AI models
AI models on Wazzu: a running collection of 3 stories we have gathered and hand-picked because they are worth your time. Every post here touches on ai models in some way — the news, the analysis, the deep dives, and the occasional surprise find. Washington State University (WSU) is a leading public land-grant research university in Pullman, WA, known for its strong programs in agriculture, environmental science,… New stories are added to this page as we find them, so check back if you want to keep up with what is happening around ai models, or subscribe to the RSS feed to get them as soon as they are published. Browse the collection below, or head back to the homepage to see everything Wazzu is covering right now.
UW researchers lead and support new ‘AI-for-Science’ Genesis Mission awards
## UW Researchers Spearhead National AI-for-Science Initiative University of Washington researchers are at the forefront of a groundbreaking national effort to revolutionize scientific discovery. Leading and collaborating on four projects, they are key players in the U.S. Department of Energy’s (DOE) Genesis Mission, a Phase 1 initiative designed to build a robust “AI-for-Science” ecosystem. This ambitious program aims to dramatically accelerate breakthroughs across critical fields including energy, discovery science, and national security. The Genesis Mission prioritizes a tiered approach to AI integration. **Ranked by impact potential**, these initial projects will focus on: 1) developing foundational AI tools tailored for scientific workflows, 2) applying these tools to specific scientific challenges, 3) building a collaborative community to share knowledge and resources, and 4) establishing a sustainable infrastructure for continued AI-driven research. UW’s involvement underscores the university's commitment to innovation and its leadership in applying cutting-edge technology to address complex global challenges. This investment promises to unlock new avenues for scientific exploration and create a ripple effect of advancements across multiple disciplines. For a glimpse into other impactful UW research happening right now, check out "August research highlights," which details fascinating findings from nectar-robbing honeycreepers to the impact of ChatGPT on attachment styles.
Video: Curiosity, frustration and antipathy: How kids play with AI toys
## Decoding Kids & AI: When Curiosity Turns to Antics University of Washington researchers recently explored a fascinating, and sometimes unsettling, dynamic: how children interact with artificial intelligence toys. In a study involving eight kids on campus, researchers observed a predictable initial phase of curiosity, marked by standard introductory questions like, "What is your name?" However, the interaction quickly evolved. When communication faltered, the children’s responses shifted from inquiry to playful, and occasionally pointed, antagonism. Reports detail kids playfully taunting the toys, labeling them with terms like “ugly” and “evil,” and even jokingly threatening to discard them. This research highlights a crucial point: even with increasingly sophisticated AI, the potential for miscommunication and frustration remains. It underscores the need to understand how children process and react to technology that doesn’t always behave as expected. The findings offer valuable insights for developers aiming to create more intuitive and engaging AI experiences for young users. The study’s observations raise broader questions about the ethical considerations of AI in childhood, particularly as systems become more integrated into education and entertainment. For deeper insights into AI's impact on creative content, see our related article, "AI models nearly erase female characters when they write kids stories about animals."
AI models nearly erase female characters when they write kids stories about animals
## AI Storytelling Faces a Gender Gap: Where Are the Female Animal Characters? New research from the University of Washington reveals a concerning trend in AI-generated children’s stories: female characters are vanishing. AI systems like Google’s Gemini Storybook, designed to create personalized illustrated tales, consistently underrepresent women when crafting stories about talking animals. A recent study analyzing six leading AI models found a stark imbalance – 57% of characters were gender neutral or ungendered, 41% were male, and a mere 2% were female. This disparity highlights a significant challenge in AI development: the potential for perpetuating and even amplifying existing societal biases. While these tools offer exciting possibilities for personalized education and creative expression, the lack of female representation in these narratives could subtly shape children's perceptions of gender roles. Researchers are actively investigating the root causes of this bias, likely stemming from skewed training data used to build these models. Addressing this issue requires a concerted effort to diversify datasets and refine AI algorithms to ensure more equitable representation. This isn't just about storytelling; it's about building AI systems that reflect and promote a more inclusive world. For further exploration of scientific discoveries, consider our recent article, "NASA’s Hubble shows star formation in Andromeda galaxy winding down," which details fascinating findings about celestial activity.