Weekly Top Picks #93
ChatGPT Pro losses / Nvidia DIGITS / AI art can be good / The year of the robot / Meta AI creators flop / Anthropic $60B / Transformers, visualized / When ChatGPT flips out
The week in AI at a glance
ChatGPT Pro ($200/month) loses money
Digits: Nvidia’s $3,000 personal AI supercomputer
Artists are going mad over this image
Every year will be the year of the robot
Meta AI-generated content creators are… doing well
Anthropic valued at $60 billion
Transformers, visualized
When ChatGPT finally flips out at you
The week in The Algorithmic Bridge (last two weeks)
(PAID) You Must See How Far AI Video Has Come: Google’s Veo 2 obliterates the competition in AI video generation, delivering unmatched quality, physical consistency, and prompt adherence—it’s not a slot machine; it’s an ATM. Compared to Veo 2’s seamless outputs, OpenAI’s Sora Turbo feels rushed.
(FREE) AI Blogger Declares AI Will Take Every Job Except Blogging (About AI): A satirical take on the labor issue from the perspective of the most overconfident group of all; people who work on AI.
(FREE) OpenAI o3 Model Is a Message From the Future: Update All You Think You Know About AI: OpenAI’s o3 shattered expectations, scoring 71.7% on SWE-bench (a 20% leap), placing in the 99.95th percentile on Codeforces, and achieving a 1200% improvement on the toughest math benchmarks. It might not be AGI yet, but o3’s performance sends a clear message: rethink everything you thought you knew about AI.
(PAID) 8 Insights to Make Sense of OpenAI o3: The scaling laws of inference-time compute work / OpenAI o3 is every bit as impressive as it seems / The schism between the AI-rich and the AI-poor / Generative AI is no longer the frontier / Average users don’t need to worry (for now) / No general intelligence is dumb at times / Benchmarks are imperfect metrics for room readers / Some competitors will catch up; most won’t.
(PAID) 20 Predictions For AI in 2025: Geopolitics, bubbles, business wars, and reasoning/agentic breakthroughs of AI.
(FREE): Physics Professors Are Using AI Models as Physics Tutors: Stanford physicist Adam Brown revealed that many top physics professors now use large language models as tutors—a stark contrast to laypeople dismissing them as dumb. Just as chess grandmasters learned from computers, physicists are leveraging AI to explore and expand their understanding. What does this say about our biases—and about the limits of human intelligence?
(FREE): The Algorithmic Bridge in 2024: I don’t review the state of AI (there are better resources for that) or what I published on the Algorithmic Bridge (I did once, almost died of tedium; and if I kill you, who’s going to read it anyway?)
Instead, I’ll look at three stats (everyone loves numbers), revisit three important lessons that marked me, and share three plans for the future.
(FREE): How Meta Plans To Crank the Dopamine Machine With Infinite AI-Generated Content: AI won’t end humanity with killer robots but with brain-hacking entertainment designed to addict us. Meta’s new generative AI creators signal a shift where content isn’t just fed or tailored—it’s created on demand to exploit our pleasure centers, leaving us “plugged to our damn screens all day.” Literally (not figuratively as it is today). As AI models become digital puppeteers, we must ask: are we trading our cognitive freedom for endless dopamine hits or will we fight while we can?