MIT historian says early AI engineers added randomness to mimic human creativity
Benjamin Lindquist, an MIT historian, explains that mid‑20th‑century engineers deliberately introduced randomness into algorithms to model the “messy” human mind. He traces the idea to figures like John Pierce, who argued that math could inject unexpected bursts into art, and to early neural networks built by psychologists to observe productive mistakes.
Key points
- Early AI pioneers injected randomness to emulate creative, unexpected human thought.
- Mid‑century engineers like John Pierce saw math as a way to add irregularity to art.
- Modern systems prioritize correct answers, suppressing the original “strangeness” valued by early researchers.
Lindquist contrasts that early appetite for occasional strange or bizarre answers with today’s focus on reliable, correct outputs from systems such as ChatGPT and Claude. He notes that legal and monetary pressures have “sanded down” the strangeness that once was valued, even though the underlying randomness still exists. The shift from symbolic AI to neural networks, accelerated by the internet’s data and compute growth around 2012, illustrates the trade‑offs between creativity and predictability that persist in modern AI development.
Q&A: How AI came to model the “messy” human mind
MIT News on AI · 23 October 2026
Loading the full article…
This text was published by MIT News on AI and written by Adam Zewe | MIT News. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Researchers introduce MemPlan for text-based planning with PDDL · 1 src
- Researchers present CiteAnchor framework for detecting hallucinated citations · 1 src
- Researchers find self‑rewarded moral argumentation makes language models decisive but pliable · 1 src
- AQL-Lite improves Arabic LLMs with curated data and partial-parameter training · 1 src
- arXiv study maps gaps in AI research for low-resource languages · 1 src
Comments
via GitHub Discussions