Google has launched its new artificial intelligence software called “Gemini” to compete with OpenAI’s ChatGPT and claims it is better at math, coding, and reasoning tasks. Gemini is a large language model trained on billions of images and sentences from the internet and can understand math problems and provide advice on how to solve them. It can also take instructions in various formats and is integrated with Google’s Bard chatbot.
Gemini will be integrated into various Google services and products, starting with Bard and Pixel 8 Pro smartphones. It comes in three tiers – Gemini Ultra, Gemini Pro, and Gemini Nano – each with different capabilities. Google claims that Gemini outperforms OpenAI’s ChatGPT and is better than GPT-4. The AI model is designed for multimodal performance and has shown impressive results in various industry benchmarks. Gemini Ultra, for data centers and enterprise applications, is expected to be available early next year.
According to MIT Technology Review, while it outperforms GPT-4 on many measures, the differences between the two models are relatively thin. This could indicate that the AI hype has reached its peak.
Gemini will not be available in Europe due to regulatory hurdles. Gemini Pro will be available to developers and enterprise customers starting December 13.
And amidst all this from Google, Apple quietly has released MLX, a machine learning framework and model library designed to run efficiently on Apple Silicon. MLX allows developers to build models for generative AI apps on MacBooks. It is accessible through open-source repositories like GitHub and PyPI. MLX Data, a framework agnostic package for data loading, works with MLX, PyTorch, or Jax frameworks. Apple’s focus on generative AI applications departs from its previous emphasis on machine learning.
MIT Technology Review’s assessment that the performance differences between Gemini and GPT-4 are marginal suggests a potential plateau in AI advancements, at least in the current generation of technology. We’ll see. The tiered offerings – Gemini Ultra, Pro, and Nano – indicate a targeted approach to different user segments, from data centers to individual developers.
We should expect Apple to be quiet until they announce. Expect a strategic play to leverage Apple’s hardware ecosystem, offering an integrated, efficient solution for AI model development.
All align with the concept of service providers as advisors here.