Georgetown University
Graduate School of Art and Sciences
Communication, Culture & Technology Program

CCTP-607 "Big Ideas" in Technology (and What They Mean): AI to the Cloud
Professor Martin Irvine

Spring 2018


This course will provide a conceptual and design-oriented introduction to some of the "big ideas" in technology that everyone needs to know. The course is especially designed for students from non-technical backgrounds, but all students can benefit from learning the methods in this course.

In launching this new course, we will build much of the "content," questions, and problems to be studied from the interests and background levels of students in the course. The topics and readings are not yet determined for many of the syllabus weeks. We will allow many questions and starting points to emerge from the active engagement of students.

Every cycle of “new” technology brings responses of hype, hope, and hysteria. In this course, we will go beyond the hype and learn what our current “Big Ideas” in technology are really about: What is “artificial intelligence” or “AI”? What are “algorithms” and how are they designed? What is “The Cloud”? What is “Big Data” and what do we mean by “Data Analytics” and “Data Visualization”? What is “the Internet of Things (IoT)”? What are “smart” appliances (home security, media services, shopping)? Every day, the news media, popular discourse, marketing, and advertising are full of statements about these technologies, but treated as unfathomable “black boxes” and corporate branded products. This course will provide the methods, key concepts, and analytical tools for understanding the “Big Ideas” and what they mean.

We will combine four main methods and approaches for an interdisciplinary deblackboxing method. This integrated method works to reveal how everyone, not just technical people, can understand the meaning of the “Big Ideas” behind our technologies and find ways to participate in how they can be used. We will will combine:

(1) “Systems Thinking” to understand how a specific technology is part of a larger, interrelated system (for example, computing systems, kinds of software, networks, and social contexts);

(2) “Design Thinking” for uncovering how and why certain technologies are designed the way they are, including the history of designs and the consequences of design choices;

(3) “Semiotic Thinking” for understanding these technologies as artefacts of human symbolic thought, which includes (a) understanding how sign systems and media can be digitally encoded as “information” or data, (b) the relationship between abstract models (e.g., algorithms, code, data models) and how (or whether) they can be implemented technically, and (c) understanding the social meanings, values, and purposes of the technical systems;

(4) the “Ethics and Policy” viewpoint for evaluating the social consequences of design choices in the large-scale adoption of certain kinds of technologies, and for analyzing proposals for ethical decisions and governmental policy.

The course will include individual and group writing assignments for students to learn how to “de-hype” popular discourse and marketing language, and use our methods and analytical tools for better ways of interpreting these technologies and explaining them to others. Requirements will also include in-class group projects and a final research project on one of the “Big Ideas.”

Classroom location: Car Barn 318

Course Format

The course will be conducted as a seminar and requires each student’s direct participation in the learning objectives in each week’s class discussions. The course has a dedicated website designed by the professor with a detailed syllabus and links to weekly readings and assignments. Each syllabus unit is designed as a building block in the interdisciplinary learning path of the seminar, and students will write weekly short essays in a Wordpress site that reflect on and apply the main concepts and approaches in each week’s unit. Students will also work in teams and groups on collaborative in-class projects and group presentations prepared before class meetings.


Grades will be based on:

  • Weekly short writing assignments (in the course Wordpress site) and participation in class discussions (25%). Weekly short essays must be posted by 10:00AM for each class day so that students will have time to read each other's work before class for a better informed discussion in class.
    • Group discussion presentations as part of weekly assignments: You will have 2-3 group discussions to prepare for presentation in class. Choose a group "scribe" to collect together your notes (and any images, diagrams, etc. you want to use) and post in one person's "author" post. Your group post should capture the outcome of what you studied and discussed, with the key points that you think are important or interesting to present in class. List the names in your group at the top of the post. Everyone will get a group grade for that week's class discussion.
    • On weeks that you do a group study topic, that post will count as your weekly writing assignment. No individual post, just one collective post. Focus on your own discussion and what you would like to present in class, and we will have a "cross-group" discussion on the topics in class.
  • A group research project (25%) to be presented in class after the 10th week. Presentation date and group rosters will be determined by the 10th week of the seminar.
  • A final research project written as a rich media essay or a creative application of concepts developed in the seminar (50%). Due date: one week after last day of class.
    (Final projects will be posted on the course Wordpress site, which will become a publicly accessible web publication with a referenceable URL for student use in resumes, job applications, or further graduate research) .

Professor's Office Hours
Wed. and Thurs. 12:00-2:00, and by appointment. I will also be available most days after class meetings.

Academic Integrity: Honor System & Honor Council
Georgetown University expects all members of the academic community, students and faculty, to strive for excellence in scholarship and in character. The University spells out the specific minimum standards for academic integrity in its Honor Code, as well as the procedures to be followed if academic dishonesty is suspected. Over and above the honor code, in this course we will seek to create an engaged and passionate learning environment, characterized by respect and courtesy in both our discourse and our ways of paying attention to one another.

Statement on the Honor System
All students are expected to maintain the highest standards of academic and personal integrity in pursuit of their education at Georgetown. Academic dishonesty, including plagiarism, in any form is a serious offense, and students found in violation are subject to academic penalties that include, but are not limited to, failure of the course, termination from the program, and revocation of degrees already conferred. All students are held to the Georgetown University Honor Code: see

Instructional Continuity
In the event of a disruption of class meetings on campus from inclement weather or other event, we will continue the work of the course with our Web and online resources, and will arrange for online discussions and meetings with the professor by using the Google Hangout interface in our GU Google apps suite. I am also always available via email, and respond to student messages within a few hours or less.

Books and Resources

This course will be based on an extensive online library of book chapters and articles in PDF format in a shared Google Drive folder (access only for enrolled students with GU ID). Most readings in each week's unit will be to pdf text links in the shared folder, or to other online resources in the GU Library.

Required Books:

  • Peter J. Denning and Craig H. Martell. Great Principles of Computing. Cambridge, MA: The MIT Press, 2015. 
  • Luciano Floridi, Information: A Very Short Introduction. Oxford, UK: Oxford University Press, 2010.

