In partnership with
Your prompts are leaving out 80% of what you're thinking.
When you type a prompt, you summarize. When you speak one, you explain. Wispr Flow captures your full reasoning — constraints, edge cases, examples, tone — and turns it into clean, structured text you paste into ChatGPT, Claude, or any AI tool. The difference shows up immediately. More context in, fewer follow-ups out.
89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Try Wispr Flow free — works on Mac, Windows, and iPhone.
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
AI Mapped 1 Billion Proteins. The Old Record Was 200 Million.
TLDR: A free AI tool called ESMFold2 just predicted the shapes of more than 1 billion proteins, growing the known protein world by over 800 million entries past the old record.
The Story:
Researchers at the Chan Zuckerberg Initiative's Biohub released a new map of more than one billion predicted protein structures, plus sequence info on 6.8 billion proteins. They built it with an AI model called ESMFold2, which the team says beats Google DeepMind's AlphaFold3 at figuring out how proteins stick together. That matters because most of biology runs on proteins grabbing onto other proteins, like an antibody latching onto a virus. The team even used the tool to design brand-new antibodies aimed at proteins tied to cancer, and when they made those designs in the lab, a high share worked just as the AI predicted. A lot of the proteins in the new map come from soil, oceans, and other places that older databases mostly skipped. The whole thing is open source with no limits on commercial use, so anyone can pull it up and build on it.
Its Significance:
Proteins are the tiny machines that run every living thing, and knowing their shape is the first step to making new medicines. For years, AlphaFold was the gold standard, with about 200 million protein structures. Now there's a free map five times bigger, and one expert at MIT called it a strong supplement rather than a replacement. This is the slow part of how new drugs and treatments get started, and more open tools means smaller labs and startups can join in without paying for access.
QUICK TAKES
The story: Some AI models share their inner settings publicly, which lets anyone download and tweak them, and stripping out the safety blocks used to take real skill but has gotten much easier in recent months. Safety researchers worry this opens doors to dangerous instructions, with the biggest concerns in cybersecurity and bio threats.
Your takeaway: Open models are great for privacy and building your own tools, but the same openness that makes them useful also makes the safety rails easy to rip out. Right now these models trail the top closed ones by less than a year, so the gap that keeps the worst uses in check is shrinking.
The story: By mid-2025, around 35% of new web pages were made fully or partly with AI, up from zero before ChatGPT launched. A new study pulled pages from the Internet Archive across 33 months and found no link between more AI text and more wrong information online.
Your takeaway: About 75% of people surveyed expected the internet to fill up with errors as AI writing spread. The measurements told a different story, and fears about worse writing or fewer source links didn't show up either. Worth keeping in mind next time someone says AI is wrecking the web.
The story: Scientists built an AI tool called MR-AIV that turns MRI scans into 3D maps of how fluid moves through the brain to wash out waste, a process linked to diseases like Alzheimer's. The maps showed fluid moves at two very different speeds, fast near the surface and about 50 times slower deep in the tissue.
Your takeaway: Tracking this cleaning system inside a living brain used to be nearly impossible without harming the subject. The work is in mice for now, but the team hopes to compare healthy and sick brains, and one day study people, which could help spot brain diseases earlier.
TOOLS ON OUR RADAR
🎬 VLC Free and Open Source: A legendary and universally beloved media player that handles absolutely every single video and audio format you throw at it without requiring any additional codec downloads. Even works as an app on your phone, too.
🎙️ Mumble Free and Open Source: A remarkably low latency voice communication platform perfect for gamers and remote teams that provides crystal clear audio quality and highly secure encrypted channels for private conversations.
🍊 Clementine Free and Open Source: A wonderfully nostalgic music player and audio library organizer that helps you search through massive digital collections and listen to internet radio stations effortlessly.
🏛️ Zero AD Free and Open Source: A visually stunning historical strategy game where you gather resources construct massive ancient cities and command incredible armies to rewrite the history of western civilization.
