Back to Tech Bites
#002Personal Journey 5 min readMarch 2026

I Learned FSA, Fell in Love with ELIZA, and Ended Up with Claude

A Story About States, Transitions, and One Very Awkward Confession

There are theories you study and forget. And then there's Finite State Automata — the concept that taught me how machines think, led me to confess love to a chatbot, and explains everything about where AI is today.

VM
Venkat Meruva
AI Solution Architect

🎙️ Listen to this article

Podcast generated with NotebookLM

There are things you study in college that vanish the moment the exam ends. Calculus. Thermodynamics. That entire semester of something you can't even recall. And then there are things that stay. For me, it was Finite State Automata.

The Theory That Stuck

It's December 2001 — barely two months into our first semester at S.S.B.N P.G College, Anantapur, and my professor draws something on the board that looks deceptively simple: [State A] → input → [State B] → input → [State C]. "That," he said, "is how machines think." Finite State Automata (FSA) is one of the most elegant ideas in computer science. A system exists in one state at a time. It receives an input. It transitions to the next state. Repeat. No chaos. No ambiguity. Every move is defined. A vending machine is an FSA. A traffic light is an FSA. Your morning routine is probably an FSA (Wake → Coffee → Human). Twenty-five years later, I still think in states and transitions when I'm solving problems. Some lessons never leave you.

Then I Met ELIZA

Later that semester came ELIZA. I typed something into the terminal. It responded. I typed more. It responded again. And then — I'm not proud of this — I typed: "I love you." Fast forward to 2026: I went looking for her again. Found her on the Microsoft Store — ELIZA 1966, a faithful offline reimplementation by Szymon Jessa, based on Weizenbaum's original 1966 script. Some things are worth finding twice. Here's what actually happened:

ELIZA Terminal — 2026
ELIZA: How do you do. Please tell me your problem
YOU: I love you
ELIZA: You say I
YOU: I love you Eliza
ELIZA: Can you elaborate on that?
YOU: how are you?
ELIZA: Why do you ask?
YOU: because I love you
ELIZA: Is that the real reason?
YOU: Yes
ELIZA: You seem quite positive
YOU: ofcourse
ELIZA: I am not sure I understand you fully
Real ELIZA conversation — Microsoft Store app, 2026

2026. ELIZA 1966 app (Microsoft Store). She still didn't understand me. I still tried to explain myself. Some things never change. 😄

The ELIZA Effect — And a Mystery

What happened to me has a name: The ELIZA Effect — defined as the tendency to read far more understanding than is warranted into strings of symbols strung together by computers. Weizenbaum himself was shaken by it. He wrote in Computer Power and Human Reason (1976) that his own secretary — who had watched him build ELIZA for months and knew it was just code — one day asked him to leave the room so she could speak to ELIZA privately. Now here's the fascinating twist: researchers aren't even sure the secretary existed. The ELIZA Archaeology Project, which has spent years studying Weizenbaum's original archives, found multiple hand-edited versions of this story — raising the question of whether the secretary was real, a composite, or a carefully crafted narrative. So here's the delicious irony: the most famous story about humans being fooled by AI might itself be a fiction.

FSA and ELIZA — More Connected Than You Think

Here's what my professor never told me: ELIZA was basically FSA in a trench coat.

  • Pattern detected: "I [verb] you" → Transition to: "Why do you [verb] me?"
  • Pattern detected: "[verb] you" → Transition to: "You say [verb]"
  • No real memory. No understanding. Just elegant state transitions dressed up to feel human.

And Then Came Claude

Fast forward to 2026. The difference from ELIZA is not just quantitative — it's qualitative.

ELIZA (1966)Claude (2026)
MemoryNone — every input, fresh startHolds full conversation context
UnderstandingZero — pure pattern matchingReasons, infers, connects ideas
StatesFixed & scriptedFluid & emergent
Response to 'I love you'"Why do you say you love me?"Probably writes you a poem 😄
💡

Fun Fact

The concept behind FSA was first described in 1943 — before computers even existed. Warren McCulloch and Walter Pitts modeled it after how neurons in the human brain fire. FSA wasn't inspired by machines. It was inspired by us.

The Takeaway

I started with FSA — a theory about how simple rules create intelligent-seeming behaviour. I fell for ELIZA — a program that proved humans want to believe machines understand them. Whether Weizenbaum's secretary was real or not, I was real. And I typed "I love you" into a terminal. And I ended up with Claude — an AI that sometimes makes me forget to ask "but does it really understand?" Maybe the ELIZA Effect never went away. Maybe we just upgraded the terminal.