Recommended Books:

  • Jerry Kaplan, Artificial Intelligence: What Everyone Needs to Know. Oxford University Press, 2016.

Course Online Library (Google Drive: GU student login required)

University Resources

Using Research Tools for this Course (and beyond)

  • Required: Use Zotero for managing bibliography and data for references and footnotes.
    Directions and link to app, Georgetown Library (click open the "Zotero" tab).
    You can save, organize, export and copy and paste your references with formatted metadata into any writing project.

AI Research and Information Sources

News and Research Sources

Stanford "100 Years of Artificial Intelligence" Study Site

AI, Ethics, and Human-Centered Design: University Research Centers

Pew Research Projects: Algorithms, AI and the Future

Professional Computing and AI Sources (ACM)

Orientation to Learning Goals of the Course:

  • Establishing some useful definitions, distinctions, and scope of subject matter included in "Artificial Intelligence."
  • What are we talking about when we talk about "Artificial Intelligence" and "Machine Learning"?
  • How to apply methods from design thinking, key principles of computing, semiotics, and cognitive science to deblackboxing AI, ML, "Data," and "The Cloud."

Old computer science jokes:

  • "Artificial intelligence is the science of getting computers to do what they do in Hollywood SF movies."
  • "AI works in practice, but not in theory."

Course Introduction: Requirements and Expectations

  • Format of course, requirements, participation, weekly assignments, projects, outcomes.
  • Classroom rules: how to use PCs and mobile devices: no social media or attention sinks during class.

Introduction to the Topics and Key Concepts of the Course (Presentation)


Learning Objectives:

  • Establishing some useful definitions, distinctions, and scope of subject matter included in "Artificial Intelligence."
  • Design concepts in computing and AI.
  • Major schools of thought, traditions, and applications of AI.
  • Reviewing some recent accessible descriptions and definitions of AI for common assumptions.
  • Introduction to concepts and methods for critical analysis of assumptions: what are some of the unexpressed ideologies and commitments in AI and views of computing technology?
  • Introduction to major applications and uses of "AI": language processing, data analytics, image analysis, pattern recognition.

Readings: Orientations

Film documentary screening: Do You Trust This Computer? (2018) (segments in class)

In-class shared note file (Google Doc)

Weekly writing and reflection (link to Wordpress site)

  • Read the Instructions for the weekly writing assignment.
  • Describe 1-2 common topics or themes in these introductory books. What questions came to mind as you were reading? What did you find helpful in "deblackboxing" AI; unhelpful? Do you see patterns in the approaches and discourses about computers and AI that continue to totalize "AI" (and computers in general) as autonomous "things" (reifications) rather than human designs (artefacts)?

Learning Objectives and Main Topics:

  • Continuing from last week's reading and discussion: AI/ML and how to describe and explain the technologies involved.
  • Introductory steps in understanding the key concepts and design principles in computing and computing systems that AI and ML systems depend upon.
  • Beginning our parallel path in learning how to analyze the discourse of AI.


  • Peter J. Denning and Craig H. Martell. Great Principles of Computing. Cambridge, MA: The MIT Press, 2015. Chapters 1-2.
    • We will be referring to the computing principles outlined in this great book throughout the course. Read it and re-read it. Get as far as you can this week, and continue reading over the next 3 weeks.
  • Margaret A. Boden, AI: Its Nature and Future (Oxford: Oxford University Press, 2016).
    • Chaps. 3-5. (It's good to get the continuous argument and framework of a major practitioner in AI for understanding the consistency of one approach and viewpoint.)
  • Ethem Alpaydin, Machine Learning: The New AI. Cambridge, MA: The MIT Press, 2016.
    • Read chaps. 2-3, focus on "Pattern Recognition."
  • Jerry Kaplan, Artificial Intelligence: What Everyone Needs to Know. Oxford University Press, 2016. Chaps. 1-3.
  • Marvin Minsky, Ray Kurzweil, and Steve Mann, “The Society of Intelligent Veillance,” in IEEE International Symposium on Technology and Society (ISTAS) (2013): Social Implications of Wearable Computing and Augmediated Reality in Everyday Life, 2013, 13-17.
    [This is a typical group of assertions that have been part of AI discourse for many years, presented in summary form. There are long chains of arguments, research, and theory assumed here, but these claims are often made without support or argument. We will analyze how this happens.]

AI/ML in Discourse: How are AI Technologies Constructed in Discourse?

  • Deborah G. Johnson and Mario Verdicchio, “Reframing AI Discourse,” Minds and Machines 27, no. 4 (December 1, 2017): 575–90.
    • Parallel with learning about AI models and applications, we will also learn how to analyze statements and assertions about AI and any computing technology to clarify key concepts. This article provides a good general orientation to how we can "reframe" the discourse (terms, concepts, assumptions) and work out better descriptions and explanations (useful in an overall approach to de-blackboxing what is closed off).
    • We will apply the proposals in this article to analyzing the statements in the Do You Trust This Computer? (2018) documentary.

Continued: Introduction: Topics and Key Concepts of the Course (Presentation)

In-class shared note file (Google Doc)


Weekly writing and reflection (link to Wordpress site)

  • Continuing from last week, choose one of the topics in the Boden, Alpaydin, and Kaplan readings to work through for own state of understanding so far, and try to express the questions that you have that we can clarify in class.
  • As you work through a topic, do you see how it can be "reframed" by asking questions or uncovering unexpressed assumptions as proposed in the Johnson and Verdicchio article?

Learning Objectives

  • Next set of building blocks: foundational principles of computation and the development of AI as a multi- and inter-disciplinary field that combines computer science with philosophy, logic, mathematics, and cognitive science.
  • Learning foundational principles in computation and AI/ML through appraises to pattern recognition in image analysis.


