Godwin's Law in Robotics, AI and LLMs

[Sergei Grichine]

There is a law coined by American attorney and author Mike Godwin in 1990.

The official wording of Godwin's Law is:

"As an online discussion grows longer, the probability of a comparison involving Nazis or Hitler approaches 1."

Well, psychological laws are universal. Nowadays, a new rule applies:

"As an online discussion about robotics grows longer, the probability of a comparison or debate involving AI and LLMs approaches 1."

Let's call it Sergei's Law ©. I am tempted to add a corollary: "...longer than two interactions." This theorem is easily proven by looking at almost any email thread.

As the sole creator of Sergei's Law (properly copyrighted, as you might have noticed), I have a respectful request. The next time you feel an urge to drop your (or Claude's) valuable thoughts about AI and LLMs into a thread that has been light-years away from those topics, please start your post with: "As per Sergei's Law ©..."—and then, by all means, fire away!

Or, better yet—open a new thread and link back to the original.

Hint: There are many other laws of nature waiting for an overhaul. Hurry up before somebody else puts their name on one! 😉

Best Regards,

-- Sergei

[Thomas Messerschmidt]

😆

Thomas Messerschmidt

-  

Need something prototyped, built or coded? I’ve been building prototypes for companies for 15 years. I am now incorporating generative AI into products.

Contact me directly or through LinkedIn:   

https://www.linkedin.com/in/ai-robotics/

On Jul 31, 2026, at 1:50 PM, Sergei Grichine <vital...@gmail.com> wrote:



--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/CA%2BKVXVNtUtUs4a%3DZyfT_k%2BB1QwThzZ0oUSvv5BNNDakzZCnFVg%40mail.gmail.com.

[Chris Albertson]

I think your law is too good; the discussion doesn't need to be very long.

Out of curiosity, I just asked Gemini to calculate the percentage of recent robotics papers that made reference to AI or ML.     The answer was that between 75% and 85% of all robotics papers do reference AI.

In other words, the field of AI and robots has nearly converged.       Historically, it seems if we go back more than 10 years, the overlap was roughly half, but now we are at about 80%.   This assumes Gemini was good enough to follow my instructions about finding peer-reviewed work and looking for references to AI.   I did not read all those papers myself.

I think tht’s happened is that our expectations have changed.  Back in thr 2000s we’d ask a robot “what is you location and then a SLAM algorithm would as “X= 32.435, Y=16.485” but today you. want the robot to say “I’m in the laundry room, about 2 meters in front of the dryer."

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/85063518-97CA-41DC-A623-4E50B5E4045B%40gmail.com.

[Michael Wimble]

I live by three laws. Here’s one:

Wimble’s First Law: You can’t cure stupidity by telling people they are stupid.

[Sergei Grichine]

Oops, Michael - it's too late to put your name on that law - but you can be first if you could come up with an effective punishment ;-)

On a serious note, we hobby roboticists are practicing escapists, and escaping into the dream of AI compounds it.

Modern AI is built by teams of incredibly bright people under powerful, mature organizations with virtually unlimited capital and computing power. Let me know if anyone reading this fits that description—I definitely don't earn a single point.

With a framework like ROS, we scratch the surface at a user level to produce a working solution. It is at most 10% original work: configuration, launch files, descriptions, and a tiny bit of coding. Even those who dig deeper rarely touch the core architecture. Now, with "vibe coding", we have a tool to do it faster, learning by asking for explanations and debugging help. We are having a great time with our toys: sensors, movement, navigation, and simple behaviors. The price tag? A couple grand at most. To me, that is the bright side of hobby robotics. Fun.

The dark side emerges when we assume that adopting AI shares the same process, price, and outcome. Slapping an API chatbot onto a robot to make it talk, detect objects, or follow voice commands is easy. Is it impressive? Sure. Does it move you forward? It depends.

If the goal is simply to learn a few tools, build a quick demo, and have fun, then absolutely. The problem is that the AI industry moves so fast that tools and models live for mere days. What reaches us are just crumbs from the tech giants' table while the foundation is still churning. Building anything substantial on top of it feels impossible. Any real attempt leaves you high and dry, trapped between slow network, obsolete black boxes and high subscription prices. It is a dead end, if you ask me.

I naturally asked our Overlords (Gemini) to add to my rant - here is their contribution:

Strategic Points You Could Add

If you want to expand your argument, consider introducing these two missing realities for the hobbyist:

• The Illusion of Autonomy: When a hobbyist connects a robot to a cloud LLM, the robot isn't actually "smart." It is just a physical terminal for a massive remote server. If the internet drops or the API changes, the robot instantly becomes a paperweight.
• The Compute Bottleneck: True robotics requires real-time, edge processing (on-device). The "big guys" run models on arrays of $30,000 enterprise GPUs. Trying to squeeze those capabilities onto a hobbyist budget (like a Raspberry Pi or Jetson Nano) creates a massive hardware wall that "vibe coding" cannot fix.
• The "API Sandbox" Trap: Big tech companies actively encourage hobbyists to use their APIs because it provides them with free telemetry, testing data, and user lock-in. Hobbyists are essentially paying subscriptions to act as unpaid QA testers for products they will never own.

So, be warned - get back to the bright side (ROS) and stop listening to the Pied Piper, kids! ;-)

Best Regards,

-- Sergei

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/94D31E7E-C0D0-4DCC-BD9A-A33EF43D4C9E%40gmail.com.

[Michael Wimble]

Well, that made me smile. The next time we get together, the cookies are on me.