Saturday, October 11, 2025
Happy Saturday!
Welcome to a SPECIAL EDITION of HOW TO // AI, we’re glad you made it.
Here’s what we’ve got for ya:
📝 Turn Your Next Meeting into Action-Ready Minutes
🎓 Top Grads Can’t Beat the Bots
🎸 AI Band Tops Spotify Charts
🌍 Can AI Cool the Planet?
🗣️ Quick Hits: Product Upgrades, Policy & Governance, Content & Discovery
Let’s get started!
📝 Turn Your Next Meeting into Action-Ready Minutes
Work the Model into Your Workflow
A single, well-structured prompt can turn a 45-minute call transcript into a share-ready minutes summary, with decisions and action owners, in under five minutes.
We’ll put the Role + Task + Format + Tone pattern from Weeks 2 and 3 to work and show how one follow-up line polishes the result.
Grab the transcript.
Export text from Zoom, Meet, or Teams. Paste it into a fresh ChatGPT chat so the model has the full conversation.
❓ How to grab a transcript, step-by-step
Zoom, Google Meet, and Microsoft Teams all let you download a text transcript after a call. Here’s the quick path in each:
Zoom – Host stops the recording ► Zoom emails you a link ► click Download transcript (.txt).
Google Meet – In the meeting controls, open Activities → Recording & Transcripts ► click Save; the file lands in the meeting organiser’s Drive folder.
Microsoft Teams – Go to the meeting chat ► click the three dots next to the transcript ► choose Download (.docx).
Open the .txt or .docx file, copy the entire text, and paste it into a brand-new ChatGPT chat.
Add a four-part brief.
Type this below the meeting transcript:
Role: You are an operations assistant
Task: Create concise minutes for a 12-person product stand-up that lasted 45 minutes
Form: Return three sections: Summary (≤ 40 words), Decisions (bullets), Action Items (table: owner, task, due date)
Tone: Use plain English, no jargonPress Send. ChatGPT will draft a short recap and a table you can paste directly into Sheets, Notion, or Confluence.
Verify against the brief.
Line-check the draft:
Are all key decisions listed once?
Does every action include an owner and a deadline?
Do numbers and dates match what you heard?
Note any gaps or errors. Don’t jump to correcting the model yet, just take notes.
Refine with one sentence.
Try to narrow your notes to a single line you can feed back to the model; just one clear instruction that fixes the biggest gap without inviting a rewrite.
Something like,
Shorten the Summary to one sentence under 30 words.Add the sprint ticket number to each action item.Replace ‘ASAP’ with specific due dates in MM/DD format.Move the owner column to the first position in the table.Pick a lone sentence as your follow-up prompt and press Send. ChatGPT will update the entire draft while keeping the structure intact.
Save as a template.
Save the refined prompt, tone line and all, in a file called Project Prompt Template. Next time you have to plan an event, just swap in the new variables (audience, size, timeline, theme) and hit Send.
Export and share!
Copy the action-item table, paste it into your team tracker, and tag each owner. The CSV-friendly format means no extra cleanup.
📌 Key takeaway: State Role, Task, Format, and Tone in the first prompt, then use one focused follow-up to close any gaps. A two-step loop turns raw transcripts into minutes you can share without extra editing.
The Rundown: College graduates face a historic 6.6% unemployment rate over the past year, surpassing the national average as employers increasingly automate entry-level positions.
Details:
Oxford Economics research links this troubling trend directly to rapid AI adoption across multiple industries, particularly affecting roles filled by recent graduates.
Employers now use AI systems to perform basic tasks that traditionally served as career entry points, significantly reducing job opportunities for young professionals.
Experts recommend that graduates develop AI-related skills to remain competitive, as training in artificial intelligence tools can help candidates stand out.
Takeaway: The AI revolution creates a challenging catch-22 for new graduates: the technology eliminates traditional entry-level positions while simultaneously creating demand for workers who already possess AI expertise.
The Rundown: The Velvet Sundown gained over 850,000 Spotify listeners within weeks, but its uncannily perfect promotional images and sound have sparked intense debate about whether humans or algorithms created it.
Details:
Music producer Rick Beato identified "artifacts" in the band's tracks suggesting AI generation, while streaming platform Deezer's detection tool flags 100% of The Velvet Sundown's music as artificially created.
