Week 1 — Intro to AI, Course Logistics & Historical Foundations

Week 1 – Course On-Ramp

Class ambience
• 1st session of the 12-week “Professional Diploma in Artificial Intelligence for Business” (UCD).
• 26-30 remote learners from diverse industries: finance, pharma, media, telecoms, aviation, retail, engineering, etc.
• Lecturer: Elton Mungani – data-scientist & master-data/business-process manager (ex Coca-Cola, currently pharma).
• Class duration each Tuesday ≈ 2.5–3 h; Zoom recording released $48\text{–}72\,\text{h}$ later on the LMS.

Lecturer’s expectations

Professional environment → self-directed reading + peer knowledge-exchange.
Sessions mix theory + tool demos ("1–2 tools per class").
Students should:
• Actively post in Week-Discussion forum (ideas, links, tools).
• Reflect weekly for the Learning Log rather than cramming at the end.
• DO NOT hand-in AI-written assessments (plagiarism system + ethics).

Course Infrastructure (LMS walk-through)

Navigation bar
• Announcements (quick panel top-right).
• Student Handbook – rules, appeals, academic integrity.
• Assessment Overview – brief; detailed rubrics further down.
• “Contact Us” form – use for admin issues (deadline extension, enrolment queries, personal matters).
• FAQ page.
• “Join live session” link (Zoom) + “Recorded Session” (playback).
• “Session X Handout” – slide PDF for pre-reading.
• Additional Reading list (articles, standards, videos).
• Week-Discussion Forums (public peer Q&A, resource sharing).
• End-of-course satisfaction survey (optional).
Assessments

Assignment 1 – Action Learning Log
• $1{,}500\;\text{words}\pm10\%$ covering reflections for all 12 units.
• Submission date: $17\,\text{Jul}\,2025$ .
• Grading: Pass / Fail (must Pass to unlock final grade).
• Strategy: write reflection immediately after every class; use learning-outcome bullets as skeleton.
Assignment 2 – Business Report
• $3{,}000\;\text{words}\pm10\%$ .
• Due: $29\,\text{Jul}\,2025$ .
• Marking rubric ⇒ Pass / Merit / Distinction; emphasises depth, critical analysis, original insight, Harvard referencing.

Turnitin originality check embedded; University ethics policy applies.

Unit 1 – Introduction to Artificial Intelligence

Official Learning Outcomes (map each to Learning-Log)

Define core AI concepts/terminology + articulate AI’s role in digital transformation.
Examine capabilities of 3 common AI types & business use-cases.
Analyse linkage between business goals & technical capabilities.
Evaluate the People–Process–Technology balance.
Assess ethical, social, legal & governance considerations.

Working Definition (course)

“AI deals with methods that enable a computer to solve tasks that, when solved by humans, require intelligence.”

Inspiring quotes

Andrew Ng: “AI is the new electricity.”
Sundar Pichai: “More profound than fire or electricity.”

Concept Map – AI & Its Sub-Fields

Artificial Intelligence (AI) – umbrella of computational techniques that emulate human cognitive functions (perception, reasoning, learning, decision-making).
• Machine Learning (ML) – algorithms that learn patterns/relationships from data, improving over time without explicit re-programming.
• Deep Learning (DL) – ML subset using multi-layered artificial neural networks (ANNs); excels at images, speech, text.
• Natural Language Processing (NLP) – enabling machines to parse, interpret & generate human language.
• Computer Vision (CV) – interpreting visual data; object detection, face ID.
• Robotics & Automation – embodied AI executing physical tasks autonomously.
• Expert Systems – rule-based inference engines (IF–THEN); earliest commercial AI.
• Cognitive Computing – systems mimicking human reasoning to tackle ambiguous problems.

Demystifying Jargon

Heuristics – rule-of-thumb shortcuts to limit search space (e.g., “look for lost keys on tables first”).
Neural Network Anatomy
• Inputs $xi$ weighted $wi$ ; compute dot-product $\sum wi xi$ ; pass through activation $\varphi(\cdot)$ → output.
Reinforcement Learning (RL) – agent $\rightarrow$ environment, receives state, action, reward; learns optimal policy $\pi^*(s)$ .

Historical Trajectory

Antiquity ➜ Pre-Computer

$\sim200\,\text{BCE}$ Antikythera Mechanism – geared analog “computer” forecasting eclipses ⇢ illustrates ancient desire to offload cognition.
18th century Mechanical Duck (Automaton) – flapped, quacked, “digested”; shows public fascination with life-like machines & early anthropomorphism (attributing human traits to machines).

19th Century

Charles Babbage – Analytical Engine blueprint (first general computer).
Ada Lovelace – wrote first algorithm; envisioned machines composing music → seed of generative AI.

1940s

Isaac Asimov – “Three Laws of Robotics” ⇒ birth of AI ethics discourse.
George Pólya – heuristic problem-solving framework (basis of search strategies).
Donald Hebb – Hebbian learning “neurons that fire together wire together.”
Alan Turing – Turing Test / Imitation Game; theoretical foundation for digital computers; asked “Can machines think?”

