🦖AI Turns Smartphones Into Tools for Dinosaur Discovery

🦖AI Turns Smartphones Into Tools for Dinosaur Discovery

Scientists have released DinoTracker, a new AI-powered mobile app that lets users identify ancient dinosaur footprints by uploading images of track silhouettes, a major shift in how paleontological field data can be analyzed outside the lab. Rather than training on potentially flawed, pre-labeled datasets, the system was fed about 2,000 unlabeled footprint outlines and taught itself which shape features matter most (e.g., toe spread, heel placement, ground contact). The AI then groups footprints based on these learned features, achieving classifications that agree with expert human judgments roughly 90% of the time.

Unlike traditional methods that require paleontologists to match prints to known species by eye, and risk bias when labels are wrong, DinoTracker offers a structured, data-driven way to explore how tracks relate to each other and to possible dinosaur makers. Users can not only compare an uploaded print with similar ones in the database, but also interactively adjust shape features to see how classifications shift.

The project also reinforces ongoing debates about evolutionary history. Some clusters of prints from the Triassic and early Jurassic resemble the shape of bird feet despite being tens of millions of years older than the oldest known bird fossils. Researchers caution, however, that this may reflect how feet interacted with sediment rather than proving early birds existed at those times.

Practical takeaway: DinoTracker doesn’t replace expert paleontology, but it amplifies it, turning pattern recognition from a manual, subjective task into a scalable, interactive one accessible to scientists and enthusiasts alike. It’s a powerful example of AI lowering barriers to domain-specific analysis while still leaving room for expert verification and contextual judgement.

🔨 AI Tools & Updates: Claude Turns Into a Unified Workplace Interface

Anthropic has expanded Claude from a standalone AI assistant into a single interface for interacting with core workplace apps. Using an extension of the Model Context Protocol (MCP), Claude now surfaces interactive versions of tools like Slack, Asana, Canva, Box, and Figma directly inside the chat window, not just text summaries or one-off API actions. Users can build project timelines in Asana, draft and send Slack messages with formatted previews, or edit Figma diagrams without switching browser tabs or apps.

Previously Claude could trigger actions in connected tools behind the scenes; the update lets users actually interact with those tools’ interfaces within Claude’s UI itself. This move positions Claude as a central productivity hub rather than a siloed assistant, potentially reducing friction in cross-tool workflows and cutting down context-switching during work sessions.

Practical takeaway: Claude is no longer “just” an AI chat assistant, for paid plans on web/desktop, it’s now a front door into everyday work apps, letting you view and manipulate actual app content inside a single interface. If this works reliably, it could meaningfully streamline how teams handle tasks, communication, design, and project management without opening multiple tabs or tools.

🗺️Google Maps Brings Hands-Free Gemini to Walking & Cycling

Google Maps is expanding the reach of its built-in Gemini AI assistant beyond driving to walking and cycling navigation, letting users engage in real-time, conversational queries while they’re on the move. With this update, you can ask Gemini practical questions like “What neighborhood am I in?”, “What are top-rated restaurants nearby?”, or “Are there places with bathrooms along my route?” without stopping or switching screens, addressing the awkwardness and safety risks of typing mid-stride or mid-pedal.

On a bike, the hands-free mode is explicitly oriented toward safety and convenience: you can check ETA, confirm calendar events, or even send a text like “Text Emily I’m 10 minutes behind” without taking your hands off the handlebars (texting features are currently available on Android). Gemini activation works via voice (“Hey Google”) or by tapping the Maps interface, effectively turning Maps into a conversational travel co-pilot.

This rollout is global on iOS now and expanding to Android, and it builds on Google’s prior work embedding Gemini into driving navigation, moving Maps away from static turn-by-turn directions toward contextual, interactive guidance tailored to what users need in the moment.

Practical takeaway: This shift illustrates how AI assistants are creeping into every part of our daily movement, not just desktop tasks or voice queries at home. If the hands-free conversation model works reliably, it could reduce friction and cognitive load on foot or bike, turning navigation from passive instruction into dynamic, on-route decision support, but real-world safety and distraction issues will be worth watching as this gets broader adoption.

🧩 Closing Thought
Taken together, these updates suggest that the next phase of AI adoption will be defined less by raw model capability and more by interface design, context awareness, and restraint. When AI becomes the layer through which we navigate cities, manage work, or interpret historical evidence, small design choices start to carry outsized consequences. The risk isn’t that AI fails spectacularly, it’s that it works well enough to fade into the background before we’ve fully thought through how much agency we’re handing over. The winners in this phase won’t just build smarter systems, but ones that know when to stay quiet, when to assist, and when to defer back to human judgment.