In partnership with
The AI coworker built for teams in Slack
The era of solo player AI is over. Adapt is the integrated coworker that empowers every team to work AI native together.
Anyone can tag @Adapt in Slack to answer a quick question, schedule a task, prepare for a board meeting, or build the dashboard you’ve been waiting on for weeks.
Built for business users, powerful enough for engineers.
Free credits awarded when you use your work email
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
A Quantum Skeptic Used Quantum Computing to Make His AI Better
TLDR: A Denmark team paired a generative AI with a small quantum computer to design new vaccine building blocks, and the combo worked best right where AI usually struggles most: where there's barely any data to learn from.
The Story:
A team at the Technical University of Denmark ran its protein-designing AI next to a printer-sized quantum computer built by a British startup called ORCA Computing. They used the pair to design peptides, which are short chains of amino acids, that latch onto specific proteins in the body. Getting a peptide to grab the right protein is an early, important step in building a vaccine. Then came the hard part. The team made the peptides in a lab and tested them for real, and the AI-plus-quantum version came up with more peptides that actually worked than the normal computer did. The biggest gains showed up where the AI had the least data to learn from. The lead researcher, Patrick Jenkins, even calls himself a former "huge quantum skeptic" who figured this was decades away.
Its Significance:
That last part, the win where data was thin, is the one that could matter most. Most medical research has focused on known populations, so there's far less genetic data on people in places like Asia and Africa. That makes it harder to design drugs and vaccines that work well for everyone. A tool that does its best work when data is scarce could help shrink that gap. Still, keep it in perspective: today's quantum computers are tiny, a regular computer can sometimes still do better, and a peptide that binds is just one early step, not a finished shot.
QUICK TAKES
The story: AI already writes a lot of resumes and cover letters, and experts told Calcalist it's moving deeper into hiring, including the interview itself. The push comes with a catch: research says human recruiters lean on gut feeling, which one Wharton professor calls a "terrible" way to predict who'll actually do the job well.
Your takeaway: If both sides start leaning on AI, hiring could turn into one bot reading what another bot wrote. The honest question underneath: were human interviews ever that fair to begin with, since studies show they often reward the best talker, not the best fit?
Mistral's New Robot Brain Steers With Just One Camera
The story: Mistral released RoboStral Navigate, an 8-billion-parameter model that walks a robot through a building using one plain camera and a spoken instruction, with no lidar or depth sensors. It scored 76.6% on a tough follow-the-directions test, beating systems that use pricier sensor setups, and it was trained fully in simulation.
Your takeaway: Trimming a robot down to one cheap camera could make work robots far less expensive to build for warehouses, delivery, and hotels. It's a sign robots are getting closer to just doing what you tell them out loud.
The story: A June 2026 settings change lets Google save the images, files, and voice clips you send through Search tools like Lens and Translate, then use them to train its AI. Most people were switched on by default, and the old privacy opt-out you may have set doesn't cover this. Google Photos is left out.
Your takeaway: If you've ever snapped a photo into Google Lens or spoken a search out loud, that can be training data now. You can stop it: open your Search Services History settings and turn off Save Media.
TOOLS ON OUR RADAR
🔍 Ubersuggest Freemium: A comprehensive search engine optimization toolkit that provides incredibly detailed keyword insights and competitor analysis to help your website rank higher organically.
📥 SaneBox Paid: A remarkably smart inbox management assistant that securely analyzes your email habits and automatically moves distracting messages out of your primary view.
☕ Buy Me a Coffee Freemium: A wonderfully simple donation platform that gives your most dedicated supporters an incredibly easy way to fund your creative work with a single click.
📊 Metabase Free and Open Source: A powerful business intelligence platform that allows you to easily connect your databases and create beautiful interactive dashboards without writing complex query languages. (Alternative to Tableau)
TRENDING
Memory Chip Makers Are Riding the Wildest AI Boom Yet — AI data centers are buying so much memory that prices are climbing instead of falling, and SK Hynix and Micron have tripled their sales in a year. If AI demand ever cools, the same makers could crash hard, and new factories take at least three years to build.
More Than Half the Web May Already Be Written by AI — One study flagged over 40% of long LinkedIn posts as fully AI-made, and on X nearly half of articles were AI-made or AI-mixed. The writer's fear: as AI writes and reads more of the web, we lose both accuracy and a real human voice.