The mysterious band has a Spotify "Verified Artist" badge despite confusing statements about their authenticity.
Human artists express concern about being outcompeted. One musician noted it took him six years to build what this potentially AI band achieved in two weeks.
Takeaway: The Velvet Sundown controversy demonstrates how AI blurs the line between human and machine creativity. Can platforms authenticate genuine artists? Will we value art differently when algorithms can rapidly generate content rivaling human work?
The Rundown: New research from the London School of Economics reveals AI applications in just three industries could cut global greenhouse gas emissions by 3.2-5.4 billion tons annually by 2035, far outweighing AI's carbon footprint.
Details:
AI could transform power generation, meat production, and passenger vehicles by optimizing renewable energy, improving plant-based proteins, and enhancing mobility systems.
The technology's climate benefits substantially exceed its carbon costs, as AI data centers are expected to produce only 0.4-1.6 billion tonnes of annual emissions.
Researchers emphasize that government intervention is crucial, calling for "an active state" to ensure AI benefits are equitably distributed through green research incentives.
Takeaway: While climate tech discussions often focus on hardware solutions, this research suggests AI's most powerful climate contribution may be its ability to intelligently optimize existing systems, though the potential only materializes if policymakers comply.
Product Upgrades 🚀
👓 AI glasses projected live subtitles onto the lenses for hard-of-hearing users, with real-time transcription, speaker labels, and even emotion tags.
🌱 AI-powered farm robots replaced herbicides and laborers in California fields, using solar energy and onboard vision to pluck weeds without chemicals.
🤖 China hosted its first AI robot football match, with humanoid bots making real-time decisions and scoring goals without human control.
Policy & Governance 🏛️
🌫️ xAI secured an air permit to run 15 gas turbines in Memphis despite protests and a pending lawsuit over Clean Air Act violations in a historically polluted neighborhood.
🇪🇺 The EU said it would enforce its AI Act on schedule, rejecting tech industry pressure to delay. Strict compliance rules will take effect by mid-2026.
🏛️ US Senate killed Trump’s AI moratorium in a 99-1 vote, preserving state power to regulate AI and protecting IP rights for creators.
Content & Discovery 📚
💰 CoreWeave agreed to buy Core Scientific in a $9 billion deal, turning a crypto mining giant into an infrastructure ready to meet computing demand.
🌐 LLMs showed cultural bias in a new MIT study, shifting tone and values depending on whether the prompt was in English or Chinese.
🗣️ Customers still preferred human agents over chatbots. 71% said bots didn’t understand them, and human reps kept users more engaged.
🔊 Transcribe Meetings With AI
Dwayne Brown walks through a simple workflow for bringing tools like Read AI and Supernormal into face-to-face meetings and letting the AI handle transcription and summaries even when everyone else is in the room.
📫 Ask the Inbox
Q: “Our weekly strategy call is pretty chaotic between the jokes, side chatter, and people talking over each other. When I feed the transcript to ChatGPT, it misses half of the real meat of the meeting. How do I get it to focus on what matters?”
A: If you need the model to filter the content you’re feeding it before giving an output, add guardrails:
Decision cue
Tell the model exactly what to listen for:
“Only capture sentences that start with ‘We will,’ ‘Let’s,’ or ‘Action:’ and treat them as decisions or tasks.”
You may have to tailor this to your team’s unique lingo, but this guardrail should filter out banter and keep the action list tight.Noise filter
Give it permission to drop the fluff:
“Ignore any text that includes laughter tags, off-topic jokes, or small talk.”
ChatGPT will skip over the chatter and focus on commitments.
Paste those two lines right after your usual Role-Task-Format-Tone prompt. The model will re-scan the transcript and surface the statements that drive the project forward, with no more missed action items!
📌 Key takeaway: A clear inclusion rule (“capture”) plus an exclusion rule (“ignore”) guides ChatGPT past the noise to make outputs more meaningful.
🤔 Got Questions?
Have a question you want answered? Email [email protected] and you just might be featured in an upcoming issue!
This newsletter is your starter kit for mastering AI with confidence. We keep things simple, show real examples, and focus on quick wins you can repeat.
Stay curious, stick with it, and watch your skills grow week by week!
Until next time :)