- Chaos Theory
- Posts
- 🥟 Chao-Down #263 TikTok owner Bytedance launches its own answer to ChatGPT, AI is revitalizing San Francisco's commercial real estate industry, Google Maps gets "supercharged" with generative AI
🥟 Chao-Down #263 TikTok owner Bytedance launches its own answer to ChatGPT, AI is revitalizing San Francisco's commercial real estate industry, Google Maps gets "supercharged" with generative AI
Plus, Mastercard releases a GPT-like model to help banks detect fraud.
Could we be one step closer to understanding how babies learn to speak and comprehend words? By attaching a camera to an infant's head, researchers from New York University gained valuable insights into language acquisition.
From the age of six months to two years old, a baby boy Sam wore a camera on top of his head for approximately one hour twice a week. Researchers then trained an AI system on video frames and transcribed words spoken to Sam. After some time, the system learned to recognize specific words like 'crib' and 'ball.' Interestingly, this learning occurred without any prior knowledge about language, challenging the notion that babies require an innate understanding of how language works to be able to start using it.
It’s a fascinating look at how we are still just trying to figure out how humans work, let alone figuring out how tech like ChatGPT learns its knowledge.
-Alex, your resident Chaos Coordinator.
What happened in AI? 📰
AI revives San Francisco's commercial real estate industry (qz.com)
Shopify’s ‘Magic’ AI image editor can make any product pics look professional (The Verge)
TikTok owner ByteDance launches its answer to OpenAI’s GPTs, accelerating a generative AI push amid ChatGPT frenzy (South China Morning Post)
Mastercard launches GPT-like AI model to help banks detect fraud (CNBC)
AI Officer Is the Hot New Job That Pays Over $1 Million (Bloomberg)
Google Maps is getting ‘supercharged’ with generative AI (The Verge)
Always be Learnin’ 📕 📖
GenAI Can Help Small Companies Level the Playing Field (hbr.org)
RAG vs Fine-tuning: Pipelines, Tradeoffs, and a Case Study on Agriculture (arxiv)
In the AI science boom, beware: your results are only as good as your data (Nature)
Projects to Keep an Eye On 🛠
FlagOpen/FlagEmbedding: Dense Retrieval and Retrieval-augmented LLMs (Github)
run-llama/rags: Build ChatGPT over your data, all with natural language (Github)
X-PLUG/MobileAgent: Mobile-Agent: Autonomous Multi-Modal Mobile Device Agent with Visual Perception (Github)
The Latest in AI Research 💡
Corrective Retrieval Augmented Generation (arxiv)
Large Language Models for Mathematical Reasoning: Progresses and Challenges (arxiv)
A Comprehensive Survey of Compression Algorithms for Language Models (arxiv)
The World Outside of AI 🌎
Google will no longer back up the Internet: Cached webpages are dead (Ars Technica)
Don’t Be a Borrower If You Can Help It (The Atlantic)
Meta's metaverse is back - Meta's Reality Labs hit $1 billion in revenue for the first time (qz.com)
The Race Is on Between Elon Musk’s Neuralink and China (Gizmodo)
Are 'love languages' the key to healthy relationships? Scientists say no. (Mashable)
Why it feels so hard to get a job right now (CNBC)