Alibaba Cloud invests in ShengShu, aiming to develop AI that understands and replicates the real world through video and physical scenarios.
Alibaba Cloud invests in ShengShu, aiming to develop AI that understands and replicates the real world through video and physical scenarios.
  • Alibaba Cloud shifts focus from large language models to "world models" for AI.
  • The company invested $290 million in ShengShu, a startup specializing in AI video generation.
  • ShengShu aims to create a "general world model" connecting digital and physical domains.
  • Alibaba is investing in startups that build AI tools grounded in physical space.

The Dwindling Appeal of Blah-Blah Bots

Right, so apparently everyone's suddenly realized that those chatbot things are about as useful as Brian after a wine cooler binge. I mean, honestly, babbling text isn't going to cut it when we're trying to build robots that don't trip over the cat. So, Alibaba, bless their cotton socks, is throwing wads of cash at this 'world model' idea. Essentially, they're building AI that understands that when you throw a ball, it doesn't just vanish into the ether like my hopes for a decent martini. It's all about replicating the real world, darling.

ShengShu's Grand Plan to Conquer Reality

This ShengShu outfit, apparently the apple of Alibaba's eye, is all about bridging the gap between the digital and physical realms. Think video games that bleed into autonomous driving, and robots that actually know which end is up. Founder Zhu Jun, a man after my own heart, wants AI that can predict real-world behaviour consistently. And wouldn't that be a novelty. It seems that the world of shipping is also going through a transformation as companies are trying to get ahead of the competition. Find out more about FedEx Strikes Back Same-Day Shipping Alliance Targets Amazon and see how the shipping world is evolving with time.

From Pixels to Physicality: A Multimodal Revolution

Now, they're not just chucking code at the wall and hoping it sticks. They're talking about 'multimodal data' - vision, audio, touch. It's like giving AI a proper education, instead of just force-feeding it Wikipedia. ShengShu's Vidu Q3 Pro model is already making waves, apparently churning out videos that don't look like they were drawn by a caffeinated chimpanzee. Which, let's face it, is a low bar to clear.

Following the Money: Alibaba's Startup Spending Spree

Alibaba isn't just betting on ShengShu. They're spreading the love, investing in outfits like Tripo AI, which conjures up 3D models from photographs, and PixVerse, which lets you direct videos as they're being generated. It's like they're building their own little army of AI geniuses. A genius army that hopefully won't turn on us and demand endless servings of caviar and world domination. Although, the caviar wouldn't be so bad.

Robots Incoming: The Embodied AI Era

And here's where it gets interesting: embodied AI. Robots, you see, need more than just fancy language models to function. They need to understand that fire is hot, cats are furry, and Lois is… well, let's just say 'complicated'. ShengShu is teaming up with companies building these humanoid robots, for use in factories, shops, and even our own homes. Imagine, a robot butler who actually understands sarcasm. The possibilities are endless, and mildly terrifying.

The Future of AI: A World Model Renaissance

Kevin Kelly, some tech guru from Wired magazine, reckons AI needs reasoning, physical understanding, and continuous learning to truly mimic human intelligence. We've got the knowledge part down with these LLM thingies, but the world models are the key to unlocking the rest. So, buckle up, because if Alibaba has its way, we're about to enter an era where AI can finally tell the difference between a martini and a glass of dishwater. A development, I assure you, that is long overdue. Quite right.


Comments

  • No comments yet. Become a member to post your comments.