How did we rating last time round? Our 4 hot trends to observe out for in 2024 included what we called customized chatbots—interactive helper apps powered by multimodal large language models (check: we didn’t realize it yet, but we were talking about what everyone now calls agents, the most popular thing in AI at once); generative video (check: few technologies have improved so fast within the last 12 months, with OpenAI and Google DeepMind releasing their flagship video generation models, Sora and Veo, inside per week of one another this December); and more general-purpose robots that may do a wider range of tasks (check: the payoffs from large language models proceed to trickle right down to other parts of the tech industry, and robotics is top of the list).
We also said that AI-generated election disinformation could be in all places, but here—happily—we got it flawed. There have been many things to wring our hands over this 12 months, but political deepfakes were thin on the bottom.
So what’s coming in 2025? We’re going to disregard the apparent here: You may bet that agents and smaller, more efficient, language models will proceed to shape the industry. As an alternative, listed below are five alternative picks from our AI team.
1. Generative virtual playgrounds
If 2023 was the 12 months of generative images and 2024 was the 12 months of generative video—what comes next? In the event you guessed generative virtual worlds (a.k.a. video games), high fives all round.
We got a tiny glimpse of this technology in February, when Google DeepMind revealed a generative model called Genie that would take a still image and switch it right into a side-scrolling 2D platform game that players could interact with. In December, the firm revealed Genie 2, a model that may spin a starter image into a complete virtual world.
Other corporations are constructing similar tech. In October, the AI startups Decart and Etched revealed an unofficial Minecraft hack through which every frame of the sport gets generated on the fly as you play. And World Labs, a startup cofounded by Fei-Fei Li—creator of ImageNet, the vast data set of photos that kick-started the deep-learning boom—is constructing what it calls large world models, or LWMs.
One obvious application is video games. There’s a playful tone to those early experiments, and generative 3D simulations may very well be used to explore design concepts for brand new games, turning a sketch right into a playable environment on the fly. This may lead to thoroughly latest sorts of games.
But they may even be used to coach robots. World Labs desires to develop so-called spatial intelligence—the power for machines to interpret and interact with the on a regular basis world. But robotics researchers lack good data about real-world scenarios with which to coach such technology. Spinning up countless virtual worlds and dropping virtual robots into them to learn by trial and error could help make up for that.
