Google’s AI to Train on Public Data
Plus: Give personality to your LLMs, LEDITS by Hugging Face Research.
Hello, Engineering Leaders and AI Enthusiasts!
Welcome to the 56th edition of The AI Edge newsletter. This edition brings you Google’s new policy update for using publicly available data to train AI.
A huge shoutout to our incredible readers. We appreciate you! 😊
In today’s edition:
🌍 Google’s AI models to train on public data
🧠 Pick your LLM’s personality type
🤗 LEDITS: Image editing with next-level AI capabilities
📚 Knowledge Nugget: Get the most out of your custom LLMs by
Let’s go!
Google’s AI models to train on public data
Google has updated its privacy policy to state that it can use publicly available data to help train and create its AI models. As of July 1, here is what the tech giant's newly updated policy reads
This suggests that Google is leaning heavily into its AI bid. Plus, harnessing humanity’s collective knowledge could redefine how AI learns and comprehends information.
Why does this matter?
Privacy and intellectual property concerns have constantly plagued the AI landscape. But leveraging public data could contribute valuable insights and broaden the scope of training data. Moreover, Google has always been an advocate of responsible AI.
Will it be able to strike a balance between data accessibility, privacy, and ethical considerations to ensure the development of trustworthy and beneficial AI systems?
Pick your LLM’s personality type
LLMs are trained on vast amounts of human-generated data, enabling them to mimic human characteristics in their outputs and enact convincing personas—in other words, exhibit a form of synthetic personality. Thus, personality becomes a crucial factor in determining the effectiveness of communication.
New research has proposed a comprehensive method for administering validated psychometric tests and quantifying, analyzing, and shaping personality traits exhibited in text generated from widely-used LLMs.
Why does this matter?
This could offer new insights into the behavior of LLMs. Moreover, LLMs are becoming the dominant human-computer interaction (HCI) interface. But many have displayed unintended behaviors such as manipulative language, bais, hallucinations, etc. This method could help engineer LLM-synthesized personality profiles for safety, appropriateness, and effectiveness.
LEDITS: Image editing with next-level AI capabilities
Hugging Face research has introduced LEDITS- a combined lightweight approach for real-image editing, incorporating the Edit Friendly DDPM inversion technique with Semantic Guidance. Thus, it extends Semantic Guidance to real image editing while harnessing the editing capabilities of DDPM inversion.
Why does this matter?
Editing in large-scale text-guided diffusion models can become challenging due to the inherent nature of editing techniques, especially for attaining one-shot generation. Plus, current methods can add latency. LEDITS is an efficient approach and helps achieve versatile edits without additional training and optimizations.
Knowledge Nugget: Get the most out of your custom LLMs
It's no surprise that custom LLMs offer numerous benefits over generalized models.
Recently, Bloomberg released BloombergGPT, a large-scale generative AI model purpose-built from scratch for finance. It outperforms similarly-sized open models on financial NLP tasks by significant margins– without sacrificing performance on general LLM benchmarks.
In this insightful article,
discusses the importance of custom LLMs and the four stages of training them. It also introduces WeightWatcher (WW), an open-source tool that can be instrumental in building custom LLMs and foundation models.
Why does this matter?
While this tool eases the building of custom AI models, its emergence also suggests that the exceptional performance of custom LLMs and foundation models in AI is driving increased investment in tools that help develop better AI systems.
What Else Is Happening❗
🧙♂️Harry Potter in anime avatar by Stable Diffusion (Link)
🚫OpenAI disables ChatGPT's "Browse" beta to do right by content owners (Link)
💰Celestial AI raises $100 million for its Photonic Fabric technology platform (Link)
🚀Inflection AI develops supercomputer with 22,000 NVIDIA H100 AI GPUs (Link)
🚲Urtopia unveils Fusion e-bike with ChatGPT integration (Link)
🛠️ Trending Tools
Storify AI: AI-based interactive content creation with human voice, boosting customer engagement.
AI Webcam Effects: Enhance video calls with virtual backgrounds, beautification, and more.
Artistic AI QR Generator: Delightful scanning experience with artistic AI-generated QR Codes, democratizing genAI technologies.
Real AI: AI Photo Generator. Easily create or transform photos using AI technology with Real AI.
Castup AI: ChatGPT-powered podcast assistant for recording and promoting amazing episodes with the help of AI.
Later: AI Caption Writer generating social media post captions that maximize engagement potential across multiple platforms.
Quoradata: AI-powered investment research app for actionable insights within 30 seconds by analyzing public company information.
Promptify: Free Chrome plugin optimizing AI-generated image prompts for authors.
That's all for now!
If you are new to ‘The AI Edge’ newsletter. Subscribe to receive the ‘Ultimate AI tools and ChatGPT Prompt guide’ specifically designed for Engineering Leaders and AI enthusiasts.
Thanks for reading, and see you tomorrow. 😊