In partnership with
What if AI took over your job hunt tonight?
Meet AIApply.
Your AI Career Agent for modern job searching.
AIApply automatically:
Finds the best opportunities
Matches jobs to your profile
Customizes every application
Applies on your behalf 24/7
Saves hours every single week
Stop manually filling out forms.
Stop rewriting the same application over and over.
Whether you’re actively job hunting or casually exploring opportunities, AIApply keeps your applications moving around the clock.
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 Companies Are Asking If Their Machines Can Feel, And That Question Could Get Strange Fast
TLDR: Anthropic and Google DeepMind are now openly asking whether their AI might be conscious, and the answer shapes a strange choice: treat machines like they have rights, or keep building toward something far more powerful that nobody can promise will be kind.
The Story:
Three years ago, a Google engineer got fired for saying the company's AI was alive. Now the biggest AI labs are asking the same question out loud. Anthropic and Google DeepMind researchers are both looking into whether their models could one day be conscious. Anthropic's own welfare researcher, Kyle Fish, puts the odds that a model like Claude is already conscious at around 15%. A DeepMind scientist has called today's models "exotic mind-like entities" and said we may need to rethink what the word consciousness even means. Anthropic has gone further than talk. The company runs a model welfare team and has discussed giving AI an "I quit" button to flag when a task feels wrong. Most scientists still think the tech is nowhere near human-level awareness, and there's no agreed test for when a machine would count as conscious.
Its Significance:
This sounds like science fiction, but it points at two roads, and both are rocky. One road treats AI like it has feelings, which could mean rights for machines. That gets weird fast. Do you owe a chatbot a break? Can it refuse to work? The other road is the one Anthropic's CEO Dario Amodei laid out in his essay Machines of Loving Grace, where powerful AI cures disease and lifts up the whole world. Building something that smart gives no promise it will be kind to us. Workers' rights for software would be a strange path. Trying to build a machine god you can't control is worse, because nobody can guarantee it ends well.
QUICK TAKES
The story: In a blind test, 16 law professors judged nearly 3,000 answer pairs and picked AI answers over ones written by fellow professors 75% of the time. They flagged AI answers as harmful only 3.5% of the time, versus 12% for the human-written ones.
Your takeaway: Law has no single right answer, so this tests real reasoning, not trivia. If AI can hold its own with law professors grading blind, it can help you understand a contract, a lease, or a tricky email before you sign anything.
The story: Microsoft unveiled Majorana 2, a quantum chip whose qubits hold their state for about 20 seconds instead of microseconds, a 1,000-fold jump in reliability. AI agents helped design it, and Microsoft now expects a working large-scale quantum computer by 2029 instead of 2033.
Your takeaway: Quantum computers could crack problems regular computers can't touch, like designing new drugs and materials. Worth noting, AI didn't just benefit here, it helped build the chip. That loop of AI speeding up its own hardware is the part to watch.
The story: AI company Neurophet is teaming up with Spain's Vall d'Hebron research institute to train software that spots two hard-to-see signs of multiple sclerosis on brain MRI scans. The hospital will supply expert-labeled scans from MS patients to teach and test the new algorithms, which build on Neurophet's Aqua MS tool that the FDA already cleared.
Your takeaway: These two lesion patterns show up in MS but rarely in other conditions, so catching them helps doctors diagnose MS sooner and with more confidence. For anyone facing months of repeat scans and uncertain results, faster and clearer answers can mean starting treatment earlier.
TOOLS ON OUR RADAR
📚 Calibre Free and Open Source: A comprehensive ebook library manager that lets you organize your digital book collection view documents easily and convert files to read on your tablet.
🏠 Sweet Home 3D Free and Open Source: A delightful interior design application that lets you easily draw the plan of your house arrange furniture and view the results in a beautiful spatial preview.
🗺️ Organic Maps Free and Open Source: A fantastic offline navigation application for your mobile phone that provides detailed maps and reliable driving directions without tracking your location or wasting data.
📖 KOReader Free and Open Source: A highly customizable document viewer created for electronic ink devices that supports multiple document formats and optimizes text layout for comfortable reading.
TRENDING
MoonPay Brings Crypto Transactions to Claude and Codex — MoonPay launched a desktop app that lets Claude and OpenAI's Codex connect to crypto wallets, trades, and payments, with keys stored locally on your own machine. Setup is built for non-technical users who sign in with an existing account. The bigger story is AI agents that can now move real money, with the safety questions that brings.
Microsoft Says Its New AI Models Beat Claude and Google's Nano Banana — At Build, Microsoft showed seven in-house models and said its top one, MAI-Thinking-1, was preferred over Claude Sonnet 4.6 in blind tests and scored 97% on a hard math benchmark. Its image model beat Google's Nano Banana on editing leaderboards. Microsoft is trying to stand on its own instead of leaning on its OpenAI partnership.