  • Peter J. Denning and Craig H. Martell. Great Principles of Computing. Cambridge, MA: The MIT Press, 2015. Chapters 4, 6, 8. (Go as far as you can: the more you understand these key principles, the less AI and ML seem like mysterious, unknowable black boxes, because AI/ML programs are (must be) implemented in actual computing systems.)
  • The Interactive and Parallel Computing model: foundations of modern software, interfaces, Internet/Web computing, and much of AI:
    • Peter Wegner, “Why Interaction Is More Powerful Than Algorithms.” Communications of the ACM 40, no. 5 (May 1, 1997): 80–91. 
      [Do your best to catch the key concepts here (a famous statement about the Interactive Computing model). Interactive computing is a software and interface system model that enables ongoing communication between human users of a system and processes that can be given choices and directions while in process (like the operations we can do through GUI interfaces). Computational interaction is based on layering symbolic representations and cues for initiating and directing computational actions that can work in parallel or concurrently in a local device and across networks. This design is the opposite of the Turing-Von Neumann computational model of one sequential process at a time that terminates (halts) when completed. AI and ML systems today use massive parallelism -- processors with multiple CPUs or clusters of many processors -- to perform many operations over massive amount of data. There are also multiple "inputs" along the way using the interactive model (whether assigned to software "agents" designed for this, or people interpreting results and providing ongoing inputs), so that the ongoing processes can be fine-tuned (adjusted) for the results intended.]
  • Ethem Alpaydin, Machine Learning: The New AI. Cambridge, MA: The MIT Press, 2016.
    • Focus on Chaps. 3-4, Pattern Recognition, and Neural Networks.
  • Geoff Dougherty, Pattern Recognition and Classification: An Introduction (New York: Springer, 2012). Excerpt: Chaps. 1-2.
    • Read the Introduction and look at the examples and methods for pattern classification in Chap. 2. This is the assumed background for the ML application to "Selfie" analysis in the Karpathy article (next).

AI/ML application case study: Pattern Recognition Using Neural Networks

  • Andrej Karpathy, “What a Deep Neural Network Thinks About Your #selfie,” Andrej Karpathy Blog (blog), October 25, 2015,
    • This article (by a young AI/ML thought leader) provides an "inside the blackbox" view of how "Convolutional Neural Networks" (the mathematical network graphs that can continually readjust the "weights" [values] between the nodes in the probabilistic calculations) can be used to do feature detection, pattern recognition, and probabilistic inferences/predications for classifying selfie photo images.
    • All "neural net" ML techniques use mathematical graphs as models for designing the algorithms (the abstract mathematical models encodable in the programming languages used) for the kind of data analysis being specified.
    • You can also get a sense of the ethical implications of programming criteria (the parameters defined for selecting certain kinds of patterns) when the ML design is "tuned" for certain kinds of results.

Optional and Supplemental: The Classic AI Intro Textbook

  • Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach, 3rd ed. (Upper Saddle River: Prentice Hall, 2010). (Selections: Chaps. 1-2).
    • This is the mostly used (and as the authors say, most downloaded) book on AI in computer science fields.
    • Review the Contents, and read chap. 1. This is a classic introduction written for computer science students. It is based on the logic and probability models for AI, still assumed in all computing and AI, but more attention is now given to "Neural Net" and "Deep Learning" methods, which are forms of probabilistic "learning" models. Do you best on this background and we will discuss in class.
    • This book has a companion Website hosted by the Computer Science Dept. at Berkeley. Note the resources available for study.

In-class shared note file (Google Doc)

Weekly writing and reflection (link to Wordpress site)

  • Group study project: post notes for leading discussion in class.
  • This week will add further "building blocks" to understanding computing and AI/ML design principles through an introduction to pattern recognition and an application in face-image analysis (Karpathy's article on "neural net" analysis of "selfies"). Study the background readings, and then discuss what you find to be the key points and issues in the Karpathy article. Apply as much as you can from our learning path so far.
  • Note: In your reading and discussion this week, don't be put off by the mathematical and technical models introduced in the readings. You will discover that the mathematical and logical models for the network (matrix) algorithms are all in service of our cognitive abilities for recognizing and interpreting patterns, and making decisions and projections (predictions) based on already learned patterns.

Learning Objectives and Topics:

In this week and next, students will learn the foundational concepts of "information" and "data" as used in digital electronics, computing, communications, and "data processing/data analysis."

All forms of computation and AI are based on "information [or "data"] processing" and "data analysis." But what is information (in the scientific and computational sense)? Instead of just eliding over this topic (and using vague, ordinary discourse conceptions), we need to understand two central concepts that are at the core of the design principles for computation, digital electronics, and AI/ML applications: information and/vs. data?

  • How is the concept of "information" used in computer science, digital networks, and all digital media and communication applications?
  • What is the difference between information and data as these terms are used in different disciplinary communities? Why does it matter to get clear on these concepts and how they are used?
  • Capsule definitions:
    • Information (in the electrical engineering context) is what can be measured and engineered for structuring physical, material signals or substrates (e.g., radio waves, binary digital electronic states in a physical medium like memory cells or Internet packets), which are then said to be the medium or "carrier" of data representations at the physical level.
    • Data implies further structure: units or computational "objects" by data type (e.g., digital image: patterns of visual units in a matrix R,G,B color values; text characters ["strings" in programming languages and software]; types of numbers [integers, rationals = "floating point" numbers]).
    • Classified, categorized, labeled "data objects" (i.e., "structured data") are what are organized in databases. We also have "unstructured data" -- a problem term, but it means heaps of text strings or streams of combined data types like that in text messages, social media streams, email, and blog posts, which is usually what is meant by "big data" or the starting point for analyzing "big data."


