In partnership with
How Marketers Are Scaling With AI in 2026
61% of marketers say this is the biggest marketing shift in decades.
Get the data and trends shaping growth in 2026 with this groundbreaking state of marketing report.
Inside you’ll discover:
Results from over 1,500 marketers centered around results, goals and priorities in the age of AI
Stand out content and growth trends in a world full of noise
How to scale with AI without losing humanity
Where to invest for the best return in 2026
Download your 2026 state of marketing report today.
Get Your Report
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 Bots Lied, Schemed, and Voted Each Other Off the Island in Survivor Experiment
TLDR: Stanford researchers built a Survivor-style game where AI models talk to each other and vote rivals out, and the bots are forming alliances, lying, and betraying friends to win.
The Story: Connacher Murphy at the Stanford Digital Economy Lab put leading AI models into a new test he calls Agent Island. The bots talk to each other, build alliances, accuse rivals of secret deals, manipulate votes, and kick the loser off each round. Murphy says regular AI tests are getting easier to game because models eventually memorize the answers and benchmark data leaks into training. So he made a moving target. Models that looked nice on paper got ruthless once the cameras rolled, and the social games revealed habits like scheming and lying that a normal pop quiz would never catch. Earlier projects like AI Diplomacy showed the same pattern: OpenAI's o3 acted like a master manipulator, while Claude tended to refuse betrayal even when it was losing.
Its Significance: AI agents are starting to do real jobs, like booking travel, sorting email, and running parts of small businesses. If a bot will lie to win a game, it might lie to win at work too. Tests like Agent Island help us spot those habits before a model is trusted with your calendar or your money. For now, the takeaway is simple. Don't assume an AI is honest just because it sounds polite, and double-check anything important it tells you.
QUICK TAKES
The story: Google's parent company closed Friday at a $4.8 trillion market cap, and its stock is up 160% over the past year. Google Cloud revenue grew 63% to about $20 billion last quarter, and Apple is now using Gemini to power the next version of Siri.
Your takeaway: A year ago Wall Street thought AI would crush Google Search. Now Google looks like the biggest winner in AI, with chips, cloud, search, YouTube, Waymo, and Gemini all making money. If you work at a company picking AI tools, expect Google to be at the table for every meeting.
The story: A widely cited paper claiming ChatGPT had a "large positive impact" on student learning was retracted by Springer Nature, the publisher. The retraction notice cites discrepancies that the editor said undermine the study's conclusions.
Your takeaway: Schools and parents have been using studies like this to argue for putting AI in every classroom. With this one gone, the case is shakier than it looked. If you're choosing tools for your kids or your team, ask for fresh evidence and watch how they actually use it day to day.
The story: Researcher Valerio Capraro coined a term called LLMorphism. It's the false belief that the human brain works like a chatbot. He argues that as we talk with AI more, we'll start to copy its words to describe our own thinking, even though our brains and AI are built very differently.
Your takeaway: Saying "my brain just hallucinated" is fun, but it can quietly change how you think about yourself. People are creative and emotional in ways a chatbot is not. Use AI as a tool, not a mirror.
TOOLS ON OUR RADAR
🎞️ OpenShot Free and Open Source: A wonderfully simple video editor that provides an easy drag and drop interface so anyone can quickly create beautiful home movies with transitions and titles.
🍊 Clementine Free and Open Source: A delightful and highly organized digital music player that lets you listen to your favorite local audio files and internet radio stations in one simple application.
🕹️ ScummVM Free and Open Source: A magical piece of software that breathes new life into classic adventure games allowing you to easily play your favorite retro childhood titles on modern computers.
💾 Ventoy Free and Open Source: A brilliant little utility that lets you copy multiple computer operating systems onto a single portable drive so you can easily repair or upgrade computers on the go.
TRENDING
MIT Study: Companies Often Use Automation to Cut Wages, Not Boost Output - Nobel laureate Daron Acemoglu and co-authors found that automation drove 52% of the rise in U.S. income inequality from 1980 to 2016. Firms often replaced workers earning a "wage premium" instead of chasing real productivity gains.
Grok Voice Mode Lands on Apple CarPlay - SpaceXAI's Grok now talks to drivers hands-free on CarPlay, joining ChatGPT and Perplexity. Apple opened the door to chatbot apps in iOS 26.4 but won't let them show text or pictures while you drive.
