In partnership with
Someone just spent $236,000,000 on a painting. Here’s why it matters for your wallet.
Late last year, a Klimt sold for the highest price ever paid for modern art at auction.
An outlier sure, but it wasn't a fluke. U.S. auction sales grew 23.1% in 2025. The $1-5mm segment even grew 40.8% YoY.
Meanwhile, Apollo’s chief economist Torsten Slok said to expect ‘zero in return in the S&P 500 over the coming decade.’
Each environment is unique, but after dot-com, post war and contemporary art grew about 24% annually for a decade. After 2008, about 11% for 12 years.
It’s also had near-zero correlation with the S&P 500 since ‘95.*
Now, Masterworks lets you invest in shares of artworks featuring legends like Banksy, Basquiat, and Picasso.
$1.3 billion invested across over 500 artworks.
28 sales to date.
Net annualized returns on sold works held 12 months+ like 14.6%, 17.6%, and 17.8%.
Shares can sell quickly, but my subscribers can skip the waitlist:
*Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at 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's AI Search to Change How 8.5 Billion Daily Searches Work
TLDR: Google is rebuilding its iconic search box around AI, and that single change could reshape the open internet more than any AI model release this year.
The Story:
Google announced this week that its homepage search box is getting its biggest makeover in roughly 25 years, according to NPR. The new box still looks like a single line of text, but it expands as you type, accepts videos and pictures, and now blends AI answers with traditional web links. Liz Reid, who runs search at Google, said users are typing longer, more conversational questions, like they're talking to a person. Google is also adding agent features so the search box can do tasks for you over time, like watching for ticket prices or scanning for local events. This comes after a year of AI Overviews sitting at the top of many search results, which already cuts how often people click through to the original websites.
Its Significance:
Google processes 8.5 billion searches a day and still holds about 90% of the global search market, with Bing at roughly 5% and AI chatbots like ChatGPT and Perplexity picking up 2-3% of search-like questions. So when Google rewires its front door, the whole web feels it. Critics are calling this the "Google Zero" moment, where AI summaries replace clicks and starve the websites that fund the internet through ads, including news sites, online stores, and small blogs. Bing has had Copilot built in for a while and saw some gains, but it never broke Google's grip, and now both giants are pointing the same direction.
The same SEO audit program built and used by Beginners in AI. Lifetime updates. Use Memorial Day Weekend discount code MEMORIAL2026 for 15% off until Monday.
QUICK TAKES
The story: Leonardo added a 3D tab inside its app, so you can generate a character or product as a 2D image and then turn it into a 3D model in the same place, with export to .glb files for game engines or 3D printers.
Your takeaway: Indie game makers, product designers, and 3D printing hobbyists used to bounce between three or four tools just to get a usable model. One workflow, one app cuts hours off small projects.
The story: Anthropic and about 50 partner companies used a private model called Claude Mythos Preview to scan major software, including more than 1,000 open-source projects. The model flagged over 10,000 high or critical bugs, and Cloudflare alone found 2,000, with bug-finding rates jumping more than tenfold.
Your takeaway: The same AI skills that worry people on the attack side are now being aimed at defense, and they're working. Banks, cloud providers, and open-source code you probably depend on are getting patched faster because of this work.
The story: Microsoft Research released Fara1.5, a family of open-weight browser agents in 4B, 9B, and 27B parameter sizes. The biggest version scored 72% on the Online-Mind2Web benchmark, beating OpenAI Operator at 58.3% and Google Gemini 2.5 Computer Use at 57.3%. Even the mid-size 9B model beat both.
Your takeaway: Browser agents are the AIs that click, scroll, and type for you on real websites. Microsoft just made the best one free to download and run yourself, which puts pressure on every paid agent product out there.
TOOLS ON OUR RADAR
☁️ Nextcloud Free and Open Source: A powerful personal cloud storage solution that lets you host your own secure file synchronization and document sharing platform without relying on massive technology companies.
✂️ LosslessCut Free and Open Source: A blazingly fast video utility that allows you to instantly trim away bad footage from your digital home movies without sacrificing absolutely any visual quality.
🔐 KeePassXC Free and Open Source: An exceptionally secure digital vault that stores all of your complex passwords safely offline on your own computer keeping your digital identity protected from online hackers.
🎬 Kdenlive Free and Open Source: A fantastic digital video editor designed for everyday creators that provides multiple audio tracks gorgeous visual transitions and powerful color correction tools for your movies.
TRENDING
Perplexity Is Open-Sourcing Bumblebee, Its Internal Security Scanner - Perplexity released Bumblebee on GitHub, a read-only tool that checks Mac and Linux developer machines for risky packages, browser extensions, and AI tool configs during supply-chain attacks.
