Beyond the First Choice: Later Effects Matter More
the second and third order effects of decisions are often more important than the initial decision itself
Non‑generative AI Crucial for National Security and Legal Search
non-generative AI and the training of non-generative AI is integral to our national security. drones don’t generate prose, images or audio. the search and target, exactly what ROSS Intelligence built but for locating the correct passage of verbatim law.
AI Training Relies on Unsettled Fair Use Assumption
base AI models (domestic & foreign) are all trained on copyrighted material. this assumes that training AI is fair use. now we await the judiciary’s confirmation to ensure this is all kosher, which it must be in order for AI...
Fair Use Needed for Courts and Lawyers Using AI
fair use must apply for the judiciary to continue to use AI as well as lawyers. think of this as an LSAT logical reasoning question folks
Predicted Outcome: AI Training May Lose Fair Use
things are playing out exactly as i said they would. if fair use doesn’t apply to training AI in the US, it’s all over
Bet on Patterns You See, Not Overanalyze
i learned to just start betting on the patterns i spot instead of trying to figure that out
License‑Gated Data Centralizes AI, Stifles Startups
large-scale pretraining data is the substrate of AI. if access to this substrate becomes license-gated in the united states but not elsewhere, then foundation-model innovation centralizes, startup entry collapses at the base layer, and long-term technological leadership shifts
China Deploys Latest AI Models in Military, Data Sources Unclear
"China is not using the oldest model. They're using the newest Qwen. They're using the newest Kimi. They're using the newest DeepSeek in their military applications." - AND WHAT DATA DID THEY USE TO TRAIN THESE MODELS?
Major Outlets Reveal Journalists Turning to AI Writing
yesterday both WSJ and WIRED published pieces about journalists using AI to write their stories https://t.co/G8CQwY7lRr
Agent-Powered Startups Will Outpace Incumbents with Speed, Cost, Agility
this means both older incumbents and younger market participants will face an onslaught of agent enabled startups. faster, cheaper, nimbler.
Flexpa Powers the Emerging AI-Driven Healthcare System
.@flexpa is integral to enabling the AI-powered healthcare system currently being built. and we're ready for all of it.
Explore Undersea Cables: Unlock High-Value Market Insight
start learning about the cables under the oceans, you won't regret it. lots of alpha in knowing about the cables under the ocean now
Flexpa Shows Patient‑Agent Future, Decentralization Neutral
“The @Flexpa experience was my big ah-ha. Decentralization isn’t inherently good or bad - it’s more of the current, unchangable state of things.” - @AndreaCoravos 👌 - patient to agent is the future of healthcare and flexpa has built the infra...
China's Open-Source AI Trained on 1.4M arXiv Papers
and how are china's open-source AI trained? here's what deepseek said, on the record: “we collected source code and compiled PDFs from 1.4 million arXiv articles ... we cleaned 860K English and 180K Chinese e-books from anna’s archive (anna’s archive,...
Old Metrics Still Rule Media Success
same as it always was though: neilson ratings, top 40 charts, new york times best sellers list etc.
Model T Succeeded After Nine Failed Ford Prototypes
“success did not come easy. the model T was ford's ninth attempt at producing a profitable automobile, after models A, B, C, E, N, R, S, and K all proved relative failures.” https://t.co/afusdFZddJ
America's WWII Industrial Mobilization: The Untold Story
“how this remarkable mobilization of american industry, technology, and material production happened remains the great untold story of world war II.” https://t.co/uLEMiAoxFI
Check Citations on Westlaw to Dodge AI Hallucinations
“And it’s not even hard to avoid citing AI-hallucinated cases. All you have to do is use Westlaw or Lexis to cite check the brief.” 👀 - a fish in water never realizes it’s wet. legal data is born free and...
AI Legality Hinges on Output Reproduction, Not Learning
if learning from existing media makes AI outputs illegitimate, then learning from existing media is illegitimate. copyright doesn’t ban learning from works, it bans reproducing them. our neurons make brain copies. the real issue is whether outputs reproduce protected expression.
