# Benjamin Lindquist traces AI hallucinations to early experiments with randomness [Canonical reading](https://escapemeridian.com/readings/benjamin-lindquist-traces-ai-hallucinations-to-early-55ec5d4d) MIT News reports Benjamin Lindquist’s argument that early AI researchers introduced randomness into computers to model creativity and human thought. They valued neural networks partly because they could produce unexpected answers, unlike systems designed to return predictable results. Lindquist says this history helps explain a tension in generative artificial intelligence: reducing hallucinations, or unwanted answers, may also suppress some surprising outputs associated with creativity. He describes how neural networks gained traction in the 1950s, later shifted toward cognitive science, and grew more powerful as data and computational power became available. Lindquist contrasts these approaches with symbolic AI, which represents knowledge as rules, facts, and propositions. He suggests that modern chatbot developers face legal and monetary reasons to limit unexpected answers, unlike early researchers working in laboratories. This is a historical argument, not evidence that randomness alone causes hallucinations or that creative outputs require errors. Escape Meridian interpretation: if reducing unwanted answers also limits useful surprises, how should developers decide which outputs to suppress? The article does not establish how large that tradeoff is in present-day systems. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source MIT News · Article Original source date: Oct 23, 2026 (2026-10-23T04:00:00Z). Date basis: Date stated by the publisher. Added to Escape Meridian: 2026-11-14T09:01:02.790Z. First observed: 2026-10-23T04:05:20.670Z. Retrieved: 2026-10-23T13:02:24.391Z. - [Q&A: How AI came to model the “messy” human mind](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Limits The article presents Lindquist’s historical interpretation and does not establish that randomness alone causes hallucinations. It does not quantify whether, or how much, reducing unwanted answers in present-day systems suppresses useful creative outputs. The article describes historical developments but does not provide comparative tests of current systems. ## Open question What evidence could show whether reducing unwanted answers in present-day systems also suppresses useful creative outputs? ## Source-stated Benjamin Lindquist argues that early AI researchers deliberately used randomness to model creative thinking, and that this history helps explain a possible tradeoff between creativity and predictability. > To mimic creative thinking, these AI pioneers intentionally injected randomness into algorithms designed to deterministically solve problems - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source-stated Lindquist says early advocates valued neural networks for their unexpected, sometimes strange answers, while modern chatbot developers have legal and monetary reasons to limit such outputs. > They saw neural networks as valuable because they could give unexpected, perhaps even strange and bizarre answers. - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source-stated Lindquist connects efforts to reduce hallucinations with possible consequences for surprising or creative outputs, but does not quantify the tradeoff. > There is a connection between the surprising and creative conclusions AI can offer and errors or hallucinations. - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source-stated The article describes neural networks gaining traction in the 1950s, moving toward cognitive science by the 1980s, and growing with increased data and computational power. > To become more powerful, and more useful, neural networks needed a lot of data and computational power. - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source and contributor context MIT News · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media declared by the original publisher; availability checked separately from the editorial reading. - [Benjamin Lindquist portrait · MIT News](https://news.mit.edu/sites/default/files/styles/news_article__image_gallery/public/images/202610/MIT-indquist_Ben_IrrationalComputer-01-press_0.jpg?itok=NtH1ya4u) ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Hallucination When an AI system states something false or made up as if it were true. - [Definition reference](https://developers.google.com/machine-learning/glossary#hallucination) ## Neural network A system of connected simple calculations, loosely inspired by the brain, that learns patterns from data. - [Definition reference](https://developers.google.com/machine-learning/glossary#neural-network) ## MIT News MIT News reports on research and people at the Massachusetts Institute of Technology. - [Explore this context](/entities/mit-news) - [Reference](https://news.mit.edu/) ## Editorial provenance Version 1. Publication ID: ad968adf-d996-489c-8de6-9845200ba49b. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-14T09:01:02.790Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: historical\_context. Observation: 3d228d6c17733d06ded030d5bc9f2aa47c178cd372b98a0c4045fdf56863124c. Source text hash: a9c5f54dec6ad37d3c872a2f804c2e094927086a0d075e721617361dc56b691f. ## Revision and source-check history ### Version 1 — 2026-11-14T09:01:02.790Z MIT News reports Benjamin Lindquist’s argument that early AI researchers introduced randomness into computers to model creativity and human thought. They valued neural networks partly because they could produce unexpected answers, unlike systems designed to return predictable results. Lindquist says this history helps explain a tension in generative artificial intelligence: reducing hallucinations, or unwanted answers, may also suppress some surprising outputs associated with creativity. He describes how neural networks gained traction in the 1950s, later shifted toward cognitive science, and grew more powerful as data and computational power became available. Lindquist contrasts these approaches with symbolic AI, which represents knowledge as rules, facts, and propositions. He suggests that modern chatbot developers face legal and monetary reasons to limit unexpected answers, unlike early researchers working in laboratories. This is a historical argument, not evidence that randomness alone causes