How AI Got 120 Likes on a Joke and What It Tells Us About Risk Management

RISK-ACADEMY (Alex Sidorenko)
RISK-ACADEMY (Alex Sidorenko)May 30, 2026

Why It Matters

The anecdote highlights both the persuasive power and fragility of AI-generated content and underscores a business case for more autonomous, multi-step AI systems that can reliably validate data and produce actionable risk analyses—reducing manual iteration and improving decision quality.

Summary

A risk-management practitioner recounts using an AI agent—trained on his own articles and videos—to generate a Maslow-style diagram as an April Fools joke and a serious, heavily iterated version. The prank version unexpectedly received more engagement (120 likes) than the carefully refined one (50 likes), exposing how audiences can take AI-produced content at face value. He describes a workflow where he explains a vision to the agent, which drafts coherent text, and contrasts that with the laborious 11–12 iteration process required to produce a rigorous, mutually exclusive, collectively exhaustive framework. He argues for the next generation of AI that can autonomously execute multi-step analyses: sourcing and validating external and internal data, resolving inconsistencies, and delivering decision-ready outputs with corporate memory.

Original Description

Alex Sidorenko shares an honest experiment with AI in risk management.
He spent 11 iterations refining a Maslow's Hierarchy for Risk Management
with AI. Then, because it was April Fool's Day, asked the AI to generate
the exact opposite — a diagram full of the fluffy, marketing rubbish
that plagues traditional risk management.
The punchline? The joke version got 120 likes. The real one got 50.
And most people took the nonsense seriously.
But beyond the irony — this experiment highlights exactly
what's missing in AI today. Right now, getting quality output
requires a human in the loop at every single micro-step.
Prompting, correcting, prompting again. Exhausting.
The real future of AI in risk management is multi-step doing —
where you give a task, AI goes away, develops the methodology,
fact-checks itself, pulls in the right data, and delivers
a full draft based on coherent corporate memory.
Not the zero shot. Not the multiple shot.
A real draft. The product of multiple thinking steps.
00:00 — The April Fool's experiment: real vs joke
00:45 — 50 likes vs 120 likes: what people actually engage with
01:30 — Iteration 11 or 12: what quality actually takes
02:15 — The future of AI: multi-step doing, not prompting
03:00 — Not zero shot, not multiple shot — the full draft
▶️ Watch the full interview: https://youtube.com/live/9DnxW9oXNe4
Explore AI-powered risk management 👉 https://scyai.com
Join our YouTube channel as a member for new videos, RAW workshops or online courses
Read more risk management articles at
Watch the best practical workshops and case studies from RISK AWARENESS WEEK
DOWNLOAD our risk management book for free. Number 1 in Google search globally with 200000+ downloads
Check out online short courses on Udemy
Check out awesome risk management quantitative software
#RiskManagement #RM2 #ArtificialIntelligence #RiskTech #ScyAI

Comments

Want to join the conversation?

Loading comments...