Hey Synapticians,

AI makes headlines every day, but let's get real: the foundation of any AI success is solid data governance and architecture. Clean, well-organized data is what makes AI actually work.

We're particularly interested in ontology-driven development, and we're not alone - Google just announced a major partnership in this space (we'll dive deeper into what this means in our next article).

When you combine a solid ontology framework with modern language models, you open up new possibilities for data analytics. Instead of complex queries, you can interact with data using natural language.

Top AI news

1. Google and App Orchid Advance Ontology-Based Business Data Analytics
App Orchid has expanded its partnership with Google Cloud to transform business data analytics using an ontology-based approach. By integrating with Google Cloud’s Cortex Framework and leveraging Gemini AI models, App Orchid enables users to extract insights through natural language queries. This ontology-driven method structures enterprise data into knowledge graphs, significantly enhancing comprehension and accuracy. App Orchid claims a 99.8% accuracy rate for text-to-SQL conversions, outperforming competitors. This collaboration marks a shift toward broader applications, making enterprise data more accessible while ensuring high precision in analysis.

2. OpenAI Introduces European Data Residency for ChatGPT Users
On February 5, 2025, OpenAI announced that new ChatGPT workspaces can be configured with data residency in Europe, enabling customer content to be stored at rest within the region. This move addresses local compliance and data sovereignty requirements, enhancing trust among European users. OpenAI remains committed to providing solutions tailored to the needs of its international clientele, considering region-specific regulations. This initiative is part of broader efforts to improve data security and privacy, facilitating the global adoption of its technologies. Interested customers can now set up their ChatGPT workspaces with this European data residency option.

3. Researchers Develop $50 AI Model “S1” Challenging OpenAI’s o1
In a significant advancement, researchers from Stanford and the University of Washington have developed an AI reasoning model with a training cost of less than $50 in cloud computing credits. This achievement highlights the potential for creating efficient and effective AI models without substantial financial investment, potentially democratizing AI research and development. The success of this project may pave the way for more accessible and open-source AI initiatives in the future.

Bonus. AI-generated ads lack human touch, expert says
The article discusses the growing role of artificial intelligence (AI) in advertising, highlighting that while AI, especially generative AI, offers efficiency in content creation, it often lacks the authentic emotional connection crucial for impactful ads. Denish Shah, a marketing professor at Georgia State University, emphasizes that the core of compelling advertising—emotional resonance, cultural nuance, and humor—must originate from human creativity. He advocates for a collaborative approach where AI augments human ingenuity, enhancing elements like video footage and personalized messaging. Shah also cautions about ethical considerations, noting that AI can reflect and amplify existing biases present in its training data.

Video of the Day

Karpathy dropped a new video to learn more about LLMs, as usual we expect his videos to be very high quality!

Theme of the Week

AI Generated Music - Scientific paper review
AI models, especially Transformers, are revolutionizing music composition by generating structured and emotionally impactful pieces, with applications in personalized therapy and relaxation.

Stay Connected

Feel free to contact us with any feedback or suggestions—we’d love to hear from you !