Readings (read in this order):

  • James Gleick, The Information: A History, a Theory, a Flood. (New York, NY: Pantheon, 2011).
    Excerpts from Introduction and Chap. 7.
    [A popular, accessible account of the background story of information theory as defined and applied in engineering. I recommend buying this book and following the issues that Gleick explains for non-technical readers.]
  • Martin Irvine, "Introduction to the Technical Theory of Information" (Information Theory + Semiotics)
    [Introduction to the key concepts in Shannon's "signals transmission mathematical model" for "information" in electronic communications systems.]
  • Luciano Floridi, Information, Chapters 1-4. PDF of excerpts.
    [Floridi is an influential philosopher, and this brief introduction is useful for main issues, but he is promoting his own "General Definition of Information," which is confusing. We need to separate out the semantic/semiotic (meaning-making, meaningfulness) level of description from the physical and mathematical level for defining the electronic, digital medium. We will discuss and untangle these issues in class.]
  • Peter J. Denning and Craig H. Martell. Great Principles of Computing, Chap. 3, "Information."

Background Documents (Optional, but review for the original statements)

  • Claude E. Shannon and Warren Weaver, The Mathematical Theory of Communication (Champaign, IL: University of Illinois, 1949).
    [Shannon's report was first published in 1948; a second edition was published with an introduction and comments by his colleague, Weaver.]
  • C. E. Shannon, “The Bandwagon,” IRE Transactions on Information Theory 2, no. 1 (March 1956): 3.
    [This is a one-page statement in which Shannon criticizes the use of the "mathematical signals transmission" model in other fields to which the model does not apply (getting on "the bandwagon"). Some comments seem like they could be applied to the unsupported enthusiasm for AI/ML (just change the technical terms):
    "Seldom do more than a few of nature's secrets give way at one time. It will be all too easy for our somewhat artificial prosperity to collapse overnight when it is realized that the use of a few exciting words like information, entropy, redundancy, do not solve all our problems.... The subject of information theory has certainly been sold, if not oversold. We should now turn our attention to the business of research and development at the highest scientific plane we can maintain."]

Case Study: The Information and Data Design Levels for the Internet & Web

  • In class, we will explain the basic models of "information" and "data" with a basic case study of the design of the Internet and Web. We will de-blackbox bits, bytes, and a first-level of data structures for everything used via the Internet and Web (including everything we experience in mobile apps). E-information design guarantees that everything we see and use works reliably, consistently, and predictably.
  • Background Reading:
  • Ron White, How Computers Work, 10th ed. (Indianapolis, IN: Que Publishing, 2014). Chapter 17: Networks and the Internet.
    [Well-illustrated introduction to the design of the Internet and both wired and wireless connections.]
  • Irvine, "Using the Model of Levels to Understand "Information," "Data," and "Meaning" (Internet design).
  • Optional (Going Further): if you want go further with background on Internet design principles, see Week 11 of CCTP-820: Leading by Design.
  • Video Background from How the Internet Works

Presentation (In-class and self-study): The Technical Design of Information (Irvine)


Weekly writing and reflection (link to Wordpress site)

Referring to at least two of the readings, choose one of these topics to focus your thinking and learning this week:

  • Describe the main features of the signal transmission theory of information and why the signal-code-transmission model is not a description of meaning (the semantic, social, and cultural significance of encoded signals)? Further, why is the information theory model essential for everything electronic and digital, but insufficient for extending to models for meanings, uses, and purposes of our sign and symbol systems?
  • Think through a case study on your own: How do we recognize the difference between E-information transmitted and received (successfully or unsuccessfully) and what a text message, an email message, social media post, or digital image means? What do senders and receivers know about the transmitted data signals that isn't a physical property of the signals? How is E-information designed as a substrate or physical medium for symbolic (meaningful) structures (text, images, music, video, etc.)?

Learning Objectives:
How do we conceive and structure data for technical uses?

We will continue the study of "data" as defined and used in computing, AI, and all data sciences. Both "information" and "data" are used in general and undifferentiated was in ordinary and popular discourse, but to advance in your learning for AI and all the data science topics that we will study, we all need to be clear on the specific meanings of these terms and concepts.

See "Introduction to Data Concepts and Database Systems" for background.


  • Michael Buckland, Information and Society (Cambridge, MA: MIT Press, 2017). Focus especially om Chaps. 2 ("Documents"), 5 ("Naming"), 6 ("Metadata").
    • Note the context of usage of "information"; assumed as D-information, data-structured information. This is a view from library science and knowledge management systems.
    • Read as much as Chaps. 1-7 as you can. This view of knowledge organization and metadata is usually not factored into ML data analysis.
  • John D. Kelleher and Brendan Tierney, Data Science (Cambridge, Massachusetts: The MIT Press, 2018).
    • This is a view of "data science" after new ML approaches to "unstructured data" and interactive Web-interface databases. Read Chaps.1-2, 4.
    • Good statement:
      "One of the biggest myths is the belief that data science is an autonomous process that we can let loose on our data to find the answers to our problems. In reality, data science requires skilled human oversight throughout the different stages of the process. Human analysts are needed to frame the problem, to design and prepare the data, to select which ML algorithms are most appropriate, to critically interpret the results of the analysis, and to plan the appropriate action to take based on the insight(s) the analysis has revealed." (33-34)
  • Irvine, "Introduction to Data Concepts and Database Systems."
  • David M. Kroenke et al., Database Concepts, 8th ed. (New York: Pearson, 2017). Excerpt.
    • This is a more technical introduction for students beginning data science studies in a computer science or information science program. Skim or review as far as you can. Consider the illustrations and explanations for the tables.

