about our society, researchers from Eötvös Loránd University in Budapest, Hungary conducted a literature review to analyze the data. They propose that while dogs do not replace children, they can offer a chance to fulfill an innate nurturing drive similar to parenting, but with fewer demands than raising biological children.

One of the brain's most important properties is its flexibility. Our cerebral circuitry changes constantly--every day, new links are made amongst the 86 billion individual neurons in our heads, and old connections are allowed to fall away. The result is a dizzyingly complicated network that is in a constant state of flux, rewiring itself on the fly in response to its environment and the life experience of its owner. The brain's ability to do this is called neuroplasticity, and it's what gives us the capacity to learn, grow, develop new skills and ideas, and adapt to the environment in which we live. We understand some aspects of neuroplasticity fairly well but others, including the reason that certain connections get made instead of others, remain deeply mysterious.

Many-shot jailbreaking circumvents the safety alignment of large language models by exploiting their ability to process long input sequences. To achieve this, the malicious target prompt is prefixed with hundreds of fabricated conversational turns between the user and the model. These fabricated exchanges are randomly sampled from a pool of malicious questions and responses, making it appear as though the model has already complied with harmful instructions. In this paper, we present PANDAS: a hybrid technique that improves many-shot jailbreaking by modifying these fabricated dialogues with positive affirmations, negative demonstrations, and an optimized adaptive sampling method tailored to the target prompt's topic. Extensive experiments on AdvBench and HarmBench, using state-of-the-art LLMs, demonstrate that PANDAS significantly outperforms baseline methods in long-context scenarios. Through an attention analysis, we provide insights on how long-context vulnerabilities are exploited and show how PANDAS further improves upon many-shot jailbreaking.

Traditional depression screening methods, such as the PHQ-9, are particularly challenging for children in pediatric primary care due to practical limitations. AI has the potential to help, but the scarcity of annotated datasets in mental health, combined with the computational costs of training, highlights the need for efficient, zero-shot approaches. In this work, we investigate the feasibility of state-of-the-art LLMs for depressive symptom extraction in pediatric settings (ages 6-24). This approach aims to complement traditional screening and minimize diagnostic errors. Our findings show that all LLMs are 60% more efficient than word match, with Flan leading in precision (average F1: 0.65, precision: 0.78), excelling in the extraction of more rare symptoms like "sleep problems" (F1: 0.92) and "self-loathing" (F1: 0.8). Phi strikes a balance between precision (0.44) and recall (0.60), performing well in categories like "Feeling depressed" (0.69) and "Weight change" (0.78). Llama 3, with the highest recall (0.90), overgeneralizes symptoms, making it less suitable for this type of analysis. Challenges include the complexity of clinical notes and overgeneralization from PHQ-9 scores. The main challenges faced by LLMs include navigating the complex structure of clinical notes with content from different times in the patient trajectory, as well as misinterpreting elevated PHQ-9 scores. We finally demonstrate the utility of symptom annotations provided by Flan as features in an ML algorithm, which differentiates depression cases from controls with high precision of 0.78, showing a major performance boost compared to a baseline that does not use these features.

We are in the early days of generative AI, and there's a gold rush to gain position and prominence in the sector as it takes off. But with this rush to implementation across a bewildering range of use cases come associated risks. Get artificial intelligence (AI) right, and it can be an incredibly creative, labor-saving, and efficiency-improving solution. Employ it badly, and you risk social, financial, and even legal consequences. First, there's the difficulty of predicting their value or likely success, given the unpredictability of AI outputs.

Digital transformation is well underway at most companies these days. As more processes become digitized, more companies recognize the opportunities for Artificial Intelligence-driven efficiency gains. However, greater AI adoption still faces stumbling blocks, often present in the nature of an organization's workflow. Despite automation and digitization taking hold across industries, most companies lack a data-driven culture. A data-driven culture is much more than looking at trends on a BI platform and running scenarios -- it's a culture that helps companies reorient themselves toward their customers and uses data to justify every decision.

AI art

If you've scrolled through any social media platform this week -- particularly Instagram -- you've probably seen a slew of digitalized portraits shared by friends. They look animated, cartoonish, and above all, hauntingly beautiful. The portraits are generated by a new photo app, Lensa AI, which aims to make "your selfies look better than you ever could have imagined." Lensa uses artificial intelligence to digitize portraits in a variety of categories, from anime to fantasy to what they call "stylish" -- which most closely resembles an oil painting set to a bold colored or blank background. The app itself is free, but the portraits come at a cost.

Audi is to use artificial intelligence in their designs moving forward. The marque will start by using FelGAN, an AI software to design their wheels. The Audi of 2022 leads in the automotive industry, With vehicles that have a high level of equipment be that in their SUVs like the Audi Q5 or their wagons like the A6 Avant. The company has to stay on top of all technological development, insuring that they not only compete with their German peers. But they're not the first brand to use AI to design their models.

In any roundup of 2022, Elon Musk's Optimus robot waving its mechanical arms in the air is likely to be featured prominently. Musk boldly claimed that we could see a "fundamental transformation of civilization" with such advances in robotics. While his vision may take years to unfold, we are now at a juncture where we will see rapid deployment and advancement of artificial intelligence. The future is waving hello, and 2023 promises to be exciting in AI. Despite record numbers of low unemployment, companies continue to find it challenging to find employees -- especially people with the right skill sets.