Itd AI guidance

[Poonam-raj]

This PR will

• Update small Slack and ITD references that are out of date.
• Extend and amend Step Three instructions to reflect the wider CYF Ai Guidance.
• remove answer 8 from the third question in the Step Three task
• Amend step 6 and 7 where AI is encouraged to make it more generic to the internet whilst still exploring AI, also referencing step three more.

Areas review is especially needed

• Whether step six and step seven still push AI in a way that doesn't align with our AI guidelines. Is it pushing it too hard? Or is it enabling exploration to inform trainees?
• Do the questions in step three align with AI guidance, does anything stand out as misleading? (I was thinking of making these multiple choice elements but then realised a submission relies on it going into a google doc so couldn't. Would love a way of getting feedback to a student and make step three have a submission that makes sense at the same time.)
• Does the opening explanation in step three ramble too long and lose the trainee before getting to the "acceptable" criteria?

[netlify]

✅ Deploy Preview for cyf-curriculum ready!

Name	Link
🔨 Latest commit	3f4c9eb
🔍 Latest deploy log	https://app.netlify.com/projects/cyf-curriculum/deploys/6a70c929a249db00086fa249
😎 Deploy Preview	https://deploy-preview-1950--cyf-curriculum.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.
Lighthouse	1 paths audited Performance: 90 (🟢 up 5 from production) Accessibility: 100 (no change from production) Best Practices: 92 (no change from production) SEO: 100 (no change from production) PWA: - View the detailed breakdown and full score reports
🤖 Make changes	Run an agent on this branch

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[netlify]

✅ Deploy Preview for cyf-common ready!

Name	Link
🔨 Latest commit	3f4c9eb
🔍 Latest deploy log	https://app.netlify.com/projects/cyf-common/deploys/6a70c92a6ede6a0008e22f77
😎 Deploy Preview	https://deploy-preview-1950--cyf-common.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.
🤖 Make changes	Run an agent on this branch

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[Poonam-raj]

Related ticket: #1901

[illicitonion]

This all LGTM, left a few suggestions, the side-by-side CV may be a little work to put together but I think is probably really worthwhile? I've definitely had a lot of experiences of pointing out to a trainee (particularly on ITD when we used to have them generate CVs/cover letters with GenAI) how bad their AI-generated cover letter was

[illicitonion]

I'd be tempted to split this into two sub-sections:

1. We need to build understanding
2. Sometimes it lies

I think the "we need to question and verify and explain" could do with being split into those two sections too:

• In the "building understanding" section we can talk more about how we should be able to spot problems in AI output, and we should be able to delete AI output and write it ourselves because we know how.
• In the "sometimes it lies" section, we can talk a bit about looking up other sources

[illicitonion]

Suggested change

	- Letting AI write your CV
	- Letting AI write something for you (e.g. your CV)

And let's explain why this is bad - that it will generate generic CVs which don't sell you, which aren't personal, that use filler words, that make things up about your background, etc.

Maybe we could even include side-by-side a bad AI generated CV and a good personally written CV, and give trainees a prompt for some criteria to use to compare them, and have them submit their comparison? (On average our ITD trainees don't know what a good CV looks like, so I think we would need to give them criteria to evaluate from)

[Poonam-raj]

Thought dump after two weeks of chewing on how to make AI CV creation make sense

Rn the learning objectives are going in the direction of:

1. Highlight the ways AI can make mistakes or go wrong.
2. Highlight the disadvantages of AI-generated work.
3. Identify good and bad uses of AI for learning during the ITD.
4. Create a Google Doc with answers to questions about AI usage.

Point 2 is where CV comparison could come into play. Issues I'm running into atm:

Upskilling ITD folks to know what a good CV looks like and how to spot issues. A lot of content for a page focused on AI.

Workaround: I am delegating upskilling trainees in what a good CV looks like by pointing to the guide and evaluator tool. It feels like a lot to explain in the prep.

I can currently force AI to hallucinate if I give it a more mid/senior role and a junior "profile" and ask it to generate a CV for it (just getting it to write a CV based on a profile was actually hard to make look bad, common LLMs seem to reasonably make a CV on its own - ChatGPT is fluffier than claude for sure).
Now the issue is, getting people to spot the hallucinations and fluffed up language is hard when they have no idea about the language, ideas, engineering concepts mentioned.

Workaround: I can try to make the profile someone who has warehouse/taxi/shop assistant experience and go for a more customer service/intern/office role. But again hard to tell if people will have enough understanding of the skills to spot the hallucination.

If the core q is "why do I bother writing a CV myself" we can drill in these advantages:

• accuracy in a CV, hallucinations avoided
• English lang skills
• Understanding the contents of the CV for interview fluency

And an alternative good use of AI might be finding out more about what a good structure of a CV would be for a uk-based SE, and a bad use would be "write a CV for me" (and why these are good and bad).

I'm still chewing on this but needed to put a bookmark in where I'm at with this