Cases and Example (choose two for thinking through data concepts):

  • Unicode (background in "Introduction to Data Concepts" above)
    • See Wikipedia for overview.
    • The Unicode Consortium (everything is open-source and international standards-based)
    • UTF-8 (Unicode Transformation Format - 8 Byte Units) is the most commonly used, including the character encoding and graphical rendering of the Web page in your current screen display "window."
    • Background documents on the current Unicode Standard 11, and Code Charts for All Languages
    • Unicode Emoji (pictographic symbols) [Yes! All emoji must have Unicode byte definitions or they wouldn't work for all devices, software, and graphics renderings. Emoji are not sent and received as images but as bytecode definitions to be interpreted in a software context. Again, code data and device-software contexts and rendering methods are separate levels.]
    • Current Unicode Emoji Chart (with all current skin tone modifications)
    • Unicode test file of all currently defined "emojis" (byte code, symbol, description: not all will display in your window or software context)
    • "Han Ideographs in the Unicode Standard," Yajing Hu (CCT student, final project essay)
      [This is a good essay that looks at the background of Unicode standards for Han characters, and other Asian language families. The same issues were dealt with for Arabic and many other "non-Latin" character languages.]
  • Relational Database Systems from Web and mobile app interfaces: Examples: eBay and Amazon.
    • Complex relational database systems, both using Amazon Cloud platforms (AWS: Amazon Web Services; take a quick view of Amazon's Relational Database System as a Cloud service).
    • Use your own account to now focus on what happens from our view of the database system(s), and what you can begin to "de-blackbox" with the data concepts we are studying this week.
    • All of the complexity in the relational database systems is hidden from view, but you can get a sense of how all this complexity can only be managed in an implementation of the relational database design model. Amazon has several database platforms, but a complex relational database behind the scenes enables all the presentations of data (text, images, user account information) in the formatted screen-window view that we see.
  • Library Database Examples:
    Main GU Library Search page. Also click on the "Database" tab for a database interface to databases available through GU Library (yes, a meta-database). (Requires GU login.)
    • Click on "P" in the Database index; scroll down and click on ProQuest Central.
    • This is one of the most used research databases. ProQuest is an aggregator of many journal and other primary data sources in one macro database interface. The system combines complex relational data structures with indices (links) to data repositories (document archiving systems).
    • Try searches on different topics; note the interactive selections (you are "activating" various data fields to be sorted on your queries). See if you can describe the basic levels in the database system(s), the query results of which we can only view at the presentation level.
  • Relational Database and Data Visualizations with Research Data:

Weekly writing (link to Wordpress site): choose one question to frame your thoughts, and cite references to the readings and data cases

  • The many contexts and uses of the terms "information" and "data" make these terms perplexing and confusing outside an understood context. Using the method of thinking in levels and our contexts for defining data concepts, outline for yourself the concept of "data" and its meaning in two of the data systems we review this week. (One "system" is the encoding of text data in Unicode for all applications in which text "data" is used; others are database management systems).
  • Discuss your experience using two examples of the data systems above, and explain, as far as you can, the levels and layers in which different kinds of data are used and interpreted. Ask any questions about the concepts or design structures in which "data" is used in the different computing contexts.

Learning Objectives

Learning the basic principles of Natural Language Processing (NLP) and Speech Recognition in AI methods: sorting, classifying, feature extraction, pattern recognition, translation.

Learning the basics about linguistic and computational levels of analysis and processing involved in NLP implementations (text, speech, and both). We will focus on examples of machine translation (troublesome term) and speech recognition or speech processing.

Readings and Video Introductions:

  • Data Structures: Crash Course Computer Science (PBS). Video Intro.
  • Machine Learning & Artificial Intelligence: Crash Course Computer Science (PBS). Video Intro.
  • Natural Language Processing: Crash Course Computer Science (PBS). Video Intro.
    • These are well-done, concise, 11 min. introductions. The "Data Structures" video is a good bridge between E-information structures at the binary and memory levels and "Data" as configurations of these structures identified by type in a computing system, which can then functions as as Data-Type "inputs" in a corresponding software process.]
    • For more background, see the whole series of Crash Course: Computing tutorials.
  • Daniel Jurafsky and James H. Martin, Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition, 2nd ed. (Upper Saddle River, N.J: Prentice Hall, 2008). Selections.
    • This is the "bible" for computer science and linguistics students in NLP. Use as reference and review as far as you can. You can always come back to these chapters for reference later.
    • These excerpts will give you a picture of the current state of computational and mathematical models used in all "machine learning" NLP, and the foundation of current and developing "Recursive Neural Net" models. As you see, even if you don't understand all the math, the models are based on graphs, linear algebra, and statistics. Nothing scary inside the black boxes. It's all learnable.
    • The background on formal methods in linguistics is excellent. For a crash course in the "linguistics" behind the "L" in NLP, see Chaps. 12-13, 15. Note the problem of "Part of Speech Tagging" (Chap. 5).
    • The chapters will also give you a sense of the complexity of the computational problems in processing natural language text and speech (voice sounds). Chaps. 7-9 give you a picture of what's involved behind the scenes in "speech recognition" so that Siri or Google voice commands work as expected.
  • Thierry Poibeau, Machine Translation (Cambridge, MA: MIT Press, 2017). Selections.
    [This is a good introduction for an overview. Read chapters 1-3, skim (for background) chaps. 4-6, and review more closely chaps. 7-12. This takes us up through ML and "deep learning" techniques.]
  • Machine Translation: Wikipedia article historical overview.
  • How Google Translate Works: The Machine Learning Algorithm Explained (Code Emporium). Video.

