This is the 50th edition of the rebranded newsletter.

(Don't ask me which number it is since I started this experiment as Innovation Dispatch in June 2023.)

We've grown steadily to, at last count, 536. I don't pay for advertising, so that's all you, and I can't tell each and every one of you how appreciated you are!

Your ongoing subscriptions, recommendations, and emails have given me the motivation to keep going.

You even inspired me to explore inverse prompting in greater detail through my first book, Becoming Irreplaceable: How We Evolve in the Age of AI, and several of you have been so kind as to purchase it in print and digital forms.

A lot has changed since we began.

We've seen the AI tools market explode from around 4,770 to 37,555 as of this note. That's 687% growth in a two-year period.

Insane.

The newsletter today is much different than the first edition, and it's my pledge to always make it better for you all. So, if you have any recommendations or requests, please send them my way by emailing me.

Look forward to hearing from all of you!

JUNE 25, 2025

• PromptCraft #6: Rid Yourself of Overused Em-Dashes, Emulate Author Styles, Think Like a Philosopher
• x
• THE HEADLINES
• Tool Spotlight: Which LLM Tells You the Best AI for Every Job

Rid Yourself of Overused Em-Dashes, Emulate Author Styles, Think Like a Philosopher, and many more discussions, tips, and prompts on the craft of prompt engineering

• ​Banishing Em-Dashes & Pinterest-Style Aphorisms – Two simple rules in Custom Instructions (“no contrastive metaphors, no em-dashes”) instantly strip that tell-tale AI voice from GPT’s drafts. reddit.com​
• ​Why GPT Loves Em Dashes—and How to Stop It – Post explains the punctuation overuse comes from training data; adds style constraints that swap dashes for commas or periods. reddit.com​
• ​Author’s Style Emulator Prompt – Feed GPT a sample and scenario; it mirrors Austen, Shakespeare, Hemingway, Angelou—great for themed emails or pastiche. reddit.com​
• ​Nova Alpha AI v3 – Quant/Institutional Trading Prompt – YAML config deploys $10K to options and day-trades, running IV sims, sentiment scans, and legal-risk checks—ambitious sandbox for paper traders. reddit.com​
• ​Argument Dissector – Think Like a Philosopher – Paste any claim; GPT labels premises, logic, fallacies, then offers counter-points—handy for debate prep or critical-thinking drills. reddit.com​
• ​AI Foundation Models & Bio-Weapon Risk – Paper shows GPT-4o, Llama 3.1, Claude 3.5 can walk users through reviving poliovirus, challenging “tacit knowledge” assumptions in biosecurity. reddit.com​
• ​Roof-Estimate Quote via Canva + GPT – A contractor asks how to funnel measurements into GPT and auto-populate a Canva quote; replies suggest CSV-to-Markdown then bulk-create in Canva. reddit.com​
• ​Powerful Note-Taking & Expansion Prompt – Snap a photo of board notes; GPT transcribes, simplifies, and deep-dives—all in one shot, perfect for lectures or workshops. reddit.com​

Research Shows ChatGPT Is Rotting Your Brain, But You Can Stop It

AI assistants are basically the office intern we always wished for: they never sleep, they rarely complain, and they can bang out boilerplate like nobody’s business.

Need a polite-but-firm return-policy email? Bam.

Want a meeting recap that sounds like you actually listened? Done.

The upside is obvious—time saved, friction reduced.

But every shortcut has a toll booth.

​The research shows that when we outsource thinking tasks, our noggins quietly downshift.

Fewer neural circuits fire. Memory retention drops. Participants who wrote essays with AI struggle to remember what they’d “written” minutes later.

It's cognitive credit-card debt.

You swipe the AI for instant gratification, but the interest shows up later: fuzzier recall, weaker critical-thinking muscles, and writing that sounds like it was steamed in the same corporate boiler as everyone else’s.

Why It Matters Beyond the Ivory Tower

You might say, “Cool science, Aric, but I’m just drafting social-media captions—not a doctoral thesis.”

Fair. Yet even in bite-sized content, how you think shapes what you produce:

• Trust-building: Clients can smell a “ChatGPT smoothie” from a mile away. Polished? Sure. Personal? Not so much.
• Skill compound-interest: Your unique phrasing, analogies, and leaps of logic are professional equity. Hand them off too often, and the market value dips.
• Future-proofing: The more AI handles, the more valuable real-time human judgment becomes. Keep those decision-making synapses in shape.

The Principle of 'Lead, Don’t Lag'

So do we toss ChatGPT in the recycle bin? Nope. The trick is leading the AI instead of lagging behind it. Think director vs. autopilot:

1. Draft first, polish second. Scrawl your raw ideas—messy grammar, half-formed metaphors, and all. Then invite ChatGPT to clarify, tighten, or poke holes. You stay the author; the AI is your line editor.
2. Critique mode over ghost mode. Ask, “What’s missing?” or “Challenge my logic.” Force the model to coach you.
3. Voice ownership. If a suggestion feels bland—or worse, “not you”—reject or rewrite it. Authenticity beats AI-smooth every time.
4. Fact-check as sport. Treat every AI-generated claim like a fun scavenger hunt. Verify. Source. Learn something in the process.
5. Designate AI-free reps. Set aside tasks (journaling, strategic brainstorming, coding katas) where you go full analog. I try to do this every night before bed now, but choose a time that works best for you.

Quick Reality Check

• Will AI write faster? Absolutely. But speed isn’t always the KPI that matters.
• Will clients notice AI fingerprints? Increasingly, yes. Uniform voice can erode brand differentiation.
• Will your brain thank you for mixed usage? Unequivocally. Balancing AI help with human sweat keeps cognitive debt in check.

If You Feel Like AI Is Doing Too Much for You and to You, Try This!

1. “Two-Pass Rule” – Write a sloppy pass solo; run a critique prompt (“Highlight vagueness, bias, weak transitions”) on pass #2.
2. Weekly AI-Free Day – Pick one day per week where you draft core content with zero automation. Compare clarity after a month.
3. Hyper-Specific Prompts – Instead of “Rewrite this,” ask: “Shorten sentences, keep my punchy tone, preserve this joke, or keep the full fidelity of my original work.” The more constraints you give, the more you steer.
4. Source Buddy System – Create a bookmark folder titled “Verify” and drop every AI-provided citation in there before publishing. Five-minute audit = long-term credibility.
5. Voice Journal – Keep a living doc of phrases, anecdotes, and analogies that sound like you. When ChatGPT dulls your shine, draw from it to put your trademark flair back in.

Use the tool—don’t let the tool use you. Your future readers (and your frontal lobe) will thank you.

ElevenLabs Rolls Out Its Voice-First Personal Assistant

​11.ai by ElevenLabs – A voice-first assistant powered by ElevenLabs’ 5,000-plus lifelike voices. Speak a request and it acts: pull answers from Perplexity, open tickets in Linear, ping Slack, or drop a calendar invite—no typing.

Free alpha taps the platform’s existing 10k-credit tier and hints at a hands-free agent future competing with Rabbit and Humane.

ElevenLabs @elevenlabsio Introducing 11ai - the AI personal assistant that's voice-first and supports MCP. This is an experiment to show the potential of Conversational AI: 1. Plan your day and add your tasks to Notion 2. Use Perplexity to research a customer 3. Search and create Linear issues 12:24 PM • Jun 23, 2025 344 Retweets 3919 Likes Read 134 replies

​

​Read All mAIn Street Back Issues Here​

Your input fuels the content. Let me know how I did today!