Yeah, I guess that’s right. They were probably only pre-trained base models at that time and didn’t have good instruction-following abilities.
RESEARCH
-
Google’s Risk Aversion: Why OpenAI Launched Chatbots First
By
–
They had the Meena and Lamda chatbots, which came out in 2020 and 2021 (
https://
en.wikipedia.org/wiki/LaMDA).
I think a startup would have launched them, but Google didn't because of risk.
They needed someone else to test the waters first. I think that is why OpenAI was first. -
How LLMs Download the Internet: Karpathy Explains
By
–
Andrej Karpathy, Eureka Labs founder & former Director of AI at Tesla, breaks down how LLMs like ChatGPT "download the Internet."
— MIT CSAIL (@MIT_CSAIL) 14 décembre 2025
Watch the full beginner-friendly breakdown here: https://t.co/doph3alGnE pic.twitter.com/EaNowJhVdSAndrej Karpathy, Eureka Labs founder & former Director of AI at Tesla, breaks down how LLMs like ChatGPT "download the Internet." Watch the full beginner-friendly breakdown here: https://
bit.ly/4o13PTx -
Programming in the Age of AI: Why Code Still Matters
By
–
Programming In The Age Of AI: Why Code Still Matters
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @demishassabis @Ronald_vanLoon @TamaraMcCleary @geoffreyhinton @goodfellow_ian @jeffdean @erikbryn -
Nokia’s AI and Quantum Sensing Networks Reshape Future Communication
By
–
Exploring the Future with Nokia: AI, Quantum Sensing Networks and Space Communication Discover how cutting edge AI, quantum sensing and satellite networks could reshape communication and sensing — redefining what’s possible. Read more https://
bernardmarr.com/exploring-the-
future-with-nokia-ai-quantum-sensing-networks-and-space-communication/
… #AI -

AI Agent Adoption Study: Perplexity Comet Browser Insights
By
–
9. AI Agent Adoption Study Presents the first large-scale field study of AI agent adoption using hundreds of millions of anonymized interactions from Perplexity’s Comet browser.
-
Expertise Gap in AI Efficiency Research Criticism
By
–
The person that wrote the article is one of the most significant figures behind the efficiency work you're claiming he doesn't understand. There is indeed sloppy thinking here, but you need to look closer to home to see it.
-
ResNet34 Achieves 94% Accuracy on IIIT Pets Dataset
By
–
BTW resnet34 gets 94% accuracy on IIIT Pets after 90s fine-tuning on an old 2080 gaming GPU.
-

Verbalized Sampling: Mitigating Mode Collapse in LLMs
By
–
Verbalized Sampling: How to Mitigate Mode Collapse and Unlock LLM Diversity Zhang et al.: https://
arxiv.org/abs/2510.01171 #ArtificialIntelligence #DeepLearning #MachineLearning -
AI Cold War And Race For Sovereign AI
By
–
The AI Cold War And The Race For Sovereign AI
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @lexfridman @sama @kaifulee @ID_AA_Carmack @karpathy @2morrowknight @ylecun