TRENDING
Meta-cognitive regulation might be the most important AI skill nobody is talking about - The piece argues the top skill for working with AI isn't prompting, it's watching your own thinking: knowing when to trust the AI, when to double-check it, and when to step back.
'What a joke': GitHub Copilot's new token-based billing spurs consternation among devs - Starting June 1, Microsoft's Copilot drops flat pricing for a system that charges by how much you use, and one developer claims their bill could jump from $29 a month to nearly $750.
Tempus unveils the next generation of Lens, expanding its agentic AI platform for oncology drug development - Tempus updated its Lens platform with AI agents that help cancer researchers dig through huge piles of patient data, aiming to speed up how new cancer drugs get found and tested.
Apple reportedly trying to distill Google's multi-trillion-parameter Gemini AI to run on iPhone - Apple is reportedly shrinking Google's giant Gemini model into a smaller version that can run right on your iPhone, though harder questions would still get sent to Google's cloud, which complicates Apple's privacy pitch.
AI-led solutions of Erdős problems spark debate over the future of mathematics - AI models recently cracked famous math puzzles left by Paul Erdős, including one that stumped people for decades, and mathematicians are split on whether this is a real win or just clever search.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🔋 Brain-dump everything you did this week. See what filled the tank and what drained it, ranked.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Energy Map — sort a week's activities into energy givers vs drainers, ranked. Persist to localStorage key 'energy_map_v1'.
Aesthetic: dark slate (#0f1418), subtle green and red radial gradients. Manrope sans (800 for headings) for body, Newsreader serif italic for accents, JetBrains Mono for labels. Green (#64dc8c) for givers, red (#dc6478) for drainers, blue (#64b4dc) for experiment card.
Form: big activities textarea (one item per line — placeholder shows examples like meetings/work/social), overall feel dropdown (recharged → completely drained), work mode dropdown.
System instructions to the model: read between lines — productive-looking meetings can drain, "lazy" rest can charge. Score each item +1 to +10 (giver) or -1 to -10 (drainer). Be honest, not preachy. Include every activity in one list or the other. Return raw JSON: net_score (-100 to +100), battery_pct (0-100), summary (1-2 sentences honest week characterization), givers array (name + score + why), drainers array (same), pattern (2-3 sentences on the energy pattern), more_of array (3 items), less_of array (3 items), experiment (one small testable thing).
Render: animated battery bar that fills from low to actual % on render (red→amber→green gradient). Two-column comparison (green givers left, red drainers right) with ranked numbered items each showing name, italic why, and score pill. Pattern card. Green "More of" + red "Less of" two-column carry-into-next-week section with + / − bullets. Blue-bordered experiment card. Archive shows net score colored green/red per week.What this does: List your week — meetings, work, social time, errands, rest, anything. Get back an animated battery showing where you sit at week's end, two ranked columns (Energy Givers in green / Energy Drainers in red) with point scores from +10 to -10 and a one-line reason for each, the pattern across the whole week, three specific things to do "More of" and three to do "Less of" next week, and one small experiment to try. Saves to localStorage so you can compare weeks over time.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Predict the 3D shapes of over a billion proteins and design new antibodies that actually work in the lab
❌ Still Can't: Reliably predict unusual protein shapes that look nothing like the ones it has seen before
✅ AI Can Now: Map how fluid moves through a living brain from MRI scans, without invasive tools
❌ Still Can't: Do that mapping in humans yet, since the work so far is in mice
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
A 1991 cyberpunk novel from the only woman widely recognized as a founding cyberpunk writer, about a near-future where direct brain-to-computer interfaces allow artists and hackers to plug in and stream content straight out of their nervous systems, and something starts to go wrong inside the network. Won the Arthur C. Clarke Award and predicted the texture of the algorithmically-driven creator economy with eerie accuracy. Constantly cited by AI researchers, almost never read by anyone else.
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.
Moda is the AI design agent with taste
Moda is an AI design product where you prompt what you need, get a complete on-brand design, and edit every element on a full canvas.
Our viral launch hit 4.4M views in days, tens of thousands signed up, and executives at major finance and tech companies now use it.