AIAny
Machine Learning Foundation Papers1950

Computing Machinery and Intelligence

Reframes "can machines think?" as a concrete test: the imitation game, now the Turing test, where a machine passes if its typed replies are indistinguishable from a human's. Rebuts nine objections and backs machines that learn like children.

Visit Website

Introduction

Seventy years before chatbots, a mathematician decided the question "can machines think?" was too vague to settle and quietly swapped it for one a referee could actually score. That move — not any claim about silicon souls — is why this 1950 paper still frames every debate about AI today. Turing's real contribution was epistemic: he made "intelligence" an operational, behavioral wager rather than a metaphysical one.

Key Findings
  • The imitation game reframes thinking as something you test from the outside: if a machine's typed answers are indistinguishable from a person's, the question of inner experience is set aside as undecidable, not solved.
  • Turing pre-empted his critics, working through nine objections — theological, mathematical (Gödel), consciousness, "Lady Lovelace's" (machines do only what we program), and others — and answered each rather than waving them away.
  • His prescription for getting there was learning, not hand-coding: build a "child-machine" with a simple initial program and educate it, anticipating machine learning decades before the hardware existed.
  • He made a falsifiable bet — that by 2000 a machine could fool an average interrogator 30% of the time after five minutes — turning philosophy into a measurable target.
Methodology

The paper's force comes from a rhetorical inversion. Instead of defining "think" and then asking whether machines qualify, Turing fixes a game with clear win conditions and lets the definition follow from performance. This sidesteps the trap that sinks most consciousness debates: arguments about what is "really" happening inside. The cost is deliberate — the test measures convincing imitation, not understanding, a gap later sharpened by Searle's Chinese Room.

Who It's For

Essential reading if you want the original logic behind the Turing test, the historical root of behavioral AI evaluation, or a model of how to make a slippery question tractable. Look elsewhere if you want technical methods: there is no algorithm, architecture, or experiment here — it is a philosophical argument. And read it critically: passing the imitation game has come to look more like a test of human gullibility than of machine intelligence, a limitation modern LLMs have made uncomfortably concrete.

Back

Information

  • Websitecourses.cs.umbc.edu
  • OrganizationsUniversity of Manchester
  • AuthorsAlan Turing
  • Published date1950/10/01

Categories

Tags

More Items

Large Language Model Papers2026
Icon for item

Multi-Stream Attention Residuals

Xiaohan Qin, Xiangdong Zhang +8School of AI, Shanghai Jiao Tong University, Shanghai Innovation Institute +3

Enables multi-head, depth-wise residual retrieval in Transformers by expanding a single residual stream into multiple context-aware streams using causal depthwise convolutions. Improves pretraining convergence and downstream benchmark scores with minimal overhead, yielding notable iso-loss compute speedups in large MoE models.

papertransformersmixture-of-expertsmoedepth+3
Large Language Model Papers2025
Icon for item

Foundations of Large Language Models

Tong Xiao, Jingbo ZhuNLP Lab, Northeastern University, NiuTrans Research

A concise textbook-style book that explains foundational concepts and techniques for large language models, covering pre-training, generative models, prompting, alignment, inference, and reasoning. Structured as self-contained chapters for readers with some ML/NLP background or those seeking a principled introduction to LLM foundations.

foundation-modelLLMNLPpaperbook+5
Large Language Model Papers2026
Icon for item

Questioning the Questions: Sustaining Self-Evolution in Reasoning Models

Jinyuan Li, Chengsong Huang +5Washington University in St. Louis, University of Michigan, Ann Arbor

Analyzes why self-evolving reasoning models collapse under repeated self-training and proposes R-Quest: a feedback-driven pipeline that trains solvers to reject invalid questions and uses a frozen base model to detect task-level repetition, filtering training data to sustain multi-round gains.

reasoningpaperllmbenchmarksmath+2