Origins of AI (Intelligences, Intelligent and Clever Ones, 1970?)

Miguel Fernández
Nov 7
3 min read

From the first time I heard the neologism “AI” as an abbreviation for “Artificial Intelligence,” it didn’t sit well with me. At first, it seemed like a translation mistake. Later, the term, probably of North American origin, was adopted worldwide. Since “artificial intelligence” is something that doesn’t exist and, God willing, never will, it was unclear what exactly people meant by it. It seems that the correct term would be “Intelligent Automatism – IA,” as some say, or “Computational Intelligence (CI in English),” as I’ve heard academics use. So it seems I’m not alone in finding “AI” an inappropriate name. “AI” is, at the very least, a linguistic error in any language.

Thinking about this, one thing led to another, and I remembered an episode from about 55 years ago (1970?), which I record here as the beginnings of the problem. At that time, commercial computers were just appearing on the market. In practice, all “programs” were custom-made by specialists using FORTRAN-MONITOR and COBOL languages, which translated commands into the basic binary language of electronics, the root of computing, as it remains today. These programs were entered through “punched cards and tapes,” which the machine “read.” There was one called the IBM360, considered the pinnacle of the time (with less capacity than my cell phone today and occupying a room of about 30 m²).

Financial institutions (banks) were ecstatic. That would put an end to the enormous amount of manual entries (pen on paper), full of errors and fraud. Little did they know that while it marked the beginning of fewer small errors and small frauds, it also marked the beginning of much greater ones.

Around that time, the old Banco do Estado da Guanabara (BEG), later BANERJ, bought some of these IBM360 machines and computerized all services in its new, modern, and impressive headquarters, made of “varnished exposed reinforced concrete with epoxy”, at the corner of México, da Ajuda, and Av. Nilo Peçanha streets, in Rio. Wonderful!

Four years after the system was implemented, a branch manager received a client who was upset about the disappearance of one cruzeiro cent (Cr$ 0.01) from his account.

Annoying customer! Just one cent? Probably some rounding or recurring decimal issue!

After much searching, the persistent manager and client dug deep and discovered that a programmer had inserted a subroutine that, on even days, took one cent from odd-numbered accounts and deposited it automatically into his own account. On odd days, it took one cent from even-numbered accounts, doing the same. It was an “intelligent automatism”, or, in today’s (2025) trendy language, an AI created by the programmer. Since then, computers, programs, and programmers have only evolved.

Why was it discovered? Because there’s always someone more alert, meticulous, careful, or simply more “annoying.” And the world doesn’t run only on “intelligence.”

It was discovered mainly because of greed. The man was already a millionaire, he could have stopped.

If he had stopped, he’d never have been caught. But human nature, the sense of impunity, the thrill of risk, didn’t let him. It’s as if getting caught took away the fun! He needed recognition for his genius, for his “natural intelligence.”

I did some math to evaluate the damage: suppose BEG-BANERJ had 1,000,000 accounts nationwide (333 branches). Assuming today’s values, with 1 USD ≈ 5 BRL, and considering that the US dollar has devalued 7.75 times since then (based on gold value), and with 230 business days per year:

1,000,000 clients / 2 × ((0.01/5)×7.75) × 230 days/year = $3,750/day × 230 = US$1,782,500/year

Over 4 years: US$7,130,000 (2025), or about R$35,000,000.

In other words, since the dawn of computing, we’ve been in the hands of programmers. No natural or artificial intelligence can change that. But compared to white-collar crimes, all “within the law”, this was a trifle.

A record remains.