America Must Lead AI to Safeguard National Security
the US relies on commercial AI models for the department of war. there is no AI manhattan project. this means we must ensure US AI models are the best in the world -- our national security depends on it
Free Online Access to Laws Solves the Mess
what if i told you there’s a relatively simple fix to this mess and it involves ensuring the law is easily available and free to all citizens on the internet?
Swarm Drones Use Simple Neural Nets to Track Warm Targets
“optically-coordinated swarming drones with simple neural nets, that simply target anything warm and moving.” - are you paying attention yet?
Living in the Past or Fantasy, Not Present Reality
“they are no longer capable of seeing contemporary reality. the world which they see is not the one that really exists. it is either the one that used to exist or the one that they would wish to have exist.”...
Congress Urged to Back Balanced AI Training on Copyrighted Works
“Congress should support a balanced approach that allows for advanced AI training on copyrighted materials” 👀
Administration Says AI Training on Copyrighted Works Is Legal
“the Administration believes that training of AI models on copyrighted material does not violate copyright laws” 👀
Instant QR Code Health Record Sharing by Flexpa
This is @Flexpa. @Flexpa enables patients to instantly share verified health records with any provider via a QR code. A big thanks to @USDS for helping enable the vision and mission of Flexpa: Flexible Patient Access. https://t.co/fD0qjuBtWy
From AlexNet to ROSS: My 2014 Journey
me since 2014 after learning about alexnet and founding ROSS at the university of toronto:
U of T CS Department Fuels Modern AI Revolution
all roads of the modern AI revolution lead back to the university of toronto’s computer science department
Copyright Should Foster, Not Hinder, AI Innovation
“‘ROSS was at the forefront of fundamental AI development,’ he said. ‘Copyright must encourage, not stop, such innovative efforts.’” 👀
Nearly All Students Use AI; Policy Urgency Looms
“95% of students report using AI in at least one way” 👀 - we’re going to have to figure out fair use and training AI quickly if this is all going to work out
Fair Use in AI Training Critical for National Security
“ROSS Intelligence Inc. told an appeals court that new filings by Anthropic PBC against the Defense Department confirm that finding fair use in AI training is "paramount" to national security.” 👀
Chinese‑linked AI Poised to Surveil U.S. Battlefields
“Companies tied to the CCP are turning AI into a battlefield surveillance tool against America. The threat from China’s technology ecosystem isn’t theoretical, it’s imminent.” 👀
Paywalls Ransom Legal Data, Blocking Citizens' Access
legal data is held for ransom behind paywalls, denying access to citizens who paid for it.
Singularity May Arrive Before Copyright Courts Decide AI Training
we may hit the singularity before courts decide whether or not AI can learn using copyrighted materials and that’s sort of amazing, no?
VPNs Bypass NY Ban on AI Legal Advice
people will use VPNs to access affordable high quality legal advice, healthcare and education from AI. odd to think that we’re here now. sad really
AI Training Fair Use: Defining Our Era's Legal Frontier
whether or not fair use applies to training AI is the fundamental legal question of our age. mark my words.
Tech Promises Affordable Justice, but Access Remains Uneven
so the technology is here to enable better more affordable justice, but not evenly distributed… yet https://t.co/VN4uMD18Qk
Predicted AI Technophobia a Decade Ago; History Rhymes
i anticipated the technophobic pushback ai would receive nearly a decade ago and gave a talk at cliocon about it. history doesn't repeat, it rhymes. watch my comments here: https://t.co/CIm9q25mQ6
AI's Exponential Leap Outpaces Even Seasoned Expectations
"we are now dealing with exponential improvements of ai, astounding even those, like me, who have been in the field for nearly a decade. we knew this would happen but it’s happening quicker than we thought."
US Court Rulings Will Shape AI IP, Economy, Security
we’re at this part of the awareness of the IP issues surrounding training AI and the national economic and security issues that flow from the decisions US courts will make in the coming months and years
Ed Lee Forecasted AI Class-Action Lawsuits in 2023
shout-out to @edleeprof for totally calling the class-action angle for AI lawsuits ahead of the curve in 2023! https://t.co/MwA9eSRAua