AI Reading for Monday March 17

ChatGPT calls OpenAI's own spokesperson's statement about US penalizing use of DeepSeek confusing and equivocal - TechCrunch

You know it’s a big threat when they start calling for bans and spreading FUD.

---

Chinese models like DeepSeek R1 and Qwen 2.5 attract big enterprise demand on Amazon, but Amazon wants people to use their homegrown Nova. - Business Insider

(Not using DeepSeek via China APIs is a no-brainer but there should be no privacy issue about using open source models hosted in Bedrock. You have one job AWS, if you can't host R1 securely and privately you can't do it with Nova either, hope you aren't spreading FUD against your own services LOL!)

---

Maybe Baidu’s new Ernie too? - Business Insider

And then there’s the Manus agent project.

---

How Jack Ma led an Alibaba pivot to AI behind the scenes, and rehabilitated both Alibaba and himself. - FT

---

Big Tech may ramp up AI investment to try to stay ahead of DeepSeek and Alibaba - Bloomberg

---

Asian AI suppliers gain ahead of Nvidia's big GTC GPU / AI event. - WSJ

---

Sun Micro said the network is the computer, now the data center is the computer and GPU loads and high speed interconnects dominate. - NY Times

---

AI will probably be used to cut costs and make customer service worse, with less human empathy and less ability to break out of Kafkaesque infinite loops. - FT

---

AI still sucks at reading analog clocks - Gizmodo

I sometimes get stories about AI Weiwei in my tech AI feeds … today it was Ai Ogura, I spent a moment trying to figure out where the AI angle was.

---

Should we store all of humanity's knowledge in an AI repository deep underground in Svalbard or Antarctica or something? - Hacker Noon

At this point we have so many massive training data sets, might as well put them in persistent storage in multiple locations for future robot civilzations to remember us by after the apocalypse 🙄.

---

Uploading your chat history to ChatGPT and asking it to analyze your relationship. Sounds like a minefield as well as recipe for self-examination and self-improvement. - Business Insider

---

Researchers use Whisper text-to-speech activations to predict brain patterns during speech better than text alone, suggesting that deep learning processes to understand spoken language are at least partially isomorphic to what the human brain does. - livescience.com

---

Follow the latest AI headlines via SkynetAndChill.com on Bluesky