BUS 658

Today’s plan

1. Intro — Organization, introduction.
2. Lab — Using LLMs + Google Scholar to generate and refine research questions
3. Discussion — What are the interesting questions in AI and Econ? (Share from the lab.)
4. Readings, questions for next week.

Grading

Deadlines: Canvas.

Materials: On Github.

Who is Seth Benzell?

• Assistant Professor at Chapman University (Argyros School of Business & Economics)
• PhD (Economics), Boston University, Postdoc at MIT 2017-2020
• Research focus: economics of digitization
  • automation + AI
  • networks + information systems
  • digital platforms
• working on a book on "The Power Law Economy"
• host of the Justified Posteriors podcast

Goals for Class

• Topic: Think about the bigger picture and longer term impact of the digital tools you are learning: specifically AI, Analytics, and Digital Networks

• Tools: Practice reading, discussing, and using LLMs as a learning assistant for foundational and cutting edge tech and econ of tech research

• Deliverables: A chance to further develop portfolio project; a refined appreciation of plausible longer-term career trajectories

Typical Weekly Format

• Do readings in the days leading up to class, assigned week before
• A small assignment due Wednesday: e.g. Developing one good question based on the readings for discussion
• An introductory lecture on the topics of the day
• A lab or breakout session
• A regrouping for a closing discussion

Breaks in between

Logistics

• Meeting: Thursdays 7:00–9:50pm (break included), Beckman 104
• Materials: no textbook; readings posted on the course site

Course roadmap (tentative)

• Unit 1 (Weeks 2–8): AI foundations; automation & the economy; alignment, regulation and the farther future
  • Reflection 1 due
• Unit 2 (Weeks 9–12): Digital platforms + the digital ecosystem
  • platform tech (e.g. APIs, MCPs, crypto); decentralized + AI ecosystems; the Power Law Economy
  • Reflection 2 due
• Weeks 13–14: Project presentations

Before Class

• Submit one question about the reading the day before class.
• Include a link that shows the conversation with the LLM.
• Include a link to an LLM answer to the question you believe is inadequate.

Some Topics I Think are Interesting

A conversation with Bo Cowgill of Columbia University

• The destruction of meaning vs. the coasian singularity
• Algorithmic bias vs. AI omniscience
• The future of data analytics careers

Who is Alexander Kurz?

PhD in Computer Science, University of Munich, 2000
Postdocs in Amsterdam and Brno
Professor of CS, Leicester, UK, 2002-2018
Chapman University, 2018-

Research Areas

Theoretical Computer Science
Theory of Programming Languages, Software Engineering (SWE)
Models of Computation
Formal Methods
Proving Correctness of Programs
Logic and Category Theory, Algebra and Topology

Teaching

Programming Languages
Compiler Construction
Algorithm Analysis
Logical Foundations of Computer Science
Natural Language Processing, Smart Contracts, Wikipedia Governance

Software Engineering (SWE)

SWE Methods for Vibe Coding
Formal Methods and AI
Incomplete Smart Contracts
Local-First AI
Substack Paia, Nearbytes, Etc
Wikipedia Bias (What is Bias?)
Democratic Technologies
Interestingness (in math and beyond)
Philosophy of AI and SWE

Lab: What makes a question interesting?

• Lab: Use LLMs + Google Scholar to elaborate an interesting question
• Discussion: Summarize your conversation at Hackmd

Method of finding an interesting question

Converse with AI until you don't get a good answer anymore. Ask the AI to link the most relevant scholarly resources on the topic. Read some of them.

Example: HI vs AI

I recently had a conversation about the future of humanity in times of AI. Here is a stepping stone I came up with after several iterations, worthy of further development

Brief history of Intelligent Machines:

• McCulloch & Pitts (1943): Neural Networks
• Von Neumann (1944): AI-as-logical-deduction
• Machines are better at I-as-logical-deduction than Hs
• Rosenblatt (1958): AI-as-probabilistic-pattern-recognition
• Machines are better at I-as-probabilistic-pattern-recognition than Hs

What is left for humans?

Hint: 210 reasons are no reason at all

Instructions for the Lab

1. Go to PantherAi, choose a model
2. Ask a question ... Ask a follow up question ... etc
3. Make a mental model of the emerging landscape
4. Ask to inline links to google scholar queries, wikipedia, etc
5. Also read outside of the AI
6. Until (end) go back to 2
7. Make a one slide summary at HackMD
8. Export the dialogue with the AI and share it via Rentry.co

Break (10 min)