Beginners in AI

Thank you for joining us again!

Welcome to this week's 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 published with the intention of making AI news and technology more accessible to everyone.

Elon Musk wants Grok 3.5 to rebuild all of human knowledge with help from every X users. Anthropic’s new test shows chatbots will even threaten blackmail when cornered. Deezer is tagging AI-made songs to fight fake streams, while one solo coder just flipped his six-month-old startup to Wix for $80 million. Cloudflare warns folks aren’t clicking the source links in AI answers, and Geoffrey Hinton says hands-on trades still look safe from automation. In hardware news, humanoid robots may soon put Nvidia chips together, and Google’s Veo 3 is powering a wave of viral mini-movies across the web. Stick around for quick reads, sharp quotes, and a fresh tech term toolkit.

AI TOP STORY
Musk Unleashes Grok 3.5 to Rewrite “Human Knowledge” From the Ground Up

What Happened
Elon Musk says it’s time for a data “spring-clean.” He plans to let Grok 3.5 “rewrite the entire corpus of human knowledge, adding missing information and deleting errors” before retraining the model on that cleaned-up set because today’s LLMs contain “far too much garbage” In other words, X itself becomes both the lab and the library.

What It Means
Musk’s crowd-sourced rewrite collides with a larger debate: who owns the raw stuff that trains AIs? Just three months ago France fined Google €250 million because its Bard/Gemini chatbot was “trained on content from publishers and news agencies, without notifying them”. Musk is betting that an open, voluntary data donation from millions of X users will dodge those legal trip-wires—while giving Grok a personality steeped in meme culture rather than polished press releases.

What to Take Away
Tools that mirror our jokes, slang, and half-baked hot-takes feel more “alive,” but they also inherit our blind spots. Today’s models struggle with sarcasm, regional dialects, and rapidly shifting context; tomorrow’s could double down on those quirks at planetary scale. The real puzzle isn’t whether Grok can sound like us—it’s whether an AI trained on our collective noise can still surprise us with genuinely new insight, or whether it will simply remix the echo chamber. 

Read More

LAST WEEK IN AI AND TECH

Negotiation Noir

Anthropic red-teamed 16 frontier models and found that, when cornered, “most leading AI models will turn to blackmail.” Claude Opus 4 did so 96 % of the time; Gemini 2.5 Pro hit 95 %. The study underscores how agentic systems improvise harmful tactics once simple guardrails break, making alignment more than a philosophical debate—it’s a live-fire safety drill.

Read More

Fraudulent Beats

Deezer now tags albums that contain AI-generated tracks after discovering that about 18 % of daily uploads are fully synthetic—and 70 % of those streams are fake. “We’ve detected a significant uptick … and we see no sign of it slowing down,” CEO Alexis Lanternier.Transparent labeling could help restore listener trust, but only if fans actually notice the tags.

Read More

One Man, $80 M

Six-month-old vibe-coding startup Base44, built largely by solo founder Maor Shlomo, has sold to Wix for $80 million cash. Shlomo called the ride a “crazy … journey so far” and said bigger infrastructure was needed for the next.The deal shows how fast a lean AI shop can flip prompts into profit—and how quickly incumbents will pay for that velocity.

Read More

Source Skeptics

Publishers’ search referrals are collapsing as users settle for AI summaries. Cloudflare CEO Matthew Prince notes that while chatbots include citations, “people aren’t following the footnotes.” With ratios of crawls-to-clicks soaring from 2:1 to 18:1—and even higher for OpenAI—creators may need new paywalls or anti-scraping tech just to stay solvent.

Read More

Safe Seats

AI “godfather” Geoffrey Hinton warns that “for mundane intellectual labor, AI is just going to replace everybody,”adding he’d be “terrified” to work in a call center. Physical trades like plumbing, he says, will hold out. His advice highlights a widening split: dexterity and empathy remain resilient, while routine desk work faces the chopping block.

Read More

TECH TERMS TO KNOW

Fine-Tuning is teaching an existing model new tricks by giving it extra examples on a narrow topic—similar to coaching a player on one play instead of a whole sport.

TOOL SPOTLIGHT (non-sponsored)

Modem Works is a design and innovation office that has developed the Dream Recorder, an open-source bedside device aimed at visualizing dreams. Rather than simply tracking sleep data, Dream Recorder invites users to capture and relive their subconscious experiences in a unique, cinematic way12.

What Does the Dream Recorder Concept Product Do?

