In partnership with
Bloomberg: "No reliable safe havens." Billionaires have been investing elsewhere. Here's how to get in.
Bloomberg's Marcus Ashworth wrote plainly recently: "No more reliable safe havens."
After all, the S&P fell over 7% from the February peak. Bonds, even with less risk, are barely keeping pace with inflation.
So-called "diversified" portfolios have gotten hit from multiple directions.
Meanwhile, the world’s wealthiest have been setting records in another asset class.
Circumstances are always unique, but after the dot-com bust, it grew roughly 24% annually for a decade. After 2008, roughly 11% annually for 12 years.
Blue-chip art.
Why? It trades globally in multiple currencies, has scarce supply, and has shown near-zero correlation to equities since 1995.*
With Masterworks, 70,000+ investors allocated $1.3B fractionally across 500+ artworks featuring Banksy, Basquiat, and Picasso.
Accredited? You can invest in a diversified portfolio of postwar and contemporary art alongside two other real assets. From 2017-2025, the mix would’ve beat the S&P 500 by 3.1x.
See if you can improve your portfolio performance all in one diversified strategy.
*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. Important Reg A disclosures: masterworks.com/cd
Beginners in AI
Good morning and thank you for joining us again!
Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.
THE FRONT PAGE
Google Just Launched an AI That Works Like a Whole Research Team, Not a Chatbot
TLDR: Google launched Co-Scientist, an AI that works like a whole research team and has already helped find a possible treatment that blocked 91% of liver scarring in lab tests.
The Story:
Google rolled out Co-Scientist, an AI built to act like a team of scientists, not a single chatbot. It runs in three stages. First, it reads through millions of research papers and comes up with possible ideas. Then different AI "agents" argue over those ideas, hunting for weak spots and missing evidence. Finally, the system picks the strongest ones and hands them to human researchers to test. Early results published in Nature show Co-Scientist spotted an old drug that blocked 91% of a scarring response in liver cells in lab tests, and it figured out how bacteria share DNA, a puzzle that took human scientists about a decade to crack.
Its Significance:
This is a big trend in what AI is being applied to and a major focus of Google. Most AI right now writes emails, makes images, or helps you code. Co-Scientist tries to do real science. If it works the way Google says, drug companies could find new treatments faster, and rare diseases that don't get much research money could finally get attention. Some scientists are worried too. They say AI mostly remixes things people already wrote, so it might miss truly new ideas. This looks like a real step forward, but the proofs are still small, and bigger tests need to happen before we know if it changes medicine in a serious way.
$39 regular. Use code LAUNCH for $10 off. First 50 buyers get it at $29. Free updates for life as more prompts and tools get added. Prompts > skills> automations
QUICK TAKES
The story: A startup called Axiom told Axios that math proofs made by its AI have now been printed in several peer-reviewed academic journals. Axiom's tools write proofs in Lean, a language built for checking math step by step. The company raised $200 million in March and earlier scored a perfect mark on the Putnam, one of the hardest college math tests in the world.
Your takeaway: Peer review is the slow, picky process that decides what counts as real science. Getting AI-made proofs through that bar is a first. New math feeds new tech, from safer self-driving software to better online security, so this matters even if you don't care about equations.
The story: Perplexity open-sourced a security scanner called Bumblebee that checks your computer for bad software packages, browser extensions, and AI tool settings without ever running the code it's looking at. Most scanners trip the trap they're trying to find. This one reads the labels instead, and it's the first to also check the connection files that let AI tools like Claude and Cursor touch your email, files, and calendar.
Your takeaway: If you let AI tools onto your computer, attackers are starting to target those connections. A free, careful scanner that doesn't make things worse is a useful thing to know about. Anyone who works with AI dev tools should bookmark this one.
The story: Los Angeles County Superior Court started a pilot in February letting civil judges use an AI tool called Learned Hand to draft research memos and even tentative rulings. Riverside County signed a $10,000 deal to test it too. The tool uses models from Anthropic, OpenAI, and Google. Current rules only require the court to tell you AI was used if a whole document was written by it, so partial use can stay hidden.
Your takeaway: AI is already touching how courts decide real cases, including ones that affect people's freedom. The pilot is in civil court for now, but the contract leaves room for criminal use later. If you ever go to court in California, you may not know if a machine helped write the decision.
TOOLS ON OUR RADAR
♟️ Lichess Free and Open Source: A phenomenal digital chess server completely free of any advertisements allowing you to play unlimited matches and learn strategies with millions of players.
⏱️ ActivityWatch Free and Open Source: An automated time tracker that securely monitors your daily computer usage helping you understand your habits and improve your personal productivity.
🏥 CorsixTH Free and Open Source: A hilarious management simulation game where you build complex hospitals hire comical doctors and cure completely ridiculous fictional diseases.
🎬 OpenToonz Free and Open Source: A two dimensional animation software originally developed for massive professional studios allowing absolutely anyone to create stunning traditional cartoons.