1950s – Birth of the Field

Term “Artificial Intelligence” coined (John McCarthy, 1956 Dartmouth workshop).
Claude Shannon – laid information-theory basis; chess-playing program.
Marvin Minsky & Dean Edmonds – SNARC neural-network simulator (rat in maze).
Frank Rosenblatt – Perceptron (single-layer ANN).
• Weight-update rule: $\Delta wi = \eta (y-\hat y) xi$ ; guaranteed convergence for linearly separable classes.
• Video demo: 20×20 pixel camera classifying circles vs rectangles.

1960s – Language & Expert Systems dawn

ELIZA (Joseph Weizenbaum, 1964) – first NLP chatbot (Rogerian psychotherapist).
• Triggered public tendency to trust machines emotionally.
Early Expert Systems (e.g., DENDRAL, MYCIN) encode domain rules; architecture: Knowledge Base + Inference Engine + UI; struggled with ambiguity.

1970s–80s – AI Winter & Privacy Concerns

Arthur Miller warns of computerised surveillance & data privacy.
Reports (e.g., UK Lighthill Report, 1973) slash funding → “AI Winter.”

1990s – Revival via Narrow Successes

Deep Blue vs Garry Kasparov (1997) – chess grandmaster defeated; rekindled investment.
Speech-to-text (Dragon NaturallySpeaking) enters market.

2000s – Embodied & Data-Driven AI

Honda ASIMO (2000) – biped robot.
iRobot Roomba (2002) – domestic autonomy.
Netflix Prize (2006) – $\$1\,000{,}000$ challenge advancing collaborative filtering.

2010s – Deep Learning Era

Smartphones: Siri (2011), Google Now, Cortana.
Human Brain Project launches; raises “strong-AI” debates.
Campaign to Stop Killer Robots (2013).
Eugene Goostman chatbot (2014) claims Turing Test pass (controversial).
AlphaGo vs Lee Sedol (2016) – deep reinforcement learning conquers Go (branching factor $\gt 10^{360}$ ).
Google Duplex (2018) – phone calls with human-like filler “um” and “uh.”
GDPR enacted (2018) – transparency & consent for data used in ML.

2020s – Foundation-Model Boom

OpenAI GPT-3 (2020), ChatGPT (2022) → fastest product to $100\,\text{M}$ users.
EU AI Act drafted; US, UK, UN frameworks; on-going debate: alignment, bias, existential risk.

Deep Learning Mechanics (high-speed tour)

Training prerequisites
• Huge labelled datasets (web scrape, sensors, logs).
• Parallel compute (GPU / TPU cloud).
Typical feed-forward layer operation: $a^{(l+1)} = \varphi(W^{(l)} a^{(l)} + b^{(l)})$ .
CNN, RNN, Transformer architectures cover vision, sequence & language tasks.

Expert Systems vs Modern ML

Feature	Expert System	Machine / Deep Learning
Knowledge source	Human-encoded IF–THEN	Data-derived weights
Pros	Transparent rules, easy auditing	Handles ambiguity, non-linear patterns
Cons	Brittle, hard to scale, rule conflict	Opaque “black box,” data-hungry
Sample domain	Medical diagnosis (MYCIN)	Fraud detection, CV, LLM

Ethics, Governance & Risk Themes (recurring)

Accountability: who is liable when model errs? human vs vendor vs model.
Bias & Fairness: training data skews propagate to decisions (hiring, credit).
Privacy & Surveillance: large-scale data scraping, facial recognition, targeted ads.
Autonomy & Control: lethal autonomous weapons, runaway optimisation.
Regulation Landscape: GDPR $\rightarrow$ EU AI Act; ISO/IEC 23894 AI risk mgmt; NIST AI RMF.

In-Class Collaborative Activity – Image Generation Demo

Task: each breakout group used non-ChatGPT/Gemini tools to create an AI image; reported:
• Tools: Freepik AI, Adobe Firefly, FacePic, DeepSeek, Bing Image Creator, Midjourney (not demoed), DALL·E.
• Observations:
– Prompt engineering critical; minor wording shifts huge output change.
– Different platforms show distinct stylistic bias tied to training sets (stock vs artistic).
– Failure cases: asked for “Eiffel Tower in Dublin” → got sunny Paris; submarine-cable query produced irrelevant images.
– Generation latency: Gemini-Flash ≈15 s vs others minutes.
Class repository: students to post tool links & sample prompts in Week-1 forum.

Practical Tips for Week 1 Learning-Log Entry

Mirror the 5 Learning-Outcomes headings; jot notes on:
• Personal definition of AI; stance on “Can machines think?”
• Favourite historical milestone & why business-relevant.
• People–process–technology insights from LMS discussion (e.g., assignment workflow, community forums).
• Early ethics reflection: privacy, anthropomorphism, AI-generated scams (deep-fake money transfer).
• Tool experimentation: which image generator you tried, prompt iterations, potential business value.
Keep citations: lecture slides, video clips (Perceptron, ELIZA, Turing Test), class discussions.

Forthcoming Units (teaser)

Digital Transformation – process re-engineering.
Process Automation – RPA + AI.
Cognitive Insight & Engagement – analytics + conversational AI.
… up to Unit 12 capstone.

“Week 1 = foundation; build literacy, network with peers, secure LMS access, and capture reflections now – future-you will thank present-you.”