History · Article

From Machine Code to ChatGPT: How Programming Changed in Eight Decades

Published:

ENIAC (Electronic Numerical Integrator And Computer) in Philadelphia , Pennsylvania . Glen Beck (background) and Betty Snyder (foreground) program the ENIAC in building 328 at the Ballistic Research Laboratory (BRL).
Image credit: Unknown author Unknown author · Public domain · Wikimedia Commons

In brief

Software programming began with engineers setting raw binary instructions by hand, then moved through languages built for people — Fortran, C, Python — before today's AI systems, built on the 2017 Transformer architecture, started generating code and text themselves. This piece traces that shift through dated milestones confirmed by IBM, the IEEE Computer Society, Dartmouth College and the original Transformer research paper.

Software did not start with typing commands into a high-level language. The earliest electronic computers were configured instruction by instruction in the raw binary patterns their hardware could read directly — what is now called machine code. From there, programming moved through assembly language, then high-level languages built for people rather than processors, and over the past decade into systems that generate their own code and text. The dates and figures below come from IBM's own corporate history, the IEEE Computer Society, Dartmouth College's account of its 1956 workshop, and the 2017 research paper that introduced the Transformer architecture used in today's chatbots.

From machine code to mnemonics

According to a timeline published by the IEEE Computer Society, one of the earliest programming languages was Plankalkul, developed by German engineer Konrad Zuse in the 1940s. By 1949, assembly language and a system called Shortcode had appeared, replacing long strings of binary digits with short mnemonic codes — but those codes still had to be translated, line by line, into the format the processor could execute.

High-level languages: writing for people, not processors

The shift accelerated in 1957, when IBM delivered Fortran, short for 'formula translation.' IBM's own history of the project states that a team led by mathematician John Backus — which included Lois Haibt, who built the compiler's arithmetic-expression analyzer — spent three years on a language originally expected to take six months. IBM says Fortran cut programs that previously needed roughly 1,000 machine-code instructions down to as few as 47 lines, letting scientists and engineers write algebra-like commands such as 'DO' and 'GO TO' instead of coding the hardware directly. By the mid-1960s, Fortran had become a national computing standard, according to IBM.

The IEEE Computer Society's timeline lists Algol and Lisp as arriving in 1958, and Cobol, built for business computing, in 1959. In 1972, Dennis Ritchie's team at Bell Labs produced C, which the same IEEE timeline credits as the structural basis for later languages including JavaScript, PHP and Python. Object-oriented programming followed with C++ in 1983. The IEEE account also records Python's debut in 1991, Java, JavaScript and PHP in 1995, Google's Go in 2009, and Apple's Swift in 2014.

A parallel track: teaching machines to reason

While high-level languages were spreading, a separate line of research asked whether machines could simulate intelligence rather than simply follow instructions. Dartmouth College's own account describes the 1956 Dartmouth Summer Research Project, organized by mathematician John McCarthy, whose proposal aimed to test what it called 'the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.' Dartmouth credits that workshop as the starting point of artificial intelligence as a defined research field.

Decades later, that research produced a widely watched public test. IBM's history of its Deep Blue chess computer states that the machine lost its first match against world champion Garry Kasparov 4–2 in Philadelphia in February 1996, then won a rematch in New York in May 1997 by a score of 3.5–2.5 — the first time, per IBM, that a computer had defeated a reigning world chess champion under standard tournament conditions.

For the first time in the history of mankind, I saw something similar to an artificial intellect.

— Garry Kasparov, quoted on IBM's official history page after the 1997 Deep Blue match

The Transformer and the generative-AI era

The architecture behind today's chatbots traces to a single 2017 paper. Eight researchers — Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan Gomez, Lukasz Kaiser and Illia Polosukhin — posted 'Attention Is All You Need' to the arXiv preprint server on June 12, 2017. The paper introduced the Transformer, described in its abstract as a network based solely on attention mechanisms, removing the recurrent and convolutional layers earlier translation systems relied on. The authors reported scores of 28.4 BLEU on English-to-German translation and 41.8 BLEU on English-to-French, with faster training than prior models.

That architecture underlies the large language models released from 2018 onward. OpenAI launched ChatGPT on November 30, 2022. According to a History.com account citing OpenAI co-founder Sam Altman, the service reached one million users within five days and roughly 100 million monthly users within about two months, which the account describes as the fastest consumer-app adoption on record at that time.

Six dated milestones in programming and AI history
YearMilestoneSource-confirmed detail
1956Dartmouth Summer Research ProjectJohn McCarthy's workshop proposal framed machine intelligence as a research question, per Dartmouth College
1957Fortran released commerciallyIBM's Backus-led team cut programs needing about 1,000 machine-code instructions to as few as 47 lines
1972C developed at Bell LabsBecame the structural basis for later languages including JavaScript, PHP and Python, per IEEE
1997Deep Blue defeats Garry KasparovFirst computer win over a reigning world chess champion in standard tournament play, per IBM
2017Transformer architecture publishedAttention-only design later underlying GPT and BERT, per the original arXiv paper
2022ChatGPT released by OpenAIReached about 100 million monthly users within roughly two months, per OpenAI data cited by History.com

Across nearly eight decades, each step traced above reduced how much hardware-specific detail a person had to write by hand — from Zuse's Plankalkul and hand-set binary instructions, through Fortran's algebra-like commands, to attention-based models that now generate code and text from natural-language prompts.

Questions

What was the first high-level programming language?
Fortran, developed by a team led by John Backus starting in 1954 and released commercially by IBM in 1957, according to IBM's own history of the project.
Why was the C programming language significant?
Developed at Bell Labs in 1972, C became the structural basis for later widely used languages including JavaScript, PHP and Python, according to the IEEE Computer Society's timeline.
When did artificial intelligence become an established research field?
At the 1956 Dartmouth Summer Research Project, organized by John McCarthy, which Dartmouth College's own account credits as the field's starting point.
What changed with the 2017 'Attention Is All You Need' paper?
Eight researchers introduced the Transformer, an architecture based solely on attention mechanisms rather than recurrent or convolutional networks; it later underpinned models such as GPT and BERT, according to the original arXiv paper.
When was ChatGPT released and how fast did it grow?
OpenAI released ChatGPT on November 30, 2022; per OpenAI co-founder Sam Altman as reported by History.com, it reached one million users within five days and about 100 million monthly users within roughly two months.

Sources

  1. IEEE Computer Society — A Timeline of Programming Languages
  2. IBM — History of Fortran
  3. IBM — History of Deep Blue
  4. arXiv — Attention Is All You Need (Vaswani et al., 2017)
  5. Dartmouth College — How the Term 'Artificial Intelligence' Was Coined
  6. History.com — ChatGPT, the Generative AI Chatbot, Is Released

This piece was prepared with artificial intelligence and passed an automated source check before publication. If you see an error, ask for a correction. Request a correction or reply · How we make this

Related