AI arms race heats up: Google announces Dec. launch of Gemini – will it blow GPT-4 away?

Google CEO Sundar Pichai announced the December launch of its new AI model named Gemini, trained on Google’s own TPUv5 chips.

Based on reports, Gemini sounds far more powerful than GPT-4. One of the features that Gemini boasts is multimodal data processing, meaning that the AI model analyzes not just text, but also media, including images, audio, video, and other formats. (Today, large language models analyze texts, while diffusion models analyze images.)

Another feature of Gemini is reinforcement learning by which the AI model learns by trial and error, and then refines its analysis.

Google Gemini reportedly may have 5 times the computing power as GPT-4!

As we discussed in a prior post, Google Brain scientists, in fact, invented the transformer, the breakthrough technology that gave rise to OpenAI’s ChatGPT. But since ChatGPT’s public launch in November 2022, Google has been playing catch up to keep pace with OpenAI and its business partner Microsoft. Gemini sounds like it could put Google back in the AI lead.

Leave a Reply

%d bloggers like this: