Someone asks how to learn AI before finishing university. I do not want the answer to become another long speech that postpones the first attempt.

In the video I offer three ways in: use the providers’ learning material, ask an assistant to help organise your learning, and build something. The third is the one that excites me most, because a project gives every new question somewhere to go.

You do not need a degree to begin using a conversational tool. Becoming a researcher or safely building a complex production system is a different level of work. Starting and mastering are different milestones.

1. Get a map of the basics

These official starting points were checked for this article:

ResourceUseful starting point
OpenAI’s ChatGPT quickstartGetting started with ChatGPT
Google AI EssentialsBeginner course with practical activities; no prior experience required
Claude AcademyAnthropic’s learning resources, including AI Fluency

Pick one that fits your immediate task. Check current access, language and any price before enrolling. A completion certificate records a course; it is not proof that you can solve every AI problem. The video’s rough learning-time estimate and user-count claim are not benchmarks used here.

2. Make the practice fit your life

Here is a prompt I would use as an exercise:

I want to make a simple information page for a small service business. I know the business but I am new to building websites. I can practise for 30 minutes at a time. Give me a short plan in plain language, one task per session, and a way to check each result. Ask for missing information instead of inventing it.

Replace the business and time with your own circumstances. Check any links the assistant suggests. If a step assumes knowledge you do not have, ask it to break the step down. A learning plan is useful when you can actually perform the next action.

Learn, personalise, build. Learn the basics: Use an official introductory resource. Personalise the practice: Give the tool your goal, context and available time. Build and check: Make one small result and test whether it works.
Three routes from the video. Move between them as you learn.

3. Build something small enough to finish

For that example page, write the real service, audience, opening hours and contact route first. Then ask for a structure and a first version. Review it on a phone. Follow every link. Confirm that it has not invented testimonials, prices or qualifications. Use fictional contact details while experimenting.

An illustrative first week could be four 30-minute sessions: define the task, create the first version, check it, improve it. That is 4 × 30 = 120 minutes, or two hours of scheduled practice. It is a time budget, not a promise that you will finish a website or master AI in two hours.

For a public site, learning also continues through hosting, accessibility, privacy and maintenance. A convincing preview does not finish those jobs.

My reflection on AI and expertise makes the same distinction: curiosity starts the work; evidence of what you can do comes from checking it.

Start with one result you can explain. Then make the next one better.

Sources and calculation notes

  • The table links directly to the providers’ resources, checked 27 September 2026. It does not promise identical course lengths, certificates or free access across providers.

  • The prompt and four-session schedule are original editorial examples. Practice-time arithmetic.

  • Original video and thumbnail: Eduard Luta on TikTok. Published 2026-06-18. Research checked 27 September 2026.