AI Is Being Used to Resurrect the Voices of Dead Pilots - The National Transportation Safety Board pulled its public docket system offline after users fed an image of a cockpit voice recording into AI tools and recreated the voices of pilots killed in a UPS plane crash. The image was a spectrogram, which was supposed to be safe to share.
Job-Seekers Are Using AI to Apply, and Now Applications All Look the Same - Greenhouse CEO Daniel Chait calls it a "doom loop," where both sides use AI and applications start blending together. Recruiters are now seeing 400% more applications than a few years ago, with some companies using AI screeners to keep up.
China's DeepSeek Makes Its 75% Price Cut on V4-Pro Permanent - DeepSeek announced its flagship V4-Pro model will stay at a quarter of its original API price, putting it at roughly $0.435 per million input tokens. That's 8 to 35 times cheaper than GPT-5.5 or Claude Opus 4.7. This follows yesterday's news about DeepSeek building its own version of Claude Code, and together it's a clear push to pull developers away from Western AI companies.
What Happens When AI Starts Improving Itself? - Santa Clara University professor Yi Fang says AI-built algorithms now outperform human-designed ones, and the gap is widening. His research focuses on keeping these systems aligned with human goals as the "human in the loop" gets pushed out of the loop.
Google I/O Showed That AI Science Is Shifting From Single Tools to Whole Pipelines - Demis Hassabis told the I/O audience we're "standing in the foothills of the singularity" while DeepMind moved focus toward agentic science systems like AI Co-Scientist and AlphaEvolve. John Jumper, who won a Nobel for AlphaFold, is now working on AI coding instead.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
⚙️ Describe how you currently work. Get a clear weekly system that fits your actual life, not someone else's morning routine.
Build a single-file HTML app with vanilla HTML/CSS/JS. Personal Operating System — a weekly system that fits the user's real life. Persist to localStorage key 'personal_os_v1'.
Aesthetic: deep slate (#0f1014), faint teal grid overlay, vignette. Geist sans 800 for headings, Newsreader serif italic for accents and body text, Geist Mono for labels. Teal (#78c8b0) primary accent, amber for morning routine, blue for evening, red for cut list. Section markers using § symbols.
Form: current week textarea, work mode dropdown (7 options), life situation dropdown (caregiver to solo), priorities textarea, what's-not-working input.
System instructions to the model: design systems that fit the user's actual life — be realistic about energy, commitments, and humans. No "wake up at 5am" advice unless they already do. Respect their constraints. Return raw JSON: name (specific not generic), tagline, principles array (3 items), week array (7 days with day/theme/focus/type where type is deep|shallow|rest|mixed), morning_routine and evening_routine arrays (3-5 steps each), rituals array (3 items with when/name/desc — Weekly/Monthly/Quarterly), cut array (3 specific things to remove), first_week (realistic guidance with permission to do less).
Render: dark OS header with v1.0 version badge and name, principles list with § markers, 7-day rhythm grid color-coded by mode type, two-column morning/evening routine cards with numbered steps, anchor rituals in 2-col grid, red-bordered cut card, teal-bordered first-week card.What this does: Tell it your real week, work mode, life situation, what deserves attention, and what's already not working. Get back a named operating system with three operating principles, a 7-day rhythm grid color-coded by mode (deep/shallow/rest/mixed), morning and evening routines, weekly/monthly/quarterly anchor rituals, three things to cut, and a realistic first-week start. Saves to localStorage.
What this looks like:
Hiring in 8 countries shouldn't require 8 different processes
This guide from Deel breaks down how to build one global hiring system. You’ll learn about assessment frameworks that scale, how to do headcount planning across regions, and even intake processes that work everywhere. As HR pros know, hiring in one country is hard enough. So let this free global hiring guide give you the tools you need to avoid global hiring headaches.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Find more than 10,000 critical software bugs across major systems in a single month (Anthropic's Glasswing).
❌ Still Can't: Reliably tell users which sources its answers came from, with critics warning Google's new AI search makes the data trail more opaque, not less.
✅ AI Can Now: Reconstruct human voices from a spectrogram image of audio that was never released as sound.
❌ Still Can't: Replace the human judgment recruiters say they still want in hiring, with most companies still requiring a person to make the final call on every resume.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
A wildly imaginative far-future fairy tale from the author of Mr. Penumbra's 24-Hour Bookstore, set 11,000 years after AI changed everything. It has dragons made of code, talking beavers, gene-engineered wizards, and a sword in a stone, none of which are quite what they appear. The first novel I've seen that genuinely sits with what large language models might mean for the long arc of human storytelling, without ever explaining itself.
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.
