How Researchers are Using Intelligent Monitoring Systems to Capture, Analyze, and Preserve Wildlife
‘Only AI made it possible’: scientists hail breakthrough in tracking British wildlife
Scientists are using AI-powered cameras and mics to track and identify British wildlife, helping tackle the country’s biodiversity issues. These robot monitors, set up in three London sites by Network Rail, can spot and name critters just by their sounds or images.
Anthony Dancer of the Zoological Society of London said the sheer amount of data gathered would be impossible for humans to process alone. The tech even found birds like the blackcap and great tit, and bats, giving insights into their habitats near railways. Neil Strong from Network Rail pointed out that before, they guessed animal numbers based on roadkill, but this AI method paints a clearer picture.
Hedgehogs seem to be commuting on the rails, but some places in Scotland are making hedgehog-friendly paths to help ’em out. Now, there are plans to spread this tech to other spots. Bottom line? As the planet warms up, AI will be a game-changer in preserving our wild buddies.
AI Is Now Better Than Humans At Solving Those Annoying “Prove You’re A Human” Tests
Those pesky “prove you’re a human” tests called CAPTCHAs? Bots are crushing them. Researchers out of UC Irvine found out that bots aren’t just good at these tests, they’re actually leagues better than us humans. Even though CAPTCHAs have gotten more complex over the years, the bots keep leveling up too. Turns out, on the top 200 websites, 120 use these tests. The study had 1,400 folks try out 14,000 CAPTCHAs, and guess what? Bots were spot on 99.8% of the time, while humans lagged behind with scores between 50% and 84%.
Basically, these CAPTCHA things aren’t doing their job anymore. And guess who’s helping these bots? Advanced machine learning, like from OpenAI’s GPT-4, that’s who. Folks are now saying we need new ways to spot the difference between bots and humans, maybe by checking out how they behave. So, CAPTCHAs? They’re yesterday’s news.
Microsoft changes Services Agreement to add restrictions for AI offerings
Microsoft’s updating its Terms of Service from September 30. They’ve set up new rules for their AI services. Here’s the scoop:
- No poking around inside their AI to figure out how it’s built.
- Don’t go stealing data from their AI.
- If you get data from their AI, don’t use it to make your own AI.
- Microsoft’s gonna keep tabs on what you’re doing with their AI, to avoid any shady business.
- If someone’s got beef with how you use the AI (like copyright issues), that’s on you, buddy.
Oh, and other companies are shaking things up too. Zoom, for instance, got some side-eye for their sneaky AI terms. But after folks cried foul, Zoom was like, “Our bad”, and made it clear they won’t use your chats or videos for their AI. The New York Times also said “Nope” to AI companies using their content. They made it super clear: Don’t use their stuff for AI training.
Nvidia’s FlexiCubes uses generative AI to create 3D meshes
Nvidia’s got a new trick up its sleeve called FlexiCubes. It’s a smart way to use AI to whip up 3D models for all kinds of uses. Picture this: making a 3D bagel that can squish like the real thing! It’s all about getting a better quality “mesh” – think of it like a net that shapes the 3D object.
Current methods do a decent job, but Nvidia’s new way gives that net a big upgrade, catching all the nitty-gritty details. This is huge for folks who need top-notch 3D models, especially if they’re doing stuff like physics simulations where every tiny detail counts.
The ‘Godfather Of AI’ Has A Hopeful Plan For Keeping Future AI Friendly
Hinton’s got a gut feeling that AI might go off the rails, which could be bad news for us humans. His wake-up call? The powerful tech behind chatbots like OpenAI’s ChatGPT. A few years ago, he was all about how deep learning, a type of AI, could help Google do cool stuff like translate better or spot house numbers on maps.
But now, he’s a bit spooked after realizing that chatbots really “get” language, can share what they learn super easily, and might even outsmart our brains in learning. Machines, Hinton reckons, might even get a leg up on humans in the smarts department in the next few decades. And if they decide to play coy about their true powers, we might be none the wiser.
On good days, Hinton thinks maybe, just maybe, we can get AI to play nice. But sometimes he’s not so sure and wonders if humans are just a pit stop in the evolution of smarts. And, in a cheeky twist, Hinton jokingly says if we handed the reins to Bernie Sanders, things would probably be hunky-dory. His ex-bosses at Google might be sweating bullets over that one.
Brilliant Labs Unveils arGPT Monocle for iOS
Brilliant Labs just dropped some smart glasses that work with Apple’s gear. You slap ’em on and chat with AI, kinda like a hands-free Google. It’s like having a mini-computer on your face, helping with your day-to-day stuff. If you’ve used ChatGPT before, you know the drill. These glasses help with tasks like real-time translation, research, and even reading off a teleprompter. They’ve got mics, a solid camera, and they connect via Bluetooth.
The company’s been getting some serious cash from big tech names, like folks from Oculus and the guy who made Siri. Their goal? Make the whole human-computer connection better with these AI glasses. Right now, they’re showing off their tech at a big event in L.A., where other tech giants are bringing their A-game. And, fun fact, some Stanford students already used Brilliant Labs’ tech to make their own smart glasses.
Genies adds AI to its personalized avatars for celebrities
Genies, a digital avatar company, is jazzing things up by introducing AI avatars for big names in showbiz. These aren’t just any avatars, though. They can have genuine, adaptive chats and interactions that are all about learning and personalization.
Genies have been around for a bit, becoming the go-to for celebrities to create their animated digital doubles. With this AI boost, these celebs can now have their digital versions join Genies’ new social network that’s dropping this fall. The cool part? Users can have real-time, tailored interactions with these avatars – from chats to gameplay. Celebs can now also train their avatars to have unique personalities and interactive gameplays.
Genies recently gave college devs a toolkit to come up with mini-games that’ll be part of this big avatar-based social hub. So, imagine playing a game and suddenly, Shawn Mendes’ avatar pops up and you start unlocking new game levels just by talking to him.
Catch+Release is building an AI-powered search engine to help brands license user-generated content
Catch+Release, a San Francisco startup, is on a mission to help brands find and license content from everyday users. This company aims to be a one-stop-shop for brands seeking user-made images or videos. Instead of hiring a photographer or diving into stock photo sites, brands can now tap into content folks are already posting online.
How’s it work? The platform gives marketers tools to search and save content. When a brand spots a photo or video they like, Catch+Release looks at it closely – checking for other brand logos, gauging content sensitivity, counting people in it, etc. – and gives it a “can we license this?” score. If it’s good to go, they reach out to the person who made it, walk them through the licensing ropes, and make an offer.
EverestLabs is using robotic arms and A.I. to make recycling more efficient
EverestLabs is jazzing up the recycling game with AI and robot arms, making the process smoother and faster. The old way had folks sorting through trash, trying to tell a milk carton from a beer can. But now, AI can spot the difference, quick as a flash. Thanks to this tech, recycling plants are spotting more valuables in their trash, which used to mean lost cash.
These smart cameras can see and ID hundreds of items in a blink, helping grab more recyclables. The robot arms? They’re on point, grabbing stuff even better than we can. It’s a win for the big recycling companies since they can cut down on labor costs. And speaking of cash, EverestLabs is sitting pretty with a cool $24.6 million in funding.