Another Crazy Day in AI: Google's New Way to Learn Languages
- Wowza Team
- May 3
- 4 min read
Hello, AI Enthusiasts.
Another week down, another excuse to celebrate. But before you toast to the weekend, here’s a peek at something refreshingly practical: Google wants to make language learning less robotic. With three new AI experiments powered by Gemini, users can now get help speaking like real people, responding in the moment, and even learning from the world around them.
Meanwhile, a landmark lawsuit against Meta is heating up. Turns out some pirated books may have fueled Llama’s brainpower—and a group of authors isn’t thrilled about it.
And yes, AI just got into business school. One professor ran OpenAI’s latest model through a Harvard case study—and it nearly beat the official answer key.
Power down your inbox and power up your weekend.
Here's another crazy day in AI:
Tools for real-world language learning
Mays Business School launches AI partnership with Perplexity
o3 nails Harvard case in one go from pdf
Some AI tools to try out
TODAY'S FEATURED ITEM: Learning by Doing and Seeing
Image Credit: Wowza (created with Ideogram)
Ever been caught in a foreign country needing the perfect phrase — and realizing you never learned it?
In a bid to make language learning more useful in real-life situations, Google has launched three early-stage AI experiments designed to help people learn languages in a more dynamic and personalized way. TechCrunch's consumer news reporter, Aisha Malik, dives into how Google's Gemini model powers tools that go beyond textbook grammar and formal speech, and instead help you speak like a local, respond in the moment, and even learn from what you see around you.
Here’s a look at the new experiments and what they aim to do:
Tiny Lessons gives you quick, context-specific vocabulary and grammar tips. You describe a situation (like “finding a lost passport”), and the tool responds with relevant phrases and possible replies.
Slang Hang helps you pick up casual, everyday language by simulating conversations between native speakers. It’s designed to show how real conversations unfold — complete with slang and informal tone.
Word Cam lets you snap a photo of your surroundings to learn vocabulary for the objects in front of you. It offers additional words too, encouraging deeper observation and more immersive learning.
These tools are accessible via Google Labs and currently support a wide range of languages, including Arabic, Hindi, Japanese, French, Portuguese, and many others.
What makes these experiments worth paying attention to is the way they attempt to meet learners where they actually are—in unfamiliar places, unexpected conversations, or simply curious moments during the day. Instead of relying solely on memorized vocabulary or structured lessons, they offer ways to engage with language as it naturally shows up in everyday life. Whether you’re trying to report a lost passport or just want to understand what a street vendor said, tools like these are built for those real, often messy moments of communication.
There’s still plenty to improve, especially when it comes to accuracy and nuance. But the idea that language tools could be more responsive, visual, and grounded in real situations is a meaningful development for independent learners. These experiments might not replace traditional study, but they could help bridge the gap between classroom learning and the real-world confidence it takes to speak up when it matters most.
Read the full article here.
Check them out on Google Labs.
OTHER INTERESTING AI HIGHLIGHTS:
Meta Faces Key Copyright Test Over AI Training
/Cristina Criddle and Hannah Murphy on Financial Times
Meta is heading to court in one of the first major legal tests of whether tech companies can use copyrighted materials to train AI. A group of U.S. authors, including Ta-Nehisi Coates, claim Meta scraped content from LibGen, a shadow library hosting pirated books, to train its Llama models. Internal emails revealed during discovery suggest Meta staff knowingly entered a legal grey area and took steps to obscure their data sources. The case could set a precedent for how copyright law applies to AI and whether “fair use” extends to training generative models.
Read more here.
o3 Nails Harvard Case in One Go from PDF
/Ethan Mollick on LinkedIn
Ethan Mollick tested OpenAI's new o3 model on a Harvard Business School case—and the results were impressive. From a single PDF, the AI generated a full set of financial projections and compared its output to the official case data, coming within 4% of the correct answers. Mollick highlights that while o3 still occasionally hallucinates, its MBA-level reasoning and modeling were remarkably coherent. The comparison revealed minor, explainable discrepancies, showcasing the tool’s potential in academic and analytical work.
Read more here.
SOME AI TOOLS TO TRY OUT:
Chikka – AI voice interviewer that chats and extracts actionable insights instantly.
Freebeat – Turn music and ideas into viral videos with one click.
ContentRadar – Create and repurpose LinkedIn and X content effortlessly using AI.
That’s a wrap on today’s Almost Daily craziness.
Catch us almost every day—almost! 😉
EXCITING NEWS:
The Another Crazy Day in AI newsletter is on LinkedIn!!!
Leveraging AI for Enhanced Content: As part of our commitment to exploring new technologies, we used AI to help curate and refine our newsletters. This enriches our content and keeps us at the forefront of digital innovation, ensuring you stay informed with the latest trends and developments.
Kommentarer