Abdirahman Jama, a Software Development Engineer @ AWS | Opinions are my own, recently posted something that made me stop scrolling: "I think a lot of people might call in sick tomorrow because ChatGPT is down".

It is a quick joke, but it lands because it points at something real: many of us have quietly woven ChatGPT (and similar tools) into our daily workflow so tightly that an outage can feel like losing a key teammate.

In this post, I want to expand on what Abdirahman Jama hinted at with that one-liner: the productivity boost is real, but so is the dependency. And when a tool becomes infrastructure, we need to treat it like infrastructure.

Why that joke hit so hard

Abdirahman Jama's line works because it compresses a modern workplace truth into a single image: people staring at an error screen and realizing how much of their day is routed through an AI assistant.

A few reasons it resonates:

Many roles now use AI for first drafts, summarization, and quick research.
The value is immediate and personal: less blank-page anxiety, faster iterations, fewer context switches.
The dependency is subtle. It builds one helpful prompt at a time until it becomes the default.

Key insight: When a tool becomes your "first step" for everything, downtime stops being an inconvenience and starts becoming a blocker.

What a ChatGPT outage really disrupts

Even if you do not rely on AI for final decisions, a lot of teams use it for the glue work that keeps projects moving. When the system is down, the friction shows up everywhere.

1) Writing and communication slows down

The most common usage is also the least visible: drafting.

Email replies
Status updates
PRDs and tickets
Meeting agendas and follow-ups

If you usually ask AI to produce a first pass, the outage does not remove your ability to write. It removes your speed and your momentum.

2) Problem solving feels heavier

Developers, analysts, and operators often use AI as a rubber duck that talks back.

"Explain this error"
"Suggest edge cases"
"Give me a checklist"

When that disappears, you still have the skills, but you lose the fast feedback loop. It can feel like going from a calculator back to mental math.

3) Context switching increases

AI reduces the need to open ten tabs, skim docs, and stitch together answers. Without it, you return to the older workflow: search, scan, compare, repeat. That is not bad, but it is slower and more draining.

Key insight: Outages hurt most when AI is not just a tool, but a workflow design choice.

The hidden risk: AI as a single point of failure

Abdirahman Jama's joke about calling in sick is funny because it exaggerates a real risk: a critical dependency with no fallback.

We typically plan for outages in systems we consider production-grade: cloud services, CI pipelines, ticketing tools, VPN access. But many individuals and teams have not yet applied the same discipline to AI assistants.

Here are a few ways this single point of failure shows up:

Knowledge gets "outsourced" to prompts: people stop maintaining their own templates and checklists.
Teams rely on AI to interpret requirements, logs, or customer feedback quickly.
Quality control gets weaker: if AI normally does first-pass proofreading, the team may ship rougher writing when it is down.

None of this means "stop using AI". It means recognize the operational reality: if it is essential, treat it as essential.

Practical ways to stay productive when AI is down

If you want the benefits without the fragility, build a simple resilience plan. Nothing fancy, just enough to keep moving.

1) Create a small offline toolkit

Keep a local folder or internal page with:

Writing templates (status update, escalation note, postmortem outline)
Checklists (launch readiness, incident triage, code review)
Common snippets (customer replies, documentation sections)

These do not replace AI. They reduce the shock when AI is unavailable.

2) Standardize prompts into reusable assets

If your best prompts live only in your head, an outage is not the only risk. You are also losing repeatability.

Turn your most-used prompts into:

Team-approved prompt docs
Form-based intake (goal, audience, constraints)
Examples of "good outputs" for calibration

That way, when the tool returns, you ramp back up faster and with more consistency.

3) Use multiple options, but keep governance

Having a backup model or provider can help, but it introduces security and compliance questions.

If you work with sensitive data:

Know what is allowed and what is not.
Avoid pasting confidential content into unapproved tools.
Prefer internal or enterprise accounts when possible.

Resilience is not worth it if it creates a data incident.

4) Practice manual mode occasionally

This is the simplest habit: intentionally do some tasks without AI now and then.

Draft a doc without assistance, then use AI for editing later.
Debug a problem for 15 minutes solo before asking.

You keep the skill sharp and reduce the feeling of helplessness during downtime.

A workplace lesson hiding inside the humor

What I like about Abdirahman Jama's post is that it is not anti-AI. It is a wink at the new normal.

The bigger lesson is about tooling maturity:

Early stage: AI feels like a nice-to-have.
Adoption: it becomes a multiplier for individuals.
Integration: it becomes part of team throughput.
Dependency: outages become operational events.

Once you hit that last stage, leaders should treat AI availability, access, and training like any other productivity platform.

Key insight: The goal is not to avoid reliance. The goal is to choose it intentionally and manage it responsibly.

Why this post went viral (and what to learn from it)

Even though the original post is just one sentence, it follows a strong viral formula that works especially well on LinkedIn.

It is timely

Outages are shared experiences. When something widely used goes down, people look for commiseration and quick takes.

It is specific

Not "AI is changing work". It is "ChatGPT is down". That specificity makes it instantly relatable.

It is low-stakes and human

The "call in sick" exaggeration gives people permission to laugh at themselves, which invites comments and shares.

It invites stories

A great LinkedIn post does not just make a point. It triggers replies like:

"This happened to me"
"Here is my workaround"
"We should not depend on it"

If you are thinking about LinkedIn content strategy, this is a strong reminder: clarity beats complexity, and one sharp observation can outperform a long essay.

Closing thought

Abdirahman Jama's joke is funny because it is uncomfortably close to true. AI tools have become a quiet layer in our daily productivity stack. When they disappear for a few hours, the interruption reveals what we have been optimizing for: speed, convenience, and cognitive relief.

The best response is not panic or denial. It is a simple upgrade in how we work: keep templates, build repeatable processes, think about backups, and maintain the underlying skills. Then, the next time ChatGPT goes down, you will still laugh at the joke, but you will not lose the day.

This blog post expands on a viral LinkedIn post by Abdirahman Jama, Software Development Engineer @ AWS | Opinions are my own. View the original LinkedIn post →