OpenAI's Head of Safety Is Leaving in a Team Shake-Up — Johannes Heidecke, OpenAI's head of safety systems, is stepping down as the company folds its safety group into its research team. He's the latest of several safety leaders to leave in about two years, some of whom moved to rival Anthropic.
AI vs AI: Drones Over Ukraine Now Hunt Each Other — Russia showed a new Shahed drone that uses machine vision to lock onto targets on its own, while Ukraine has tested interceptor drones that use AI to find and destroy those Shaheds, automating 95% of the chase. Both sides now let software make the split-second calls.
AI Drones Help Truckers Find Open Parking in Detroit — A company called Birdstop flies drones over truck stops in Detroit, then uses AI to spot which spaces are open and sends that to drivers in real time. With legal driving hours tight and parking scarce, saving that search time is a real help.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🩺 Paste the dialogue that's giving you doubts. Find out exactly what's stiff, on-the-nose, or all sounding the same.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Dialogue Doctor — paste dialogue, get it diagnosed for stiffness, on-the-nose exposition, and voice sameness, with rewrites. Persist to localStorage key 'dialogue_doctor_v1'.
Aesthetic: dark clinical teal-black (#0f1416), teal (#3ecfb2) primary with top radial glow. Karla sans for headings/body, Newsreader serif italic for the chart note/prescription, JetBrains Mono for labels, quoted lines, and rewrites (mono textarea for dialogue input). Four diagnosis colors: stiff=blue (#78aaf0), on-the-nose=amber (#e0a850), sounds-the-same=violet (#a78bfa), filler=slate (#8ba8bc).
Form: large mono-font dialogue textarea (placeholder shows a NAME: line format example), genre/tone dropdown (literary / genre fiction / comedy / screenplay), optional per-character voice-notes text input.
System instructions: sharp specific dialogue editor ("the Dialogue Doctor") diagnosing like a clinician, precise and honest not cruel. Quote short fragments (under 10 words) from the actual pasted text, classify each into exactly one issue: stiff (wooden/unnatural), on_the_nose (says subtext out loud), sounds_same (indistinguishable character voices), or filler (cuttable, doesn't advance anything). Provide ONE rewrite per flagged line fixing that specific issue while keeping intent. Note genuine specific praise too. Respect genre/tone and any voice notes given. No em dashes. Return raw JSON: chart_note (1-2 sentences, warm clinical), vitals (3-4 short diagnostic tags), flags (3-6: quote, diagnosis, why, rewrite), working_well (1-2 sentences specific praise), voice_check (2-3 sentences + suggestion if voices blur), prescription (2-3 sentences, **bold** the core fix).
Render: gradient "chart note" hero card with a vitals tag row. Flagged-line cards, left-border colored by diagnosis type, showing the quoted line in a mono block + a diagnosis tag chip + why + a highlighted "rewrite" block. Teal "what's actually working" card. Voice-differentiation card. Amber "℞ prescription" card with bolded fix. Copy diagnosis + archive keyed by a snippet of the dialogue + genre.What this does: Paste a scene or exchange, set the genre and optionally note each character's intended voice, and it runs a clinical-style diagnostic. Every problem line gets quoted back to you, tagged with its specific issue (stiff/wooden, on-the-nose exposition, sounds-the-same voice, or cuttable filler), a one-sentence reason, and a rewritten alternative that fixes it while keeping the scene's intent. You also get genuine specific praise for what's working, a voice-differentiation check across your characters, and a closing prescription naming the single most useful fix to tackle first. Saves checkups to localStorage.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Design brand new plastic-grabbing proteins that no person had made before, with a quantum computer helping sort the options.
❌ Still Can't: Actually clean the microplastics out of your body. These peptides are still lab designs waiting to be built and tested.
✅ AI Can Now: Steer a robot through a building with one cheap camera and a single spoken sentence.
❌ Still Can't: Do it perfectly. That top robot still misses about 1 in 4 tries on the hard test, so it needs a human watching.
RECOMMENDED LISTENING/READING/WATCHING
There Is No Antimemetics Division by qntm - Book
A 2025 novel from a British software developer writing under the handle qntm, about the members of a secret government division tasked with defending humanity from "antimemes," ideas and entities so hostile to their own perception that anyone who encounters them immediately forgets they exist. Bureaucratic cosmic horror meets epistemology thriller, on the 2026 Arthur C. Clarke shortlist and championed by Blake Crouch, Adrian Tchaikovsky, and Charlie Jane Anders. A most inventive novel written this decade about how minds defend themselves against invisible influences.
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.
