• Chaos Theory
  • Posts
  • 🥟 Chao-Down #284 Google targets AI spam in search results, Study finds ChatGPT-created workouts are too cautious and incomplete, TurboTax and H&R Block chatbots give poor advice 50% of the time

🥟 Chao-Down #284 Google targets AI spam in search results, Study finds ChatGPT-created workouts are too cautious and incomplete, TurboTax and H&R Block chatbots give poor advice 50% of the time

Plus, a look at how AI is already doing jobs like screening resumes and rental applications and determining medical care with little supervision.

How is AI changing the type of work people do?

Henley Chiu from Bloomberry recently analyzed data from the freelance marketplace Upwork and found that since the release of ChatGPT, AI seems to be having the most negative impact on translation work, creative writing jobs, and market research jobs. Interestingly growth showed up more in higher-order and more complex tasks like accounting, sales, and video editing/production. 

Will AI continue to eat more pieces of work?

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

Google now wants to limit the AI-powered search spam it helped create (Ars Technica)

ChatGPT’s AI-generated workouts were incomplete, too cautious, study says (The Washington Post)

TurboTax and H&R Block’s AI chatbots are giving bad tax advice (The Washington Post)

How businesses are actually using generative AI (The Economist)

China offers AI computing ‘vouchers’ to its underpowered start-ups (ft.com)

AI is already screening job resumes and rental apartment applications and even determining medical care with almost no oversight (Fortune)

Always be Learnin’ 📕 📖

microsoft/Security-101: 7 Lessons, Kick-start Your Cybersecurity Learning. (Github)

Accessibility Has Failed: Try Generative UI = Individualized UX (substack.com)

The Age of Incumbents - Distribution + Data Become Concentration Functions (substack.com)

Projects to Keep an Eye On 🛠

ConsequentAI/fneval: Functional Benchmarks and the Reasoning Gap (Github)

hpcaitech/Open-Sora: Building your own video generation model like OpenAI's Sora (Github)

Scenario Diffusion helps Zoox vehicles navigate safety-critical situations (Amazon Science)

The Latest in AI Research 💡

Large Language Models for Data Annotation: A Survey (arxiv)

Evaluating Quantized Large Language Models (arxiv)

Resonance RoPE: Improving Context Length Generalization of Large Language Models (arxiv)

The World Outside of AI 🌎

Lead-Tainted Applesauce Highlights Failings in Food Safety System (The New York Times)

Speeding tickets aren’t enough to save lives. A tech fix can make it physically impossible to drive at ridiculous speeds. (Vox)

Forced to Change: Tech Giants Bow to Global Onslaught of Rules (The New York Times)

The Meditation Start-Up That’s Selling Bliss on Demand (The Atlantic)

Climate Change and ‘Last-Chance Tourism’ (The New York Times)

Cookie Monster puts focus on shrinkflation. Here's what it means (qz.com)

One Last Bite 😋