You Probably Believed At Least One AI Lie This Year. Here’s How to Stop.
A Harvard study said AI killed junior jobs. Sam Altman said agents will 'join the workforce' in 2025. Most people assumed that brainstorming with AI outperforms brainstorming with fellow humans. All of these are misleading, exaggerated, or missing critical context.
All of these are wrong, exaggerated, or missing critical context.
If you made career decisions, hiring decisions, or AI investments based on headlines this year, this is your correction.
If You’re Using AI to Brainstorm, Read This First
You’re probably making your ideas worse, not better.
These studies I analyzed this year all point to the same conclusion: AI boosts individual “originality” scores while tanking collective diversity. Everyone ends up in the same clever box.
→ Is Brainstorming With AI REALLY A Good Idea?
What you’ll know after reading: When to use AI for ideas (and when it’s sabotaging you without you noticing).
If You Think AI Is Why Juniors Can’t Get Hired...
You’re solving the wrong problem.
A Harvard study with 62 million data points has a fatal flaw: the hiring crash started before ChatGPT existed. The real culprits? Interest rates and over-hiring corrections.
→ Are Junior-Level Jobs Really Killed by AI?
What you’ll know after reading: How to spot when prestigious research is built on shaky foundations — and what’s actually happening to entry-level hiring.
If Your Proposals Keep Getting Rejected...
The screener might prefer your competitor’s AI-written version. Literally.
I interviewed the researcher who discovered AI systems choose AI-generated content 60-95% of the time. Not because it’s better — because it pattern-matches.
→ AI Doesn’t Discriminate Against You. It Just Prefers Its Own Kind.
What you’ll know after reading: How to audit whether AI gatekeepers are filtering you out — and what to do about it.
If You’re Betting on AI Agents in 2026...
The best models complete only 24% of real workplace tasks autonomously. The rest? They fail, get stuck, or require human rescue.
→ Is AI Destroying SaaS? + AI Agents Problems No One Talks About
What you’ll know after reading: Where agents actually work vs. where the demos are lying to you.
One Question Before 2026
I spend 20 hours per week reading research so you get the signal without the noise.
But I don’t know what you need most.
Here’s a 2-minute survey: What can I best write for you in 2026?
If You’ve Been Receiving Value From This…
Some of you have been here a while.
You’ve forwarded these to colleagues, mentioned the ideas you read here in meetings, maybe used an insight to push back on a vendor or challenge an assumption by someone in your network.
If this work has earned a spot in how you think about AI, I’d love to have you as a paid subscriber.
Price goes up to $75/year on Jan 1. Four days left at $50.
What you get:
Exclusive deep-dives that don’t hit the free list
Reply access — ask me anything, I actually respond
30 mins session with the team
Independence — no sponsors and no incentive to hype
— Jing
P.S. — The survey genuinely shapes what I write. If something’s missing, tell me.
I value your unique ways of showing more nuance about AI. Have a good new year
Hi Hao — I really enjoy your practical LLM posts.
I recently wrote about why LLMs often miss user intent even when the relevant information is present. My take is that it’s not just a “knowledge” gap, but a thinning meta-layer for observability, explanation, and control — the layer where human judgment and intent get negotiated.
I think it connects directly to reliability, safety, and evaluation for real-world systems.
Here’s the piece if you’re curious:
https://northstarai.substack.com/p/ai-spoke-of-a-meta-layer-in-its-own