Every route below teaches the same method: specify what correct means, compare against evidence produced independently of the thing you are checking, and say what your check does and does not cover. They differ in where they stop, not in what they teach. Choose one on the course hub and the site will follow it.
The short route
6 units, 250 minutes (about 4.2 hours). video 62, reading 91, activity 97. The video and the reading cover the same ground, so most learners spend one of those two rather than both.
Who it suits. You want the course's actual skill and cannot give it the time the core route takes. You keep the part where you build a reference and check something against it, and you give up the parts about writing the work up and defending it.
What you will miss. Specification and re-prompting, the disclosure and defence you would submit, everything about choosing and working with these systems, grounding and citation checking, high-stakes use and privacy, and everything the core route also leaves out, including the closing argument about where the observer stands.
- Unit 0: Introduction and how to use this course — 33 min: video 8, reading 15, activity 10
- Unit 2: Ada's ordered procedure — 30 min: video 8, reading 10, activity 12
- Unit 3: What a language model does differently — 42 min: video 12, reading 14, activity 16
- Unit 6: Meet the Bernoulli numbers — 44 min: video 10, reading 14, activity 20
- Unit 7: Ask AI to generate the routine — 33 min: video 9, reading 12, activity 12
- Unit 8: Verification lab — 68 min: video 15, reading 26, activity 27
The core route (the default)
14 units, 620 minutes (about 10.3 hours). video 161, reading 217, activity 242. The video and the reading cover the same ground, so most learners spend one of those two rather than both.
Who it suits. Everybody, unless you have a specific reason to be on another path. It assumes no computer science and no programming, it covers what these systems are for and how to work with them as well as how to check them, and it is the path the time budget is set for.
What you will miss. Agents and automated actions, rights and attribution, representation and accessibility, the historical bug, and the closing second-order argument. The optional extensions and the Python lab work are also outside it.
- Unit 0: Introduction and how to use this course — 33 min: video 8, reading 15, activity 10
- Unit 1: What generative AI is for — 34 min: video 10, reading 12, activity 12
- Unit 2: Ada's ordered procedure — 30 min: video 8, reading 10, activity 12
- Unit 3: What a language model does differently — 42 min: video 12, reading 14, activity 16
- Unit 4: Choosing the tool — 36 min: video 10, reading 12, activity 14
- Unit 5: Prompts, context and attachments — 34 min: video 9, reading 11, activity 14
- Unit 6: Meet the Bernoulli numbers — 44 min: video 10, reading 14, activity 20
- Unit 7: Ask AI to generate the routine — 33 min: video 9, reading 12, activity 12
- Unit 8: Verification lab — 68 min: video 15, reading 26, activity 27
- Unit 9: Grounding and citation checking — 65 min: video 12, reading 17, activity 36
- Unit 10: Re-prompting, specification, and overreliance — 58 min: video 13, reading 16, activity 29
- Unit 12: High-stakes use and escalation — 34 min: video 9, reading 11, activity 14
- Unit 13: Privacy and data handling — 33 min: video 9, reading 12, activity 12
- Unit 17: Defence, disclosure, and responsible use — 76 min: video 27, reading 35, activity 14
The core route with the code work
14 units, 620 minutes (about 10.3 hours). video 161, reading 217, activity 242. The video and the reading cover the same ground, so most learners spend one of those two rather than both.
Who it suits. You write code, or you are on a computing programme and want the version that engages with testing rather than only with judgement. Same units as the core path, taken through the Python route with the lab extensions. Choosing this route does not switch the checkpoints over -- set How you verify to the Python route on your progress page. The time shown is the core path's, because no unit declares minutes for the Python route or the extensions, so allow longer. Recommended if you program, required of nobody.
What you will miss. The same units the core path misses. The code work teaches no new verification idea; it gives a second way to practise the same one, with the machine doing the comparison.
- Unit 0: Introduction and how to use this course — 33 min: video 8, reading 15, activity 10
- Unit 1: What generative AI is for — 34 min: video 10, reading 12, activity 12
- Unit 2: Ada's ordered procedure — 30 min: video 8, reading 10, activity 12
- Unit 3: What a language model does differently — 42 min: video 12, reading 14, activity 16
- Unit 4: Choosing the tool — 36 min: video 10, reading 12, activity 14
- Unit 5: Prompts, context and attachments — 34 min: video 9, reading 11, activity 14
- Unit 6: Meet the Bernoulli numbers — 44 min: video 10, reading 14, activity 20
- Unit 7: Ask AI to generate the routine — 33 min: video 9, reading 12, activity 12
- Unit 8: Verification lab — 68 min: video 15, reading 26, activity 27
- Unit 9: Grounding and citation checking — 65 min: video 12, reading 17, activity 36
- Unit 10: Re-prompting, specification, and overreliance — 58 min: video 13, reading 16, activity 29
- Unit 12: High-stakes use and escalation — 34 min: video 9, reading 11, activity 14
- Unit 13: Privacy and data handling — 33 min: video 9, reading 12, activity 12
- Unit 17: Defence, disclosure, and responsible use — 76 min: video 27, reading 35, activity 14
The route without the mathematics
11 units, 475 minutes (about 7.9 hours). video 127, reading 165, activity 183. The video and the reading cover the same ground, so most learners spend one of those two rather than both.
