Google has quietly opened a "backdoor" to the magical world of AI for a select group of lucky customers who have purchased the Vertex AI Google Cloud package. These customers now have private access to Google's generative AI models, Veo and Imagen 3.
As of today, these privileged users can experiment with high-tech “magic.” With Veo, they can input a text prompt or upload an image, and voilà, within moments, a 1080p HD, over-a-minute-long, coherent video is generated. Whether starting with AI-created imaginative images or their own photos, users have a fantastic base for video creation. However, Veo, like a novice magician, occasionally makes a few humorous mistakes, especially when handling cause-and-effect relationships. For instance, in a scene depicting marshmallow roasting, the marshmallow stubbornly remains “unmelted” and perfectly still by the fire, almost as if protected by a "fireproof shield." Additionally, the video sometimes features minor glitches, such as artifacts.
But don’t worry, Google has more tricks up its sleeve. Next week, it will release its latest text-to-image model, Imagen 3, to these users. Google has boldly claimed that Imagen 3 can generate the “most realistic, highest-quality images” from simple text, with rich detail and near-perfect lighting. The new version also drastically reduces artifacts, marking a significant improvement over previous iterations—it's like upgrading from a toy gun to a high-powered rifle. Yet, sharp-eyed users have quietly noted, “Google, it seems there's still room for improvement here.”
Generative AI: Opportunities and Challenges for Businesses
While these AI models may have some "quirks," Google’s enthusiasm remains undiminished. The company is determined to onboard more enterprise customers to dive into the world of generative AI. Why? Google’s own research shows that companies utilizing generative AI in production report a remarkable 86% of them seeing an increase in revenue. However, there’s a twist—Appen, a company that recently conducted a survey, found that from 2023 to 2024, the return on investment (ROI) for AI projects dropped by 4.6 percentage points. This serves as a reminder that while generative AI is advancing rapidly, companies hoping to strike a bigger “gold mine” need to refine their “gold-digging techniques” in real-world applications to maximize ROI.
Ranking