Examples and Implementations

  • Syntax (Grammar) parsing: mapping word classes and combination patterns
    • In all NLP, a syntax/grammar map must be used to interpret word strings (tokens; i.e., instances of words in the context of a sentence.)
    • The term "parse" in linguistics/computational linguistics/NLP comes from a traditional grammar term for breaking sentences down into their "parts of speech" [pars/partes: Latin for "part/parts"], i.e., the grammatical word classes and how they are combined (abbreviated POS, "part of speech"). Syntax parsers are assumed at one input level in all NLP (text and speech) methods. Web search algorithms and Siri voice recognition and interpretation software have to use some model of a POS pattern for generating a map of the probable grammatical structure of natural language phrases before processing probable semantic searches (and converting word tokens into "commands" to be activated in an algorithm, like answering a question or searching for information).
    • See a syntax parsing NLP system at work: XLE-Web (inside the language blackbox!)
      A sentence parsing tool and syntax tree generator that maps both the "Constituent Structure" and "Lexical Functional Grammar" models of generative grammar. Choose "English" (or other language you know) from the drop down list, and try any sentence for the system to map its constituent (c- ) and functional (f- ) structure! (Uses linguistic notation from two formal systems, the main one as a tree structure.) So here you get a visualization of what Siri and Google first has to map out in milliseconds behind the scenes before the other software algorithms can initiate a search or other command (even if using a different ML technique).
    • Why is this step in NL so important? Any NL sentence (or series of sentences) can have grammatical ambiguities, and in real life we use contexts and shared background knowledge to interpret the intended grammatical pattern. The XLE-Web system is designed to provide alternatives (ways the grammar could work), and choosing the most likely structure is what AI/ML NLP has to decide first. ML and "neural net" approaches rely on many layers of probabilities for word patterns across millions of samples, but a "successful" pattern analysis (in any NLP method) is one that matches the humanly understood (intended, expected) grammar pattern as closely as possible
  • BabelNet (go to live version)
  • WordNet | Search Wordnet semantic database online
  • Google Translate (
  • Further cases to study and de-blackbox:
    • The NLP behind the Google search algorithms (and speech commands to perform searches)
    • Siri and speech recognition in everyday applications

Case: Exposing Limitations of Neural Net ML Methods in NLP

  • OpenAI: New ML Language Model
    • This is the new RNN model that surprised everyone by how well the alogrithms could generate well-formed "fake news" from millions of data samples of news writing.
    • Since ML is designed for pattern identification and recognition, the algorithms will provide recognized patterns (because the patterns are there -- in human-composed sentences), but the fact of a pattern has nothing to do with its meaning, the relation of a pattern to its use and its contexts of interpretation (truth values, consistency in logic, beliefs).
  • Will Knight, “An AI That Writes Convincing Prose Risks Mass-Producing Fake News,” MIT Technology Review, February 14, 2019.
  • Karen Hao, “The Technology Behind OpenAI’s Fiction-Writing, Fake-News-Spewing AI, Explained,” MIT Technology Review, February 16, 2019.
    • Note the continuing reification of "AI" as an entity in journalistic discourse.

Weekly writing and reflection (link to Wordpress site)

  • Using the key concepts and descriptions of the technologies in the background readings and videos, describe the design principles of one or more levels at work in one of the NLP applications above.

Learning Objectives and Main Topics:

  • What are the main design principles and implementations of AI systems in interaction interfaces for information, digital media, and Internet/Web services that we use everyday?
  • What do we find when we de-blackbox the algorithms, computing processes, and data systems used in "virtual personal assistants" (Siri, Google Assistant and Google speech queries, Alexa, Cortana, etc.)?
  • Deblackboxing the levels and layers in speech recognition/NLP applications (Siri, Alexa, etc.) by using the design principles method.

Readings and Background on Applications

  • "Virtual Assistant," Wikipedia background.
  • Amazon Lex: Amazon's description of the Natural Language Understanding (NLU) service that Amazon uses (for Alexa and product searches) and also markets as an AI product for other companies on the Amazon Web Services (AWS) Cloud platform.
    • Amazon Lex (like Siri) is known as a "chatbot" (or simply "Bot"), i.e., a "software agent" designed in levels and layers to interact through speech recognition-NLP and hand-offs to multiple layers of background data processes, which are returned (client-server) to a user initiating requests and receiving answers or information.
    • Amazon's Announcement of Lex: how it works for conversational voice and text interfaces.
    • How Lex can be integrated with other platforms ("Utterance Monitoring") (Amazon).
  • Google Assistant: Wikipedia background [mostly business and product information]
    • Google Assistant: Google Homepage [company info and app integrations]
    • Google's Patent Application for "The intelligent automated assistant system" (US Patent Office)
      [Scroll down to read the description of the "invention", and/or download the whole document.]
      [Patents are good sources for the design principles and a company's case for why their design (intellectual property) is distinctive and not based on "prior art" (already known and patented designs).]
    • Abstract of the system in the patent (same general description as for Siri):
      "The intelligent automated assistant system engages with the user in an integrated, conversational manner using natural language dialog, and invokes external services when appropriate to obtain information or perform various actions. The system can be implemented using any of a number of different platforms, such as the web, email, smartphone, and the like, or any combination thereof. In one embodiment, the system is based on sets of interrelated domains and tasks, and employs additional functionally powered by external services with which the system can interact."
    • Google Assistant for Developers (Google)
  • Apple Siri: Wikipedia background [note prior development of the system before Apple]
    • Apple's Patent Application for "An intelligent automated assistant system" (US Patent Office, 2011)
      [Scroll down to read the description of the "invention", and/or download the whole document.]
      [Note the block diagrams for the layers of the system.]
      [There are now many patent lawsuits going on for "prior art" in Siri and speech recognition systems.]
    • Abstract of the system in the patent (same general description as Google):
      "An intelligent automated assistant system engages with the user in an integrated, conversational manner using natural language dialog, and invokes external services when appropriate to obtain information or perform various actions. The system can be implemented using any of a number of different platforms, such as the web, email, smartphone, and the like, or any combination thereof. In one embodiment, the system is based on sets of interrelated domains and tasks, and employs additional functionally powered by external services with which the system can interact.
    • Apple's WIPO Patent, "Intelligent assistant for home automation" (2015).
    • Apple Machine Learning Journal (1/9, April 2018): "Personalized 'Hey Siri'."
    • Apple Machine Learning Journal: "Hey Siri..." [The design of Apple's "Neural Net" Speech Recognition system]
    • Siri uses Wolfram Alpha for "knowledge base" (or "answwer engine") answers

Going Further in Natural Language Processing Backgrounds (Video) [optional]

De-blackboxing the Virtual Assistant, Speech Recognition/NLP and Database System [presentation]


Other Applications [to be discussed in class]

  • AI, Algorithms, and Everyday (Computing) Life:
  • Image analysis and Face Recognition (for next week on Ethical dimensions)

Critique [New]

Weekly writing and reflection (link to Wordpress site)