TRENDING
Famous Hacker Says AI Coding Agents Will Be a "Costly Mistake" - George Hotz, the guy who cracked the first iPhone, spent six months testing AI coding agents and says they make code that's broken in ways that are hard to spot. He thinks big companies pushing AI tools on every engineer will see software quality drop fast.
MIT Tech Review: The AI Jobs Panic Doesn't Match the Data - A former head of the Bureau of Labor Statistics says AI's effect on the job market is small so far. The numbers show a stable labor market. Computer science majors are dropping at schools, but most jobs haven't budged yet. Disruption may come, but it's not here.
Uber's COO Says AI Spending Is Getting Hard to Justify - Uber's Andrew Macdonald said the company can't draw a clear line between rising AI costs and better products for users. 95% of Uber engineers use AI tools every month, and the company burned through its full-year Claude Code budget in four months. Other firms like Microsoft and Duolingo are asking the same question.
Fake Journals Are Using Real Professors' Names to Publish AI Papers - Fake academic journals attached real professors' names to AI-written articles without permission. The professors say the mess is a warning sign about how AI is breaking the trust system behind scientific knowledge.
Northwestern Lab Releases Huge Open Protein Dataset for AI Drug Models - Gabriel Rocklin's lab at Northwestern, working with the OpenFold group, released a giant new open dataset on how proteins fold and stay stable. The data is meant to train AI models for designing new biologic drugs, the kind of medicines used for cancer and immune diseases.
Google and OpenAI Are Now Sharing Watermarking Tech - Google announced at I/O 2026 that OpenAI, Kakao, and ElevenLabs are all adopting SynthID, Google's invisible watermark for AI-made images, video, and audio. Over 100 billion images and 60,000 years' worth of audio already carry it. The watermark survives cropping and filtering, so it sticks around when files get shared.
Google Launches "Gemini for Science" With Tools for Researchers - At I/O 2026, Google introduced Gemini for Science, a set of tools for forming hypotheses, scanning piles of papers, and running simulations. It also plugs into more than 30 life science databases, so things like genome analysis that used to take hours can run in minutes.
China Launches Open-Access AI Platform for Global Research - China rolled out a new AI-for-science platform built to help with discoveries in physics, astronomy, and other fields. The platform pushes open innovation and international cooperation, putting China deeper into the global race to use AI for research.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
⚖️ Name two values pulling you in opposite directions. See the real tradeoff and how to live with both.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Tension Map — map two values in tension, show the real tradeoff. Persist to localStorage key 'tension_map_v1'.
Aesthetic: near-black (#0e0e14), warm amber vs cool blue radial gradients on opposite sides, vignette. Fraunces serif italic for body and headings, Inter for UI, IBM Plex Mono for labels. Amber (#dc8c50) for Value A, blue (#50a0dc) for Value B, violet for false-choice card.
Form: Value A input + Value B input split with italic "vs." divider, context textarea, what-you've-tried input.
System instructions to the model: map values tensions honestly. Not all tensions resolve — some are permanent and must be lived with skillfully. Take both sides seriously, refuse to collapse into clever synthesis. Return raw JSON: side_a_cost, side_b_cost (each one sentence), verdict (2-3 sentences on what this tension actually IS), for each side: protects / costs_long / lean (1-2 sentences each), false_choice (name the binary if there is one, or say it's a genuine tension), living_moves array (3 practical specific items), question_to_sit_with (one italic-worthy question).
Render: dark tension viz banner with both value names color-coded (amber italic / blue italic) and their costs in between with ↔ divider, dark verdict card, two-column protects/costs/lean grid color-coded for each side, violet-left-bordered false-choice card, dark "how to live with both" section with numbered italic moves, centered "question to sit with" in italic quotes.What this does: Type two values in tension (Freedom vs. Security, Ambition vs. Presence, Honesty vs. Kindness). Get back what each side protects and costs in the long run, when to lean each way, the false choice you might be making, three concrete moves for living with both, and one italic-worthy question to sit with. Saves to localStorage.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Help find a drug candidate that blocks 91% of a scarring response in lab tests for liver disease (Google Co-Scientist).
❌ Still Can't: Reliably tell the difference between a real and a made-up scientific paper, which is why fake journals using real professors' names are spreading.
✅ AI Can Now: Write math proofs strong enough to pass peer review at academic journals (Axiom).
❌ Still Can't: Justify its own cost at big companies. Uber, Microsoft, and Duolingo can't yet link heavy AI use to better products for users.
RECOMMENDED LISTENING/READING/WATCHING
A woman with a terminal diagnosis pays a company to make a clone so her family won't be alone after she dies, then her diagnosis is reversed and the law says she and her clone have to fight to the death to determine which one gets to live. Director Riley Stearns made this as a deadpan satire of identity, AI replacement anxiety, and the absurd ways legal systems try to handle technology. Karen Gillan plays both versions.
Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!
-James
By the way, this is the link if you liked the content and want to share with a friend.
Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.
Attio - the AI CRM for modern businesses.
Attio is the AI CRM that keeps you ten steps ahead.
Ask Attio anything. Where should I focus? What deals are at risk? Search, update, and create across your customer data.
Ask more from CRM. Ask Attio.