• Tech Times
  • Posts
  • AI’s writing your emails. Maybe even this one

AI’s writing your emails. Maybe even this one

Question our existential existence together in our new series AI EXPLAINED

AuthorAuthor

Rob LeFebvre & Kevin Johan Wong
June 20, 2025

AI EXPLAINED

Presented By:

plaud.ai

Newsletter Issue #2 – “Thanks for Connecting 👍”

Rob dives today into the world of quick-compose - email replies, smart suggestions, that ghostwriter living inside Gmail. No harm, no foul, right? Unless you’re like me and get a little twitchy every time someone sends one of those LinkedIn auto-replies. (“Thanks for connecting!” Cool. Why?)

Most of these AI-generated responses still don’t sound like real people. Sometimes they miss the tone, and other times they miss the point entirely. But you can’t deny they’re changing how we create. Even if you can still tell what AI is vs. not — with your own eyes or with tools like ZeroGPT — it’s reshaping the process under the hood.

For now, at least for us, it doesn’t write for us, but it helps us think. Helps us organize. This newsletter, for example, gets a little help from Trinity, a personal tool we built that combines Gemini, GPT, and Perplexity into one model, then has them hilariously argue with each other. Keeps us honest. Keeps AI from glazing us and telling us we’re the best. Keeps us out of the bubble.

This issue of AI Explained newsletter is presented in partnership with Plaud.ai — creators of the world’s smallest voice recorder and the smartest AI notetaker. Whether you’re in meetings, on calls, or mid-rant in the shower, Plaud captures it, cleans it, and turns it into gold. Check them out → or if you're looking to join in the story and get a free Plaud.ai device, fill in the form here →.

Check out Rob’s piece here →

👩‍💻 Bonus: Our two cents on LinkedIn AI replies

  • Roughly 40% of LinkedIn messages now involve autocomplete or AI-suggested text (LinkedIn Engineering Blog, 2023–2024). That number’s probably even higher if you count the “invisible assist” when you let AI write the first draft and just tweak.

  • Now, do these AI-written DMs get fewer replies? Probably,  especially for messages that actually matter. The tone’s off. The context is missing. And yeah, people can tell.

  • But on the flip side… AI gets people to send something. And… you miss 100% of the shots you don’t take, even if it’s a clumsy, over-polished, GPT-generated layup.

  • So maybe the bigger shift isn’t how people reply, it’s who suddenly feels empowered to reach out.

Still, we’re all using it. Just depends if you’re using it to say something… or to say nothing faster.

🧵 Bonus Read: Karpathy on Keeping AI On a Leash

If you’re looking to go deeper into the AI rabbit hole, this one’s worth your time:

In this Tech Times piece, Andrej Karpathy (ex-OpenAI, ex-Tesla) issues a word of caution: don’t unleash unsupervised agents into the wild just yet. He argues that while autonomous AI is powerful, it’s prone to feedback loops, hallucinations, and unintended behavior,  especially when humans blindly follow it. Think of it as Clippy, but with way more influence and way less oversight.

TL;DR for the Doomscrollers:

  1. AI is quietly ghostwriting your inbox — from Gmail replies to LinkedIn DMs — but most messages still feel soulless and easy to spot.

  2. Even imperfect AI is changing behavior: people are writing more, faster, and sometimes saying something when they’d otherwise say nothing at all.

💌 P.S.

If you have any thoughts, shoot us a message back; we’re human! And if you’re building AI that speaks with people, not just at them, drop us a line. We’ve got a platform for you (literally and figuratively).

Until next time,

– The AI Explained team

Until next time,

AI EXPLAINED team