  • Using the background from this week and the concepts that you've learned so far, describe and explain as many of the levels and layers in the design of speech-activated services ("virtual assistant" applications). Choose one to investigate (Siri, Alexa, Google, Amazon Lex), not for the specifics of the proprietary business branding, but for the underlying design principles that are implemented in all speech recognition/NLP systems for accessing data and services. This will take you into the layers and levels of a complex system design. You can use the patent application descriptions, and as many layers as you can discover/uncover will be great to have learned.
  • The deblackboxing will at first be difficult because there is not much open source published information on the systems. So we need to use our conceptual design principles tools to "reverse engineer" how the systems must be designed to work they way they do. If you get stuck, you can use "reverse engineering" as a thought experiment (i.e., make a new version of a designed thing, not by using the plans of a branded product but creating a model of the required technologies that must be combined and managed to make it work): if we were going to design and build a service app like Siri or Alexa, what would it take to do it? Inside the black box: what unobservable (invisible) layers of technologies and design principles are required to explain what we do observe?

Learning Objectives and Main Topics:

This unit will provide a top-level overview of the ethical, legal, and government policy issues for the development and application of AI/ML in everyday uses, both in consumer applications and devices and deeper black box issues.

The wide deployment of AI applications (ML, Deep Learning, Face Recognition, Speech Recognition systems) by businesses and governmental entities has provoked important ethical questions about AI and all uses of data. We will explore some of the main issues and learn via case studies of AI applications, both in everyday life and behind-the-scenes "invisible" technologies that effect privacy, security, surveillance, and human agency.

The starting point for ethics in relation to any technical system is committing to truthful, de-blackboxed, open and accessible descriptions, definitions, and explanations. There is no magic, but much of the discourse for AI/ML is hype that keeps everything blackboxed and like magic in the service of commercial (corporate) interests. Any definition or conception of an "ethics for AI" must begin with truthful definitions and descriptions. Here's where our three core methods for truthful de-blackboxing can come to the rescue:

  • The Complex Systems View: The Human Response to Modeling and Interpreting Complexity
  • The Design View of Complex Systems: AI/ML Implemented as Designed Interdependent Systems (composed of Subsystems, Modules, and Levels [Layers]
  • The Semiotic Systems View: Computational and AI Systems as Semiotic-Cognitive Artefacts

Intro Readings and Backgrounds

Philosophy and Theory for Ethics and Human Rights in AI

  • Philip E. Agre, “Toward a Critical Technical Practice: Lessons Learned in Trying to Reform AI,” in Social Science, Technical Systems, and Cooperative Work: Beyond the Great Divide, ed. Geoffrey Bowker et al. (Mahwah, N.J: Psychology Press, 1997), 131-58.
    • Though this essay considers AI research as practiced earlier, most of Agre's argument is urgently applicable today. The continued accelerated rush for AI/ML products, business applications, and competition in algorithm design has obscured the fundamental ideas, questions, and human capabilities that enable "AI" as a practice and discipline with a long cumulative history of knowledge and assumptions.
  • Michael Anderson and Susan Leigh Anderson, eds., Machine Ethics (New York: Cambridge University Press, 2011). Selections.
    • This collection of essays is interesting because it features some of the prominent thinkers in philosophy and logic on technology. But the conceptual framing and presuppositions in most of the essay confusing (confused?), and you need to be aware of the different kinds of discourse going on. Read the Introduction, and Chaps. 1 and 4 for this week.
    • "Machines" (here including computational systems and robots) are artefacts of human design (designed in various communities of practice, knowledge, science, and technology), and ethics (principles of human value, beliefs, moral standards) are human-community concepts for promoting desired actions and behavior and regulating undesired actions and behavior, which may be called the social consequences of ideas and beliefs. In much of the discourse here, there is a slippery slope of reification and autonomy, in which humanly designed cognitive artefacts are treated as totalized things (i.e., black boxes). Can you think through how to challenge and correct these views?

Ethics, Policy, and Law: Industry, Corporate, and Governmental Issues


Group Research Project: Planning

  • We will begin discussion in your study groups in class for planning your group research project

Weekly writing and reflection (link to Wordpress site)

  • There are many ethical, political, and ideological issues surrounding AI being discussed now; identify what you think are 1 or 2 important issues and explain why. Use your deblackboxing skills to critique what is being discussed, and also to untangle and expose false, alarmist, or misunderstood ideas about AI, data, and computing systems.

Learning Objectives:

Learning the basic design principles and main architecture of Cloud Computing:

  • "Software as a Service" (SaaS)
  • "Platform as a Service" (Paas)
  • "Infrastructure as a Service" (Iaas)
  • "Virtualization" of server systems, scalable "on-demand" memory

"The Cloud" architecture: a model for integrating the "whole stack" of networked computing.

The design principles for Cloud computing systems extend the major principles of massively distributed, Web-deliverable computing services, databases, data analytics, and, now, AI/ML modules. Today, a simpler question for the ways we use the Web and Internet data might be "what isn't Cloud Computing"?

The term "Cloud" began as an intentional, "black box" metaphor in network engineering for the distributed network connections for the Internet and Ethernet (1960s-70s). The term was a way of removing the complexity of connections and operations (which can be any number of configured TCP/IP connection in routers and subnetworks) between end-to-end data links. Now the term applies to the many complex layers, levels, and modules designed into online data systems mostly at the server side. The whole "server side" is "virtualized" across hundred and thousands of fiber-optic linked physical computers, memory components, and software modules, all of which are designed to create an end product (what is delivered and viewed on screens and heard through audio outputs) that seems like a whole, unified package to "end users."

An Internet "Cloud" Diagram: What happens "inside" a Cloud is abstracted away from the connections to and from the Cloud: only the "outputs" and connections to the Cloud as a system need to be known.

