D429 Introduction to AI for Computer Scientists, catalog number ICSC 3111, is a two competency unit course in the WGU School of Technology. The catalog describes it as an overview of artificial intelligence terminology and key concepts, the history and evolution of the field, and the process for understanding algorithms. Two CUs is small, which fools people. The material is broad rather than deep, and breadth is what makes it slippery: the course asks you to hold a vocabulary steady across sixty years of a field that renamed itself three times. Precision with terms, not cleverness with code, is what gets scored.
Why a two CU survey is harder than it looks
A deep course gives you one thing to understand well. A survey gives you thirty things to place correctly relative to each other, and placement is exactly what an assessment can test cheaply. In ICSC 3111 the recurring failure is not ignorance, it is imprecision: using artificial intelligence, machine learning and deep learning as if they were interchangeable, or describing a rule based expert system as if it learned.
Hold the nesting in your head and most of the course falls into order. Artificial intelligence is the outer set, the general project of getting machines to do things that look like reasoning. Machine learning is a subset in which behaviour is derived from data rather than written by hand. Deep learning is a subset of that, built on layered neural networks. Symbolic and search based approaches sit inside artificial intelligence but outside machine learning entirely, which is why an A star search or a minimax game tree belongs on the AI side of the map without any training data anywhere near it.
The history section is not decoration. The field's evolution explains why the vocabulary is messy: the Dartmouth workshop of 1956 named the discipline, symbolic reasoning and expert systems dominated into the 1980s, funding collapses followed twice, statistical methods took over in the 1990s, and layered networks with large datasets and modern hardware produced the current era. Every one of those turns left terminology behind in the language. Knowing which era a term came from is often the fastest way to answer a question about what the term means.
Turning scored aspects into a study or section plan
WGU publishes scoring detail inside your Course of Study rather than in the public catalog, so open the rubric before you write a line. Each aspect is scored on its own and a score of 2 in each one passes the task. There is no averaging, so a superb history section will not rescue an algorithm section that stops at a definition.
If your version of this course is assessed by written work, treat the aspect list as the outline and use the rubric's own nouns as headings. If it is assessed by a proctored exam, the same list becomes a revision schedule instead, one block per aspect, weighted by how many concepts sit under it.
The word budget, worked. A two CU course usually carries a shorter deliverable than a four CU one. Suppose five scored aspects and directions calling for about 1,500 words. Reserve 100 for an opening that states the scope and 100 for a closing, leaving 1,300 across five aspects, or 260 each. Then rebalance: any aspect that asks you to walk through an algorithm needs room for a trace, so give it 400. Any aspect that asks for definitions or a timeline compresses well, so drop those to 200. The total holds and the length now sits where the marks are.
Watch the opposite failure too. Survey courses tempt students into 800 word history essays because history is easy to write. An overweight section starves a neighbour, and the starved neighbour is what comes back.
A structure that fits an AI concepts deliverable
Where the task directions give you headings, use theirs. Where they do not, this shape keeps the concepts separable and makes each aspect easy to find.
| Section | What belongs in it | How it gets scored |
|---|---|---|
| Scope and definitions | The terms you will use, defined once, with the nesting made explicit | Scored on precision; a definition that would also fit a neighbouring term earns nothing |
| Historical arc | Named periods with what changed and why, not a list of dates | Scored on causation, so each entry needs a because clause |
| Approach comparison | Symbolic, statistical and learned approaches set side by side on stated criteria | Scored on whether the criteria are consistent across all three |
| Algorithm walkthrough | One algorithm traced on a small input, step by step, with the state shown at each step | Scored on the trace being followable, not on the algorithm being impressive |
| Evaluation | How you would judge whether the approach works: accuracy, cost, interpretability, data need | Scored on naming trade offs rather than advantages only |
| Limits and ethics | Bias in training data, opacity of learned models, and where a human decision must stay | Scored on specificity to your example rather than general concern |
| Sources | Textbook, documentation and any papers, APA formatted | Scored where citation is named in the aspect |
The walkthrough row is the one worth over preparing. A trace of breadth first search on a six node graph, showing the queue at every step, demonstrates more understanding than three paragraphs describing what search is.