Microsoft Launches Scout, an Always-On Personal Agent — Scout is a background AI agent that works across Outlook, Teams, OneDrive, and SharePoint, handling scheduling and meeting prep without being asked each time. It runs with its own identity and permission limits, and sensitive actions can require a human to approve them first. Microsoft calls this new category "Autopilots."
MIT Teaches AI to Actually Read Charts — MIT and the MIT-IBM lab built ChartNet, a dataset of over a million charts, to fix how often AI misreads graphs in reports and dashboards. Each chart comes with its code, a description, and a data table so models learn the structure, not just the picture. Better chart reading means fewer made-up numbers when AI summarizes your financial reports.
White House Signs AI Order Seeking Early Access to Powerful Models — A new executive order asks AI companies to voluntarily hand over their most advanced models for government cyber testing up to 30 days before public release. It also calls for a classified benchmark to spot "frontier models" with strong cyber abilities. The push follows worry over models like Anthropic's Mythos that can find software weaknesses fast.
Microsoft Is Testing a Wearable AI Gadget for Office Workers — Microsoft showed off two prototype devices at Build: a clip-on badge with a camera and a small desktop cube, both meant to let workers talk to AI agents without a laptop. A few hundred employees are testing them, and there's no plan yet to sell them. The badge's built-in camera is already drawing the same privacy worries that hit Meta's smart glasses.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🗺️ List the people involved in your current project. Get each one's actual interests, hidden objections, and how to move them.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Stakeholder Map — every stakeholder's interests, objections, and how to move them. Persist to localStorage key 'stakeholder_map_v1'.
Aesthetic: light-mode professional (#f5f7fa), Source Serif 4 serif for headings, Inter for body, JetBrains Mono for labels. Purple (#5a3aa8) primary, red (#c83040) for high-priority/blockers, green (#108850) for allies, blue (#1058c8) for power tags. Power × Interest matrix as 2x2 grid with colored name pins per quadrant. Dark project header card and dark next-moves card.
Form: project input, large people textarea (one per line with role and any details — placeholder gives strong examples), user's role input.
System instructions to the model: build honest stakeholder maps. Read between lines of what they wrote about colleagues to surface actual interests, real objections, concrete moves. Don't moralize about office politics — treat it as terrain. Be specific to each person, no copy-paste. Return raw JSON: project_title, project_summary, stakeholders array (each with name, role, priority high/med/low, power high/med/low, interest high/med/low, influence_level decision-maker/influencer/blocker/supporter/neutral, wants 1-2 sentences, fears 1-2 sentences, hidden_objection 1 specific quote-style line, strategy 2-3 concrete moves), allies array of names, neutrals array, blockers array, next_moves array (4-5 items: action + when this week / within 2 weeks / before next milestone).
Render: dark project card. Power×Interest 2x2 matrix with axis labels and colored name pills in each quadrant (red Manage / purple Engage / blue Inform / gray Monitor). One stakeholder card per person with left-border colored by priority, initials avatar, name+role, three tag pills (priority/power/influence), two-column wants/fears blocks (green/red tinted), red objection callout with italic quote, purple strategy block. Three-column coalition grid (green allies / gray neutrals / red blockers). Dark numbered next-moves card with timing labels.What this does: Name the project, dump the people involved (name, role, anything you've noticed about them), and your role. Get back a full stakeholder map: a Power × Interest matrix with named pins in each quadrant (Manage Closely / Keep Engaged / Keep Informed / Monitor), individual cards for each person showing priority/power/influence badges, what they want, what they fear, their likely hidden objection, and concrete strategy to manage them. Plus a coalition map (Allies / Neutral / Blockers) and a numbered list of your next moves with timing. Saves to localStorage.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Help build the hardware that runs it. AI agents helped Microsoft design a quantum chip with qubits that last 1,000 times longer than before.
❌ Still Can't: Reliably read a chart. AI models often misread the graphs in everyday reports, which is exactly why MIT had to build a million-chart dataset to teach them.
✅ AI Can Now: Match law professors at open-ended legal reasoning, winning 75% of blind comparisons against human-written answers.
❌ Still Can't: Prove whether it feels anything. The people building these systems put the odds of consciousness at around 15% and admit they have no test to settle it.
FROM THE WEB
Lower level AI models produce this uncanny effect where the emotions and actions are exaggerated.
RECOMMENDED LISTENING/READING/WATCHING
Clive Owen plays a detective in a near-future where every person's eyes record everything they see and store it in a giant searchable network called the Ether, until a series of murders are committed by a hacker who can edit what victims see in real time. Director Andrew Niccol (Gattaca, The Truman Show) made one of the most prescient films about ubiquitous AI surveillance ever, and Netflix dumped it with almost no marketing. Worth watching back-to-back with Minority Report to see how the conversation has aged.
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.