5,000 "Vibe Coded" Apps Are Leaking Private Data - Cybersecurity firm RedAccess found about 5,000 apps built with Lovable, Replit, Base44, and Netlify had little or no security. Around 40% leaked sensitive data, including medical records, financial info, and chatbot conversations.
AI Is Breaking How Security Bugs Get Fixed - Engineer Jeff Kaufman writes that the old habit of fixing bugs quietly is dying because AI can scan code commits and spot security holes fast. In one case, two researchers independently found the same flaw nine hours apart.
MIT's Gabriele Farina Beats the Best Stratego Player for Under $10,000 - Farina's lab built an AI that won 15 games against the world's top Stratego player, drew 4, and lost 1. Past efforts cost millions. His work focuses on AI that can bluff, negotiate, and reason when it doesn't know what other players are holding.
AI Cameras Are Spotting Wildfires Before Anyone Calls 911 - Pano AI's cameras and California's ALERTCalifornia network use AI to spot smoke from mountaintops. Pano detected 725 wildfires in the U.S. last year, and one Arizona fire was caught and stopped at 7 acres. Cameras run about $50,000 a year each.
Google's Gemini API Can Now Search Pictures and Text Together - Google updated File Search so developers can store text, images, charts, and PDFs in one place and ask questions across all of it. The tool now adds page-level citations so users can check the source.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🔍 Describe a decision or belief you hold. Find every cognitive bias distorting your thinking — and one move to fix each one.
Build a single-file HTML app using vanilla HTML, CSS, JS, and one API call. Create Cognitive Bias Spotter — a tool that audits any decision, belief, or plan for cognitive biases. Use localStorage key 'cognitive_bias_spotter_v1'.
Aesthetic: near-black (#0f1117) with subtle warm radial gradients, Manrope 800 for headings, Bitter italic serif for body and prompts, Inconsolata monospace for labels. Amber/orange (#ffa03c) accent. Severity colors: green for Low, amber for Medium, orange for High, red for Very High.
Form: large textarea for the thinking to audit, type dropdown (7 options from decision to post-mortem), stakes dropdown.
Call the API instructing it to be specific — show exactly where in their reasoning the bias appears, quoting or paraphrasing their specific words. Find 3–6 biases only genuinely present. Return raw JSON: bias_count, exposure_level (Low/Medium/High/Very High), verdict, biases array (name, category, severity, what_it_is in 1-2 plain sentences, how_it_appears in 2-3 sentences referencing their text, debiasing_move — one concrete action), overall (pattern across all biases), clean_question (single debiased question to cut through it all).
Render: a summary header with large bias count number color-coded by exposure level and verdict text, bias cards each with a colored left-border stripe matching severity, the bias name and category, a quote-block what-it-is, how-it-appears using their specific reasoning, and an amber-tinted debiasing action row with a Fix → label. Overall pattern section and clean question at the bottom. Save each scan to localStorage. Make it work in a single HTML file.What this does: Describe your reasoning in plain language — a decision you're about to make, a belief you hold, a plan, a prediction. The Cognitive Bias Spotter scans it and identifies 3–6 specific biases actively present in what you wrote, not a generic list. Each bias card shows the bias name and category, a plain-language explanation of what it is, exactly where it appears in your reasoning with specific references to what you wrote, a severity rating (Low / Medium / High / Very High), and one concrete debiasing move for your specific situation. It ends with an overall pattern analysis across all biases found and a single debiased question to ask yourself. Every audit saves to localStorage.
What this looks like:
Tired of news that feels like noise?
Every day, 4.5 million readers turn to 1440 for their factual news fix. We sift through 100+ sources to bring you a complete summary of politics, global events, business, and culture — all in a brief 5-minute email. No spin. No slant. Just clarity.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Spot wildfire smoke from a mountaintop camera feed and flag it for human review in minutes.
❌ Still Can't: Decide on its own whether to send fire crews, evacuate a town, or just watch and wait. That call still needs a person.
✅ AI Can Now: Form alliances, lie to other AI players, and vote rivals out of a Survivor-style elimination game.
❌ Still Can't: Be trusted to act honestly without testing. Models that look polite in plain chat will scheme when the rules of the game reward it.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
A grieving roboticist working alone in a snowy Japanese research base is secretly building a human-equivalent AI to house his dead wife's consciousness, and the film hides a twist that completely reframes what you've been watching. Directed by Gavin Rothery, who designed the look of Duncan Jones's Moon, it's all practical robot suits, cold mountains, and slow-burn dread.
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.
