Links for 2025-01-07

AI:

1. Microsoft expects to spend $80 billion on AI-enabled data centers in fiscal 2025 — For comparison: In 2021, Russia's military spending was about $65.91 billion. https://blogs.microsoft.com/on-the-issues/2025/01/03/the-golden-opportunity-for-american-ai/

2. Sam Altman: “We are now confident we know how to build AGI as we have traditionally understood it…We are beginning to turn our aim beyond that, to superintelligence in the true sense of the word…We’re pretty confident that in the next few years, everyone will see what we see…” https://www.lesswrong.com/posts/T5p9NEAyrHedC2znD/sam-altman-reflections

3. Sam Altman believes that Trump will be the AGI President, that AGI will be probably developed during his term. He also told Bloomberg that nuclear fusion will work "soon" and he expects that Helion (the company he has invested $500 million in) will demonstrate net-gain fusion soon https://www.bloomberg.com/features/2025-sam-altman-interview/ [no paywall: https://archive.is/Q7Ov7]

4. “We are starting to see pre-AGI systems shrink analyst classes, change personnel strategies, and trigger layoffs. Remember that today is the worst these systems will ever be. You should expect that they will become more capable as time goes on. As they get better, their impact on the labor market will grow rapidly.” https://www.lesswrong.com/posts/Mak2kZuTq8Hpnqyzb/the-intelligence-curse

5. Microsoft's Charles Lamanna: "By this time next year, you'll have a team of [AI] agents working for you" https://www.fastcompany.com/91254053/25-experts-predict-how-ai-will-change-business-and-life-in-2025

6. Google demonstrates scalability of AI agents using LangChain and Vertex AI, enabling complex workflow automation. https://www.kaggle.com/whitepaper-agents

7. "AI agents is likely to be a multi-trillion dollar opportunity"— Jensen Huang, Nvidia CEO https://www.youtube.com/live/k82RwXqZHY8?si=lVdJlwg51hKwIyU_&t=2638

8. What Indicators Should We Watch to Disambiguate AGI Timelines? https://www.lesswrong.com/posts/auGYErf5QqiTihTsJ/what-indicators-should-we-watch-to-disambiguate-agi

9. Language agents backed by open-source, non-frontier LLMs can match and exceed both frontier LLM agents and human experts on multiple scientific tasks at up to 100x lower inference cost. https://arxiv.org/abs/2412.21154

10. B-STaR: Monitoring and Balancing Exploration and Exploitation in Self-Taught Reasoners https://arxiv.org/abs/2412.17256

11. Human-like Bots for Tactical Shooters Using Compute-Efficient Sensors https://arxiv.org/abs/2501.00078

12. Hypothetical Minds: A LLM agent for multi-agent settings that generates hypotheses about other agents' latent states in natural language, adapting to diverse agents across collaborative, competitive, and mixed-motive domains https://arxiv.org/abs/2407.07086

13. Given sufficient context, LLMs can suddenly shift from their concept representations to 'in-context representations' that align with the task structure https://arxiv.org/abs/2501.00070

14. “We achieved a click-through rate of above 50% for our AI-generated phishing emails.” https://www.lesswrong.com/posts/GCHyDKfPXa5qsG2cP/human-study-on-ai-spear-phishing-campaigns

15. Transformer takes 22 seconds of brain state vectors, and outputs the next 5 seconds of human neural activity with decent accuracy. Also, functional connectivity emerges naturally from the predicted brain states https://www.arxiv.org/abs/2412.19814

16. Generating All-Atom Protein Structure from Sequence-Only Training Data https://amyxlu.github.io/plaid/

17. A metagenomic foundation model to help detect and prevent the next pandemic early. Open-source, decentralized AI can help differentially accelerate scientific progress to safeguard against risks like pandemics while deepening our understanding of nature. [PDF] https://metagene.ai/metagene-1-paper.pdf

18. Inarix is using open source AI models from Meta FAIR to turn smartphones into pocket laboratories for farmers. Building a foundational model on top of DINOv2, the platform enables farmers to assess crop value in real time https://ai.meta.com/blog/inarix-agricultural-supply-chain-meta-dino-v2/

19. John Deere says its self-driving tractors and trucks will help address labor shortages https://www.theverge.com/2025/1/6/24334357/john-deere-autonomous-tractor-truck-orchard-mow-ces

20. NVIDIA Cosmos: World Foundation Model Platform for Physical AI https://github.com/NVIDIA/Cosmos

21. Google: "we believe scaling on video and multimodal data is on the critical path to artificial general intelligence" https://techcrunch.com/2025/01/06/google-is-forming-a-new-team-to-build-ai-that-can-simulate-the-physical-world/

22. Test-time Computing: from System-1 Thinking to System-2 Thinking https://arxiv.org/abs/2501.02497

23. Proactive Agents for Multi-Turn Text-to-Image Generation under Uncertainty https://github.com/google-deepmind/proactive_t2i_agents

24. VideoRefer Suite: Advancing Spatial-Temporal Object Understanding with Video LLM https://github.com/DAMO-NLP-SG/VideoRefer

25. VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interaction https://arxiv.org/abs/2501.01957

26. Metadata Conditioning Accelerates Language Model Pre-training https://arxiv.org/abs/2501.01956

27. 2.5 Years in Class: A Multimodal Textbook for Vision-Language Pretraining https://arxiv.org/abs/2501.00958

https://x.com/jam3scampbell/status/1876059641404400113

“This is a joke, but you will soon see this ALL THE TIME in higher-level programs: calling out to an AI model mid-execution.” https://x.com/pbwinston/status/1876403446565544213

Health:

1. The mission to restore sight takes a big leap forward! 🚀 🌟 Large-scale RF mapping without visual input for neuroprostheses in macaque and human visual cortex https://www.medrxiv.org/content/10.1101/2024.12.22.24319047v1.full

2. Transplanting young Hematopoietic Stem Cells (or CD150 low) into old mice rejuvenates their blood's epigenetic age and boosts physical performance. Improved muscle strength and even cognitive function. https://www.nature.com/articles/s41422-024-01057-5

Tech:

1. Why China Is Building a Thorium Molten-Salt Reactor https://spectrum.ieee.org/chinas-thorium-molten-salt-reactor

2. Storing Thousands of Terabytes in a Single Gram of DNA https://www.nature.com/articles/s41586-024-08040-5

Math:

1. “I have two children, at least one of whom is a boy born on a day that I'll tell you in 5 minutes. What is the chance that both are boys, and what will the chance be after I tell you the day?” https://www.lesswrong.com/posts/7i4qTDCxf5QBYWqvg/practicing-bayesian-epistemology-with-two-boys-probability

2. How can it be feasible to find proofs? What is it about [mathematics] that makes it tractable? And how can we computationally formalize the kinds of tactics that humans use when attempting to prove things? https://drive.google.com/file/d/1-FFa6nMVg18m1zPtoAQrFalwpx2YaGK4/view

3. The subset parity learning problem: much more than you wanted to know https://www.lesswrong.com/posts/Mcrfi3DBJBzfoLctA/the-subset-parity-learning-problem-much-more-than-you-wanted

Science:

1. Why is the Speed of Light So Fast? Argues that *any* living creature in our universe, natural or artificial, will inevitably view light's speed in empty space as extremely quick. https://profmattstrassler.com/2024/10/03/why-is-the-speed-of-light-so-fast-part-2/

2. “The penguins nodded off >10,000 times per day, engaging in bouts of bihemispheric and unihemispheric slow-wave sleep lasting on average only 4 seconds, but resulting in the accumulation of >11 hours of sleep for each hemisphere.” https://www.mpg.de/21169426/1127-orni-penguins-nesting-in-a-dangerous-environment-obtain-large-quantities-of-sleep-via-seconds-long-microsleeps-154562-x

Miscellaneous:

1. This paper quantified the importance of venture capital in America: "92% of R&D spending and patent value" comes from "VC-backed companies" https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2681841

2. "First Strike" from 1979, showing a dramatic failed response to a Soviet ICBM strike, followed by arguments for increased Air Force spending https://www.youtube.com/watch?v=7lJ3mMzLZuY

Ukraine:

1. Volodymyr Zelenskyy: Ukraine, War, Peace, Putin, Trump, NATO, and Freedom | Lex Fridman Podcast https://lexfridman.com/volodymyr-zelenskyy-transcript

2. Lex Fridman asks Zelensky: “Can’t Ukrainians just forgive Russia?” https://x.com/DrewPavlou/status/1876043497545121849

3. “While the fighting has naturally tied Ukrainian forces, Russia has also brought quality units to the area, such as VDV. They've taken heavy losses while advancing relatively little this winter. Ukrainians likely want this to continue, which can explain the latest developments.” https://x.com/emilkastehelmi/status/1875884909677654264

4. “A long thread on the war and the current situation. Although the worst-case scenarios didn’t materialize in 2024, it was the most difficult period since spring 2022. There were positive developments, and bright spots, but the current trajectory is negative.” https://x.com/KofmanMichael/status/1875568044648583213

5. “Foreign support has often determined the fate of nations fighting larger enemies. The American Revolution succeeded with the help of French aid, the Soviet Union defeated Nazi Germany with U.S. Lend-Lease, and Ukraine’s survival in this war depends on the Western support” https://x.com/Tatarigami_UA/status/1875597784038175225

6. FPV ambush: Ukrainian recon eliminated the chief of staff of the Russian battalion "Storm-Ossetia." https://x.com/NOELreports/status/1875863505028731220

7. In what appears to be a first, Ukrainian USVs launched FPV drones from the Black Sea, scoring hits on a pair of Russian Pantsirs https://x.com/Osinttechnical/status/1876318974092443880

8. “Ukraine's "Sych" drones are striking Russian military targets. HUR's "Partisan" group uses strike and reconnaissance drones to detect and instantly destroy key Russian assets, including ammo and fuel depots, command posts, troop concentrations, communication and surveillance equipment, military vehicles, and electrical substations.” https://x.com/NOELreports/status/1875882898500325590

9. US-supplied M1126 Stryker (ICV) of the Ukrainian Air Assault Forces, which has run out of ammunition, crushes Russian infantry in the Kursk Region https://x.com/GloOouD/status/1876376661345898498