AI Reading for Sunday April 20

OpenAI guide - AI in the Enterprise: Lessons from seven frontier companies - OpenAI

Pairs well with the practical guide to agents (previously shared) - OpenAI

---

Tests say new OpenAI models hallucinate more - TechCrunch

be careful out there

maybe there is a scaling law for hallucination, the smarter the model is the more weird it might get, some forms of wackiness require a PhD

---

Google launches its own quantized models, letting you run Gemma 3 on your laptop with minimal performance degradation. Will be interesting to see how they stack up vs the artisanal quantized models on Hugging Face. - Google

---

"in five years, every video uploaded to YouTube could be dubbed automatically into every spoken language. Each word will sound like the actual voice of the person talking, their lips reanimated to move like a native speaker’s." - Bloomberg

---

Google’s Search for Its AI Future - The Information

---

From gifted student to AI hero, DeepSeek founder inspires a nation - South China Morning Post

---

AI removes the last vestiges of humanity from tech and the online world - FT

Are you AI, or simply a grammar pedant? Also, what has ‘humanity’ ever done for shareholder value, anyway? - FT

---

AI deepfake journalists spread Russian disinformation - Eurasia Review

---

Los Alamos uses AI to translate legacy Fortran to C++. - Kyle Dickman

Very light, want to see the eventual paper, how much they were able to translate and what performance gain/loss they achieved, how many bugs, whether the code was more maintainable.

---

Humanoid robots run 1/2 marathon in Beijing - WIRED

---

AI and open source intel is cool but might be opening the door to easy deceptions - Business Insider

---

AI said Lina was at 'medium' risk of domestic violence. Then she was killed. - BBC

---

Community Colleges Hit By Wave of ‘Bot’ Students Using AI to Receive Student Aid - PCMAG

---

Why AI Is the New Social Media - TheWrap

---

U.S. Chipmakers Fear They Are Ceding China’s A.I. Market to Huawei - NY Times

---

Meta asked Microsoft and Amazon to help fund Llama - WinBuzzer

---

A2A explained (Diamant) - Nir Diamant

---

‘Black Mirror’ Showed Us a Future. Some of It Is Here Now. - NY Times

---

Smart TV Ad Tech Analyzes Viewers' Emotions For Targeted Ads? Black Mirror stuff. - WIRED

---

Reddit users are asking ChatGPT to make images of how ChatGPT sees their dynamic and some of them are gold, Jerry! - Reddit

Seriously, these are fine art, check them out

---

Follow the latest AI headlines via SkynetAndChill.com on Bluesky