About the Good, and the Bad, of AI

Open Source for AI

The term "open source" in the context of large language models (LLMs) like Llama 2 lacks a clear definition, posing a fundamental question in the AI community. Unlike traditional software, LLMs involve complex neural network architectures with weights as a critical component. The open source definition may not neatly apply to these weights, as they can't be directly modified. Some suggest considering curated training data as the source code, but this raises ownership issues. The challenges extend to licensing the neural network architecture and handling model variations. Clear definitions are crucial as AI adoption grows, demanding clarity on usage rights and restrictions.

Google Cloud leveraging generative AI

Google Cloud is emerging as a top choice for generative AI model development and deployment. Google's Duet AI is evolving into a robust productivity tool for complex tasks. Tensor Processing Units are optimized for AI models. Google Cloud offers a range of foundation models, caters to users of all AI experience levels, and streamlines cloud-based application development with Duet AI.

Good AI product

For us to put on the Good Product list, must meet three criteria:

1. Good product

2. Company has good prospects to stay in the AI race

3. Terms of Use are acceptabl

Today’s Good is WalkMe, see details below

AI is missing……

Behind paywall, ‘AI will not get Recent Progress of Iodine Catalysis Directed toward Environmentally more Benign and Sustainable Chemistry’ Niche subject, AI is good for general knowledge (for now), a lot of research is still behind paywalls,