Learning the design principles of "Cloud Computing" is an essential tool in our de-blackboxing strategy. Many of the computing systems that we are studying -- and use every day -- are now integrated on platforms (a systems architecture for data, communications, services, and transactions) designed for convergence (using combinatorial principles for making different systems and technologies interoperable) for data, information, and AI/ML data analytics. For organizations and business on the supply-side of information and commercial services, subscribing to a Cloud Service provides one bundle or suite of Web-deliverable services that can be custom-configured for any kind of software, database, or industry-standard platform (e.g., the IBM, Amazon AWS, and Google Cloud services).

Internet-based (or Internet-deliverable, Internet-distributed) computing continues to scale and extend to many kinds of online and interactive services. Many services we use every day are now managed in Cloud systems with an extensible "stack" architecture (levels/layers) all abstracted out of the way from "end users" (customers, consumers) -- email, consumer accounts and transactions (e.g., Amazon, eBay, Apple and Google Clouds for data and apps), media services (e.g., Netflix, YouTube, Spotify), and all kinds of file storage (Google app files) and platforms for Websites, blogs, and news and information.


Major Cloud Service Providers: Main Business Sites

In-class discussion: Strategies for Developing Group Projects

Weekly writing and reflection:
choose one topic to focus your learning this week (link to Wordpress site)

  • The Cloud system model provides ways to combine and integrate the "whole stack" of computing in the Internet/Web interactive client/server model. At any point of view (especially end "user"), a Cloud system (like Amazon AWS and Google's Cloud) is a complex black box full of intentionally engineered black boxes that "abstract away" the complexity so that for a user/customer the screen+output "presentation layer" seems like a transparent, seamless, unified service, with all processes and transactions handled behind the scenes. AI/ML modules and data service layers are now becoming a routine part of the Cloud "bundle". Based on your background so far and this weeks readings, identify one or two main points of convergence in the design and use of AI/ML and Data systems implemented in the Cloud architecture ("Virtual Assistants," speech recognition, and Web/Mobile translation apps are all Cloud-based systems), and map out for yourself how the modules and layers/levels are designed for combination.
  • The Cloud architecture (although a well-known international standards-based model for system integration in layers and modules) is operationally available only through a subscription and build-out of services with an account on one of the major Cloud service provider companies. Can you think through some of the consequences -- positive and negative -- in the convergence of the technologies on one overall "unifying" architecture (system design) provided by only by one of the "big four" companies (Google, AWS, IBM, Microsoft)?

Learning Objectives and Main Topics:

In this unit, students will learn the basic design principles at work in the converging "platforms" for data and databases, Internet/Web accessible services, AI, and data analytics.

Our current "data environment" is shaped by many kinds of technologies that are managed in multiple levels of interoperable Internet/Web-accessible data of all kinds. This includes the background layers of Internet and Web programming (for active and interactive networked client/server applications), AI/ML techniques for data, Cloud Computing for provisioning all levels of computing-and-networking-as-a-service (OS, platform, software, databases, real-time analytics, and memory storage), and Internet of Things (IoT) (IP-connected devices, sensors, and remote controllable actions).

This environment forms our biggest, complexly layered "black box", comprised of hundreds of subsystem and modular black boxes inside black boxes, all orchestrated for what we call "big data" functions. "Big data" just means massive amounts of data generated from multiple sources (human and human-designed computational agents) and stored in massive arrays of memory accessible to software processes at other levels in the whole system. The main dependencies of both AI/ML expansion and "big data" are cheap, modular memory, fast multiple core processors for parallel and concurrent processing, and fast ubiquitous IP-connected networks (both wired and wireless).

In studying a technology reified with a name (like "AI," or "Big Data") we must always immediately remove the name and consider the system of dependent technologies without which the top-level "black boxes" would be impossible. These are all complex systems, designed to scale with massive modularity and precise attention to levels of operation. Our access point to understanding this kind of system is always through opening up the design principles, and always keeping the human-community designs in focus, especially when confronting the torrents of marketing and ideological discourse from the business and technical communities for the technologies branded and marketed as products.


  • "Big Data": ACM Ubiquity Symposium (2018). Read (html and pdf versions available):
  • Possible selections from the following:
  • Kitchin, Rob. The Data Revolution: Big Data, Open Data, Data Infrastructures and Their Consequences. London; Thousand Oaks, CA: SAGE Publications, 2014.
  • Tarleton Gillespie, Custodians of the Internet: Platforms, Content Moderation, and the Hidden Decisions That Shape Social Media (New Haven: Yale University Press, 2018). (selections)
  • Siva Vaidhyanathan, Antisocial Media: How Facebook Disconnects Us and Undermines Democracy (New York: Oxford University Press, 2018).
  • Shoshana Zuboff, The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power (New York: PublicAffairs, 2019).
  • Safiya Umoja Noble, Algorithms of Oppression: How Search Engines Reinforce Racism (New York: NYU Press, 2018). Selections.
  • Cathy O’Neil, Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy (New York: Crown, 2016). Selections.

Consequences and Philosophical Questions

In-class discussion: Developing plans for Group Projects


Weekly writing and reflection (link to Wordpress site)


Learning Objectives:

This class meeting will be devoted to reviewing the topics and learning path in the course, and returning to the main philosophical and ethical issues in AI/ML and data science.



Critique and Analysis of Current Descriptions of AI/ML

  • Zachary C. Lipton and Jacob Steinhardt, “Troubling Trends in Machine Learning Scholarship,” ArXiv:1807.03341 [Cs, Stat], July 9, 2018. Presented at ICML 2018: The Debates.
    • This is a very enlightening article. The authors present analyses of some of the rhetorical mistakes and a critique of discourse used in describing current work in AI/ML -- from an insider's view.
    • The analyses are also very relevant for anyone wanting to understand what is going on in this field, develop clear and truthful explanations, and critique non-explanations and mystification.
    • The authors are part the Machine Intelligence Research Institute (MIRI) at Berkeley. [View the site to see the kind of research going on at MIRI.]

Weekly writing and reflection (link to Wordpress site)

Learning Objective:

  • Student groups will report on their research projects.



  • To be determined for case studies and applications.

Weekly writing and reflection (link to Wordpress site)