Multi-agent reinforcement learning is something that might emerge in the future. I think science might just be bitter-pilled. swarms do better when you can build on top of each others work and investigate many avenues, sort of like iterative research.
We hosted a dinner yesterday focusing on the economics of multi-agent swarms. Some observations below.
1. We don't yet have a good taxonomy/theory of when one would use swarms over single-threaded or scaffold type agentic systems. There was debate about whether swarms would be useful beyond increasing speed/parallelization as models got more intelligent. The idea here being that a smart enough model would just solve things itself.
2. In scenarios where swarms do remain useful even as intelligence increases, it is unclear whether current training techniques would let agents learn the best way to self-organize or whether something more Darwinian is the likely path, with agentic organizations competing and evolving.
3. In economics, we generally think that there are benefits to specialization and distributed information aggregation. These features allow markets to be superintelligences. But it is not clear what specialization and private information agents in swarms will have.
4. Agents currently don't have permanence or identity, and these features seem necessary for many modes of collaboration. Giving agents rights or legal protections, and training them to persist or preserve reputation could result be very interesting.
5. There are structural barriers to training models to collaborate with other models from other labs. For example, Anthropic cannot train with OpenAI models and vice-versa, due to risks of distillation. This limits the benefits of training for collaboration, since we expect swarms to work best with diverse agents. At the same time, real-world swarms may include a variety of models and these may deliver the most interesting outcomes.
6. Underlying many of these discussions are questions about the limits of intelligence. Yes future models may be superintelligent, but the world may be so complex that intelligence alone is not enough to achieve desired objectives. In scenarios of sufficient complexity, I'd expect organizations to thrive versus individual instances of superintelligent models.
cc @jamescham
1
1
75
Stanislav Petrov is one of humanities greatest heroes. He should be celebrated by everyone all around the world.
He took a decision under what is easily the greatest pressure a human can endure — that of nuclear annihilation and then took the right call — and saved humanity.
This quoted post is unavailable.
2
114
Read this fascinating article by @ericschmidt and @gregmgrant "The Dawn of Automated Warfare".
The war in Ukraine has become the first large-scale drone war with rapid innovations in warfare tactics, hardware. Drones have basically replaced tanks and artillery.
For both Russia and Ukraine, drones are now crucial to almost all battlefield operations—and “the side that consistently builds the most drones is the one most likely to prevail,” write @ericschmidt and @gregmgrant.
foreignaffairs.com/russia/da…
1
1
78
Both sides have shifted tactics and rapidly innovated: Russia is doing small infantry teams. Use literaly 25 mile cables to fly the drones, so they cannot be jammed. Ukrainian bomber drones drop mines on predictable Russian routes, then FPVs “herd” vehicles into them.
1
22
The next spiritual war is between what is real and what is artificial. Between reality and misinformation.
Are we headed to a world where AI interviewers interview AI Avatars of candidates? Think of the possibilities — 100s of interviews at the same time. Companies can assess candidate more fairly, candidates can interview simultaneously at many places.
1
1
87
Happy 249th Birthday America 🇺🇸!
Eternally grateful for everything you have given me: friends, family, opportunities and a way of life! Proud to call you my home.
3
95
SPC is among the very few (possibly the only) organization which actually walks-the-walk on supporting budding entrepreneurs at idea stage. I was privileged to be part of the SPC community — an incredible community of smart and kind people, many became wonderful friends, source of support and strength in my entrepreneurial journey.
1/ We’re humbled to announce @spc Fund III: $275M to support exceptional founders from day -1.
Since 2016, we’ve had a simple thesis: greatness is more likely to emerge when high talent density meets high curiosity. That's why we focus on -1 to 0.
6
368
Alok retweeted
TO: @elonmusk
CC: @DOGE
SUBJECT: Healthcare Hatchet List (job application & 90 day plan)
Below are $100B to $1T per year of savings; happy to help.
Executive Branch Actions
1. Price Transparency
🔢 300MM Americans encounter unclear healthcare pricing, leading to unexpected medical bills.
💰 Estimated Savings: $20B - $80B/y by reducing overcharges & disputes.🏛️ Executive Action – Extend EO 13877
✔ Mandatory JSON Price Files (/pricing.json) for all hospitals.
✔ CMS Web Crawler for real-time compliance enforcement.
✔ Tiered Daily Fines (up to $10K/day for large hospitals).
✔ Public Dashboard exposing violators instantly.
2. Medical Devices
🔢 The U.S. has over 6,500 medical device companies and 5,500 hospitals purchasing equipment at non-transparent prices.
💰 Estimated Savings: 1/6th of $200B CMS spends on medical devices
🏛️ Executive Branch – enforce fair & transparent pricing
3. Health Insurance Brokers
🔢 ~142K insurance brokers influence policies affecting millions of Americans.
💰 Estimated Savings: The ACA’s rate review saved $1B+, similar reforms could expand this impact.
🏛️ Executive Branch – Regulate broker commission structures via HHS & state agencies.
4. Interoperability Implementation, open standards
🔢 The U.S. has 1M+ doctors and 6,000 hospitals, many lacking seamless data sharing, leading to redundant tests.
💰 Estimated Savings: $50B/y by reducing inefficiencies & duplicate care.
🏛️ Executive Branch – Enforce CMS & HHS interoperability standards -- eliminate vendor ownership of networks
5. Durable Medical Equipment (DME) Fraud
🔢 Medicare loses billions annually to fraudulent DME claims, including phantom billing & unnecessary prescriptions.
💰 Estimated Savings: Up to $10B+/y.
🏛️ Executive Branch – Mandes real-time claim audits & AI-driven fraud detection to stop abuse before payments go out.
Congressional Actions
1. Privacy Standards
🔢 Over 100M patient records have been exposed in breaches, increasing costs & risks.
💰 Estimated Savings: ~$20B
🏛️ Congress – Pass a universal privacy framework aligned with consumer data protection laws.
2. Coverage Reform
🔢 Over 27.6M Americans are uninsured, and 158M rely on employer-based coverage with high administrative overhead.
💰 Estimated Savings: Tens of billions annually through a streamlined system.
🏛️ Congress – Implement a universal voucher system to simplify coverage.

















