The landscape in 2026
There are now dozens of AI tools available, and the pace of change is dizzying. Rather than cataloguing every tool, this guide focuses on the ones most useful for students and covers how to think about using them well.
The big three chat models
| Tool | Best for | Free tier? | |------|----------|-----------| | ChatGPT (OpenAI) | General tasks, code, writing | Yes (GPT-4o limited) | | Claude (Anthropic) | Long documents, nuanced reasoning, safety | Yes (Sonnet limited) | | Gemini (Google) | Google Workspace integration, multimodal | Yes |
All three can: write and debug code, explain concepts, summarise documents, brainstorm ideas, and draft emails.
What AI is genuinely good at
- First drafts — get something on paper, then edit it yourself
- Explaining concepts — ask it to explain anything at multiple levels of depth
- Debugging code — paste your error, describe what you expected
- Summarising — long papers, articles, meeting notes
- Brainstorming — generating options you can then filter
What AI is NOT good at (yet)
- Accuracy on specific facts — it can confidently hallucinate. Always verify citations.
- Current events — training data has a cutoff date
- Complex maths without a code interpreter — use the code tool if available
- Knowing your context — you need to provide it
Academic integrity
UC has clear guidelines on AI use in assessments. The short version:
- Check your course outline — AI policies vary by course
- Disclose if required — many courses ask you to note if/how you used AI
- Understand what you submit — using AI to generate work you don't understand is academically risky and practically useless
Using AI to learn and to accelerate your work is generally fine. Using it to avoid engaging with the material is not — and it usually shows.
Getting started
- Create a free account on Claude or ChatGPT
- Try asking it to explain a concept from your current coursework
- Read our Prompt Engineering 101 guide to get better results faster