Google’s big finish to 2024 with DeepMind is a clear indication that it will likely dominate the AI headlines in 2025. DeepMind is the AI framework that houses all of Google’s AI projects and there are several in constant development.
ChatGPT may have caught Google off guard when it took the world by storm at launch but things have changed. The latest version of Gemini is stunningly good and Veo 2, DeepMind’s video creation tool, mostly blows OpenAI’s Sora away. But Sora is publicly availably whereas Google’s imaging system is still restricted.
I expect many more AI headlines in 2025 and although Google will take the front running, Open AI will continue challenge.
GEMINI 2.0
Released a few weeks ago, Gemini 2.0 is Google’s most advanced large language model. Its multimodal design means it can understand various types of input, including text, images, audio, and video.
Gemini 2.0 can look at your screen, make sense of it, then make suggestions. You could have a spreadsheet open or powerpoint presentation, pretty much anything and it will work on it with you in real time.
It truly is amazing.
VEO 2
When OpenAI’s text-to-video tool Sora went public a lot of us early adopters were disappointed. The launch videos were mind blowing but the paid product months later was less than inspiring.
Google’s Veo 2 looks far better than anything produced by OpenAI and those with early access are producing incredible videos. Veo 2 understands film speak and it delivers a far better result with better physics, resolution and duration options.
Look at a couple of examples below.
Google is taking its time so it can identify, understand and improve the model’s quality and safety while slowly rolling it out via VideoFX, YouTube and Vertex AI. It also plans to expand Veo 2 to YouTube Shorts and other products next year.
Visit Google Labs to sign up for the waitlist.
PROJECT ASTRA
Project Astra is a research prototype exploring future capabilities of a universal AI assistant. It really shows where Google is heading with Gemini in the real world.
Better real time conversations and the ability to remember details of past interactions means Google’s assistant will become more useful over time. Think about it remembering passcodes via smart glasses, handy stuff like that.
It can also draw context from up to 10 minutes of the current conversation.
JUST THE BEGINNING
Google will need to roll out its impressive AI models to a wider audience in 2025. OpenAI will respond in the new year and don’t forget, it’s already accessible globally. It does come at a cost while Google’s AI is free, for now.
Google’s video model is potentially a game changer for the TV and Film industry. I can already see uses for my projects. That’s why I’m on the waiting list.
It’s a brave new world and it’s going to be another big year for AI in 2025.