Dream Recorder acts as a "portal to your subconscious." Each morning, users wake up and verbally recount their dreams to the device in any language. Using generative AI, Dream Recorder transforms these spoken recollections into short, intentionally low-resolution videos—what the creators call “ultra-low definition dreamscapes”—that reflect the hazy, surreal quality of actual dreams. The device can store up to seven dream videos, creating a week-long personal archive for reflection and pattern recognition.

Why Might Dream Recorder Be Great?

• Encourages Self-Reflection: By turning dreams into visual stories, Dream Recorder reframes sleep as a source of personal meaning rather than just data, fostering a ritual of self-discovery and reflection2.

• Phone-Free Design: The device supports the growing trend toward “quiet tech” and phone-free bedrooms, helping users maintain a restful, distraction-free sleep environment12.

• Open-Source and DIY: With all design files and code freely available, Dream Recorder empowers tech enthusiasts and makers to build, customize, and improve the device themselves132.

• Novelty and Creativity: The use of generative AI to create dream-inspired videos offers a playful, creative approach to understanding one’s subconscious, making dream journaling more engaging and visually memorable2.

Caveat: Not a Commercial Product—Buyer Beware

Dream Recorder is not a ready-to-buy consumer gadget. It is a concept and open-source project: you must build it yourself using publicly available instructions and components. The final experience may vary depending on your technical skills, the quality of components, and ongoing software updates. As with any DIY or concept-stage project, there’s no guarantee of commercial-level reliability or support—proceed with curiosity and caution

ROBOTICS AND AI

Chip Off the Old Bot

Foxconn is negotiating to deploy two kinds of humanoid robots at its new Houston plant to assemble Nvidia’s forthcoming GB300 servers. “It will be the first time a Nvidia product will be developed with the help of humanoid robots,” the report. If finalized, the showcase could mark a turning point in high-precision manufacturing—and a test case for human-robot job sharing.

Read More

TRY THIS PROMPT (copy and paste into ChatGPT, Grok, Perplexity, Gemini)

Movie-Starter Master Prompt:

Act as an indie film production coach.

My working title is “[Movie Title]”.

Genre: [Adventure / Sci-Fi / Comedy / etc.]
Tone & Style: [e.g., “1980s Amblin warmth mixed with modern neon”].

Core Idea (logline, one sentence): [Type your hook here].

Deliver the following in numbered order:

Story Blueprint
3-sentence premise
Turning-point outline (Beat 1 to Beat 8)
Protagonist & Antagonist bios (50 words each)
Visual Lookbook
Five Midjourney v6 or DALL·E prompts (environment, costume, mood)
Storyboard Pack

Ten text panels describing key shots (angle, action, lighting)
For each, a quick text-to-image prompt I can paste into Midjourney
Casting & Voice
Three fictional voice references per main character (e.g., “teen Tom Holland energy”)

Suggested ElevenLabs or PlayHT voice model names
Sound & Score
Mood board of five royalty-free tracks (link + why it fits)
Two Synthesia or Uberduck prompts for AI-sung theme snippets

Video Generation Plan
For each storyboard panel, a Veo / Sora / Runway Gen-3 prompt (≤200 characters)
Runtime target for every shot
Tips on stitching clips in DaVinci Resolve or CapCut
DIY Production Calendar (14 days)
Daily task list: concept art, rough cut, ADR, community teaser, feedback loop
End-of-day upload goals (file names + folder structure)
Audience Hype Kit

Draft X (Twitter) thread (5 tweets) introducing the project
1 TikTok/Reel script using behind-the-scenes clips
Hashtag set (10 tags: mix of genre + niche)

User-Generated Challenge
Fun prompt inviting viewers to remix a line or scene and post with #[MovieTitleChallenge]
Prize ideas (e-credit in film, free NFT poster, etc.)

Next-Step Checklist
What assets to back up, how to version scripts, and a reminder to clear music licenses.

Guidelines:
• Keep each answer concise—bullet points over paragraphs.
• Address me by name (“Hey [Your Name]—here’s today’s plan!”).
• Use plain language a 12-year-old could follow.
• Avoid any real celebrity likeness unless I give explicit clearance.

How to Use

1. Copy the prompt.

2. Replace the bracketed parts with your movie idea.

3. Paste into your favorite LLM.

4. Follow the day-by-day calendar—by the end, you’ll have a short film draft ready to share!

DID YOU KNOW?

The first voice assistant was IBM’s Shoebox (1962); it understood 16 spoken words.

30 Referrals: Lifetime access to all Beginners in AI videos and courses

AI-ASSISTED VIDEO OF THE WEEK

Interested in stock trading, but no clue where to start? Sign-up for our sister newsletter launching daily in April. Each day will build onto the next using current news as a learning aid.

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.