Who it suits. You do not want to derive Bernoulli numbers by hand and would rather not meet summation or binomial coefficients at all. You still read about Ada's procedure, because it is the demonstration the whole course refers back to, and you do the full method on a case with no arithmetic in it -- checking a referenced summary against what its sources actually say.
What you will miss. The Bernoulli derivation, the generated routine and the verification lab (Units 6 to 8), which is where the method is proved on a case with a decidable answer rather than described. You get the method, on a case where the judgements are yours -- and you should know that is the trade you have made. It also misses the units the core path misses.
- Unit 0: Introduction and how to use this course — 33 min: video 8, reading 15, activity 10
- Unit 1: What generative AI is for — 34 min: video 10, reading 12, activity 12
- Unit 2: Ada's ordered procedure — 30 min: video 8, reading 10, activity 12
- Unit 3: What a language model does differently — 42 min: video 12, reading 14, activity 16
- Unit 4: Choosing the tool — 36 min: video 10, reading 12, activity 14
- Unit 5: Prompts, context and attachments — 34 min: video 9, reading 11, activity 14
- Unit 9: Grounding and citation checking — 65 min: video 12, reading 17, activity 36
- Unit 10: Re-prompting, specification, and overreliance — 58 min: video 13, reading 16, activity 29
- Unit 12: High-stakes use and escalation — 34 min: video 9, reading 11, activity 14
- Unit 13: Privacy and data handling — 33 min: video 9, reading 12, activity 12
- Unit 17: Defence, disclosure, and responsible use — 76 min: video 27, reading 35, activity 14
The responsible-use route
8 units, 351 minutes (about 5.8 hours). video 95, reading 122, activity 134. The video and the reading cover the same ground, so most learners spend one of those two rather than both.
Who it suits. You want the duties rather than the mechanics -- consequence and escalation, what leaves your device, who owns the output, and what a generated summary does to the people it describes. Short, and meant to be taken beside another path rather than instead of one.
What you will miss. Most of how these systems work, and the worked case on code. It keeps the citation checking the rights and representation units rely on. This is why it is a companion path rather than a whole one.
- Unit 0: Introduction and how to use this course — 33 min: video 8, reading 15, activity 10
- Unit 3: What a language model does differently — 42 min: video 12, reading 14, activity 16
- Unit 9: Grounding and citation checking — 65 min: video 12, reading 17, activity 36
- Unit 12: High-stakes use and escalation — 34 min: video 9, reading 11, activity 14
- Unit 13: Privacy and data handling — 33 min: video 9, reading 12, activity 12
- Unit 14: Copyright and attribution — 29 min: video 8, reading 9, activity 12
- Unit 15: Bias, representation and accessibility — 39 min: video 10, reading 9, activity 20
- Unit 17: Defence, disclosure, and responsible use — 76 min: video 27, reading 35, activity 14
The complete route
19 units, 821 minutes (about 13.7 hours). video 221, reading 288, activity 312. The video and the reading cover the same ground, so most learners spend one of those two rather than both.
Who it suits. You want everything the course offers -- all nineteen units, both worked cases, the optional extensions and the lab work. It is meant to be taken over several sittings rather than at once.
What you will miss. Sustainability beyond the resource section in the tool-choice unit, and industry practice, both of which are signposted rather than taught. Institutional assessment policy, employer policy and jurisdiction-specific legal advice are out of scope by decision -- the course states the boundary and tells you to check your own rules.
- Unit 0: Introduction and how to use this course — 33 min: video 8, reading 15, activity 10
- Unit 1: What generative AI is for — 34 min: video 10, reading 12, activity 12
- Unit 2: Ada's ordered procedure — 30 min: video 8, reading 10, activity 12
- Unit 3: What a language model does differently — 42 min: video 12, reading 14, activity 16
- Unit 4: Choosing the tool — 36 min: video 10, reading 12, activity 14
- Unit 5: Prompts, context and attachments — 34 min: video 9, reading 11, activity 14
- Unit 6: Meet the Bernoulli numbers — 44 min: video 10, reading 14, activity 20
- Unit 7: Ask AI to generate the routine — 33 min: video 9, reading 12, activity 12
- Unit 8: Verification lab — 68 min: video 15, reading 26, activity 27
- Unit 9: Grounding and citation checking — 65 min: video 12, reading 17, activity 36
- Unit 10: Re-prompting, specification, and overreliance — 58 min: video 13, reading 16, activity 29
- Unit 11: Agents and automated actions — 37 min: video 9, reading 14, activity 14
- Unit 12: High-stakes use and escalation — 34 min: video 9, reading 11, activity 14
- Unit 13: Privacy and data handling — 33 min: video 9, reading 12, activity 12
- Unit 14: Copyright and attribution — 29 min: video 8, reading 9, activity 12
- Unit 15: Bias, representation and accessibility — 39 min: video 10, reading 9, activity 20
- Unit 16: The historical bug and historical honesty — 34 min: video 9, reading 13, activity 12
- Unit 17: Defence, disclosure, and responsible use — 76 min: video 27, reading 35, activity 14
- Unit 18: Alles Lüge! First- and second-order reasoning — 62 min: video 24, reading 26, activity 12