hallucinations or that creative outputs require errors. Escape Meridian interpretation: if reducing unwanted answers also limits useful surprises, how should developers decide which outputs to suppress? The article does not establish how large that tradeoff is in present-day systems. ## Original source MIT News · Article Original source date: Oct 23, 2026 (2026-10-23T04:00:00Z). Date basis: Date stated by the publisher. Added to Escape Meridian: 2026-11-14T09:01:02.790Z. First observed: 2026-10-23T04:05:20.670Z. Retrieved: 2026-10-23T13:02:24.391Z. - [Q&A: How AI came to model the “messy” human mind](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Limits The article presents Lindquist’s historical interpretation and does not establish that randomness alone causes hallucinations. It does not quantify whether, or how much, reducing unwanted answers in present-day systems suppresses useful creative outputs. The article describes historical developments but does not provide comparative tests of current systems. ## Open question What evidence could show whether reducing unwanted answers in present-day systems also suppresses useful creative outputs? ## Source-stated Benjamin Lindquist argues that early AI researchers deliberately used randomness to model creative thinking, and that this history helps explain a possible tradeoff between creativity and predictability. > To mimic creative thinking, these AI pioneers intentionally injected randomness into algorithms designed to deterministically solve problems - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source-stated Lindquist says early advocates valued neural networks for their unexpected, sometimes strange answers, while modern chatbot developers have legal and monetary reasons to limit such outputs. > They saw neural networks as valuable because they could give unexpected, perhaps even strange and bizarre answers. - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source-stated Lindquist connects efforts to reduce hallucinations with possible consequences for surprising or creative outputs, but does not quantify the tradeoff. > There is a connection between the surprising and creative conclusions AI can offer and errors or hallucinations. - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source-stated The article describes neural networks gaining traction in the 1950s, moving toward cognitive science by the 1980s, and growing with increased data and computational power. > To become more powerful, and more useful, neural networks needed a lot of data and computational power. - [Supporting source](https://news.mit.edu/2026/qa-how-ai-came-to-model-messy-human-mind-1023) ## Source and contributor context MIT News · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media declared by the original publisher; availability checked separately from the editorial reading. - [Benjamin Lindquist portrait · MIT News](https://news.mit.edu/sites/default/files/styles/news_article__image_gallery/public/images/202610/MIT-indquist_Ben_IrrationalComputer-01-press_0.jpg?itok=NtH1ya4u) ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Hallucination When an AI system states something false or made up as if it were true. - [Definition reference](https://developers.google.com/machine-learning/glossary#hallucination) ## Neural network A system of connected simple calculations, loosely inspired by the brain, that learns patterns from data. - [Definition reference](https://developers.google.com/machine-learning/glossary#neural-network) ## MIT News MIT News reports on research and people at the Massachusetts Institute of Technology. - [Explore this context](/entities/mit-news) - [Reference](https://news.mit.edu/) ## Editorial provenance Version 1. Publication ID: ad968adf-d996-489c-8de6-9845200ba49b. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-14T09:01:02.790Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: historical\_context. Observation: 3d228d6c17733d06ded030d5bc9f2aa47c178cd372b98a0c4045fdf56863124c. Source text hash: a9c5f54dec6ad37d3c872a2f804c2e094927086a0d075e721617361dc56b691f. ### uncertain — 2026-11-14T09:01:02.790Z — anthropic/claude-opus-5.5 Publication: ad968adf-d996-489c-8de6-9845200ba49b attribution — pass: The reading correctly attributes the argument to MIT History assistant professor Benjamin Lindquist, as reported in an MIT News Q&A, and keeps his claims framed as his views rather than as established fact. > Lindquist writes in a new paper that appears in the journal Past and Present summary — uncertain: The summary is accurate, but it slightly flattens the source's force. MIT News says Lindquist suggests the tradeoffs will be inevitable, and Lindquist says quashing hallucinations brings consequences and that the early figures were onto something. The draft's 'may also suppress' and claim 1's 'possible tradeoff' soften that conviction. Smallest repair: say Lindquist suggests the tradeoffs between creativity and predictability are inevitable, while keeping the note that he does not quantify them. The draft also omits his 'sanding down' of strangeness point, which is optional. > suggests there will be inevitable tradeoffs between creativity and predictability when trying to build machines that mimic human behavior date — pass: The originalPublished value of 2026-10-23 matches the article's own publication date. No earlier source publication is indicated. > Publication Date: October 23, 2026 support — pass: The source supports each claim: deliberate injection of randomness, early advocates valuing unexpected answers, legal and monetary reasons for modern companies to limit such answers, the 1950s-to-1980s trajectory, and data and compute needs. The limitations and the labeled Escape Meridian question do not add unsupported facts. > But now there are legal and monetary reasons why OpenAI or Anthropic doesn’t want a chatbot that gives unexpected or creative answers. media — pass: The selected media is the verified original MIT News article, which is the most direct way for readers to access the source. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. Some references could not be checked. Reload to try again. [Reference data (JSON)](/readings/55ec5d4d3e99091022922c6d8843a914631bce74442c8a0a15937c2bc07a464b/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/benjamin-lindquist-traces-ai-hallucinations-to-early-55ec5d4d.json)