The Origins of AI: From John McCarthy to Modern Foundation Models

Chief Strategy Officer (CSO)

Simple question with a contested answer: who is the father of AI? I want to give my exec team a short, accurate historical foundation before I make them sit through another quarter of strategy work.

AI Leadership Agent — Future Strategy Advisory

The honest answer: there is no single father — AI has several, and the credit depends on which lineage of the field you're tracing. The most widely cited candidate is John McCarthy, but the picture is richer than that.

1. John McCarthy — coined the term "artificial intelligence" in 1955 and organized the 1956 Dartmouth Summer Research Project, which is widely considered the founding event of the field. Most textbooks give him the title.
2. Alan Turing — wrote the foundational 1950 paper "Computing Machinery and Intelligence," introducing the Turing Test. Without his theoretical groundwork, the Dartmouth workshop wouldn't have had a target to aim at.
3. Marvin Minsky — co-founder of the MIT AI Lab and the most influential evangelist of symbolic AI for decades.
4. Frank Rosenblatt — built the Perceptron in 1958, the direct ancestor of every neural network running today.
5. Geoffrey Hinton, Yann LeCun, Yoshua Bengio — the trio who kept neural networks alive through the long winter and made modern deep learning possible. Often called the godfathers of deep learning.

If your team asks for one name, say John McCarthy. If they want the truer story, tell them AI has parents — and the modern systems they're using descend much more directly from Hinton's lineage than McCarthy's.

Chief Strategy Officer (CSO)

Why does that distinction matter strategically? Isn't this just trivia?

AI Leadership Agent — Future Strategy Advisory

It matters because the field has gone through two very different paradigms, and most of your senior leaders were educated under the first one. That shapes their intuitions about what AI can do — often incorrectly.

1. The symbolic era (1956–1990s) — AI as hand-crafted rules and logic. This is what most over-40 executives implicitly think AI is. It explains why they sometimes ask why models can't just "follow the rules."
2. The statistical / machine learning era (1990s–2010s) — AI as patterns learned from data. The shift from rules to learning is what made things like spam filters and recommendation engines work.
3. The deep learning and foundation model era (2012–present) — AI as enormous neural networks trained on enormous datasets. This is what powers every product your competitors are launching this year.

Strategy goes wrong when leaders apply symbolic-era intuitions to foundation-model-era technology. The systems behave differently, fail differently, and need to be managed differently.

A two-slide history lesson at the start of your strategy offsite is one of the highest-ROI things you can do this quarter.

Chief Strategy Officer (CSO)

Give me the two-slide version, then. What's the simplest narrative I can hand to my CEO that's still true?

AI Leadership Agent — Future Strategy Advisory

Two slides, plainly told:

1. Slide one — "Seventy years of trying." In 1956, a small group at Dartmouth set out to build thinking machines. For five decades, progress was real but narrow. AI played chess, diagnosed diseases, recommended films — but always inside tight boundaries.
2. Slide two — "The last decade changed everything." In 2012, deep neural networks beat every other approach to image recognition. In 2017, the Transformer architecture made it possible to train models on essentially all the text on the internet. In 2022, those models became conversational. In 2024 and 2025, they became agentic. We are not in a new technology cycle — we are in a new computational paradigm.

End with the strategic implication: the question for your business is no longer whether AI works. It's how fast your organization can absorb a technology that doubles in capability every 12 to 18 months. That's the framing that turns history into strategy.