Turing Test Project

“A computer would deserve to be called intelligent if it could deceive a human into believing that it was human.” – Alan Turing

https://www.youtube.com/embed/DccwpcoWY0w

Introduction

As computation becomes more efficient, software is rapidly replacing human workers in the labor market. Robots now work assembly lines, the Internet has replaced the postal service as the primary means for distributing written communication, and electronic kiosks have replaced information desks and even, in some cases, waiters. These are examples of physical labor—but what about mental labor? Can computer software ever rival human intelligence? Will computers replace doctors, educators, or even artists?

What does it mean to be intelligent? How can we measure intelligence?

Assignment

Your job is to write a protocol for conducting a Turing Test that can distinguish artificial intelligence from human intelligence using only text-based chat.

Submission

Submit a write-up of your Turing Test protocol experiment. The protocol should include the following:

  • an introduction that states the purpose of the test and its underlying principles,
  • at least 10 detailed directives for someone to complete in order to identify artificial intelligence through text-based chat (these steps should be so detailed that any two people performing the Turing Test would do so in exactly the same way),
  • quotations used as necessary to indicate exact text that must be used (e.g., questions to ask),
  • example responses for each of the questions, describing in detail how text responses should be analyzed (e.g., differences in expected and actual output for a given input), and
  • an analysis of the strengths and weaknesses of the protocol.

Learning Goals

You will be able to:

  • Comprehend how the mechanical manipulation of symbols is a form of problem solving.
  • Analyze a “chatterbot” for pattern recognition and manipulation.
  • Evaluate multiple approaches for designing a Turing Test for both effectiveness and generalizability.
  • Differentiate between “strong” and “weak” AI.
  • Identify relationships between the AI subfield of Natural Language Processing (NLP) and Human-Computer Interface (HCI).
  • Describe visual and speech recognition as a form of multi-modal AI.
  • Discover, implement, and exploit strategies to discern between human and artificial intelligences.
Criteria Points
TURING TEST PROTOCOL
Describes clear and concise steps for the administration of your Turing Test experimental design. 3 pts
Explains how the final Turing Test design assures that the experiment is effective and generalizable. 3 pts
Uses key terminology from the Artificial Intelligence glossary appropriately and effectively. 2 pts
TURING TEST RESULTS
Provides background information for the analysis of the chatterbots’ AI. 3 pts
Provides a detailed, sequential account describing the administration of your Turing Test to two different chatterbots. 2 pts
TURING TEST EVALUATION
Describes the two chatterbots, analyzing their strengths and weaknesses, and pattern recognition and manipulation abilities. 4 pts
Includes specific examples as evidence. 2 pts
Commits few, if any, grammatical, organizational, or spelling errors. 1 pt
TURING TEST SYNTHESIS
Your Scratch program:
Responds generally to user input. 4 pts
Responds specifically to selective phrases. 2 pts
Utilizes previous responses (memory). 2 pts
Includes useful and clear documentation. 2 pts
TOTAL 30 pts