Evidence craft in a fast moving technical field
Artificial intelligence has a source quality problem that older computing subjects do not. A large share of what is written about it is marketing, and marketing uses the vocabulary loosely on purpose. Academic work has to be pickier.
- Prefer the textbook and the peer reviewed paper for concepts, and official documentation for anything about a specific tool.
- Date every claim about capability. A statement about what models can do is true of a moment, so name the moment and the system.
- Do not cite a vendor blog for a definition. Cite it only as evidence of what that vendor claims, and label it that way.
- Attribute the algorithm. Search, backpropagation and the perceptron all have origin literature, and naming it costs one sentence.
- Keep quoted definitions short and rare. Technical definitions are the easiest text in any submission to match against a source, and WGU runs a similarity check.
- If you use an AI tool while studying, treat its output as an unsourced claim to verify, never as a citation, and follow your course policy on tool use exactly.
One more habit separates strong technical writing here: naming what the approach cannot do. Every method in this course has a failure mode, and stating it beside the strength is the difference between a summary and an evaluation.
What separates Competent from a submission sent back
Aspects are scored independently, so returns tend to be narrow. In a concepts survey the usual cause is a definition that never became an application.
- Every term you use is defined once, early, and used consistently thereafter with no drift.
- The algorithm section contains an actual trace with intermediate state, not a paragraph about the algorithm's purpose.
- Comparisons use the same criteria for every option. Changing criteria mid comparison is the classic tell of a padded section.
- Every claim about performance or capability carries a source and a date.
- The ethics discussion attaches to the specific system under review rather than floating free.
Performance assessment work can be revised and resubmitted without a grade penalty, so a return costs days rather than standing. In a six month flat rate term those days matter: a two CU course is meant to be a two week course, and a return that turns it into a five week course quietly removes a larger course from your term.
Five mistakes that cost time in D429
- Using AI and machine learning as synonyms. This single slip contaminates a whole submission, because every later sentence inherits the confusion.
- Writing history as a timeline. Dates without causes read as copied. Each period needs the reason the previous one stalled.
- Choosing a glamorous algorithm to trace. A hand trace of a transformer is not happening. Pick something small enough to show completely.
- Describing neural networks with metaphors only. The brain comparison is a teaching aid, not an explanation, and evaluators of a computer science course want weights, layers and an update rule.
- Leaving ethics as a closing paragraph of general worry. Bias has a mechanism. Name where in the pipeline it enters and what would detect it.
How support works on this course
Send the rubric from your Course of Study and the task directions. Written work comes back aspect mapped, with the algorithm trace worked out fully and a walkthrough that explains the reasoning so the vocabulary sticks for later courses in the technology sequence.
If your plan puts a proctored objective assessment on this course, the line does not move. Proctored exams are yours to sit. We build a revision plan, drill the term boundaries that this course tests hardest, run practice traces and give an honest go or wait read. We never sit assessments, and we never ask for portal credentials.
Questions students ask about D429
Is D429 the same course as ICSC 3111?
How much programming does D429 assume?
Can you sit my proctored exam for this course?
Working through the AI survey right now?
Send your rubric and the task directions. You get an aspect mapped draft with a full algorithm trace, plus a walkthrough that makes the vocabulary stay put.
Where D429 sits in WGU's programs
The July 2026 catalog places this code in 2 current WGU programs. Open a program page for the complete standard path and term positions. The live Degree Plan remains authoritative after transfer credit, substitutions, and mentor planning.
The assessments, one by one
The public catalog does not publish this course's PA/OA identity or task count. WGU Tutors publishes at most one PA manual per course and only from a WGU-controlled public rubric. Until that source exists, PA help begins from the student's real Course of Study and OA support remains preparation only.