AI’s Evolutionary Journey: Gemini's Blunders to Stable Diffusion's Masterstrokes and Health Predictions

Feb 26, 2024

This week’s AI news contains a series of evolution and side-effects of recent announcements in the world of AI. Google's Gemini AI had a bit of a blunder, mixing up historical figures with different races and genders. On the bright side, ElevenLabs is now adding soundtracks to their mind-blowing video creations. And Stable Diffusion 3 is stealing the show with its ultra-realistic images that can handle even the trickiest prompts. Oh, and doctors have a new weapon against Alzheimer's.

Google's Gemini Gaffe: Apologies and Culture Clashes

Google's AI tool, Gemini, was supposed to be inclusive, but it went overboard. When people asked for historically accurate images, Gemini would sometimes swap in images of people from different races or genders.

Google says it didn't think about cases where accuracy was more important than diversity, like when someone wanted an image of a U.S. Senator from the 1800s. Gemini gave them an Asian woman, even though the first woman wasn't elected until the 1900s.

This shows a bigger problem at Google: people are afraid to speak up when they see something wrong. They'd rather avoid trouble than risk their jobs. But companies need people to point out problems, especially when they're building AI tools that could have unintended consequences. Read more about it here.

AI's Soundtrack Serenades Sora's Videos

ElevenLabs, the speech-synth masters, are now adding AI-generated soundtracks to AI-generated videos! Just like OpenAI's incredible Sora model that makes mind-blowing videos from scratch, ElevenLabs is making it possible to auto-tune your videos with footsteps, waves, and ambient sounds.

ElevenLabs is aiming to be your one-stop shop for all things AI-video. Remember their game-changing dubbing tool? Now they're expanding into sound effects. And let's not forget about the good ol' text-to-sfx models, like AudioGen. But hold your applause for now, ElevenLabs' sounds promise to be extra natural.

The future of AI-video looks bright, where tools can automatically analyze videos and add sound effects like pros. And with music AI tools evolving, from text-to-music to image-to-music, the day may come when we can conjure up an entire movie from a single prompt.

Stable Diffusion 3: Outpacing the AI Art Kings

Stable Diffusion 3 from Stability AI is coming, and it's about to blow the competition out of the water. Get ready for images so realistic, you'll forget they're made by a computer!

Stable Diffusion 3 has been practising hard since its previous versions and has improved its spelling skills and ability to handle complex prompts. While other AI artists like Midjourney still struggle with word puzzles, Stable Diffusion 3's got it covered! Plus, with different model sizes available, you can choose just how intricate your creations can be, from small masterpieces to mind-blowing epics. Read the full announcement here.

AI Crystal Ball Predicts Alzheimer's Years Ahead:

An AI model can predict Alzheimer's disease up to seven years before symptoms show up, giving doctors a head start on managing this brain-cruncher. The model analyzes health records and spots patterns that point to a higher risk, like high cholesterol and osteoporosis, especially for women.

The AI model isn't just a yes/no machine. It digs into the why. By combining health data with molecular databases, the model discovered that high cholesterol and osteoporosis (in women) could be linked to Alzheimer's through specific genes. This knowledge could pave the way for new treatments. Read more here.