Researchers at Stanford University are exploring practical and ethical ways to use increasingly capable artificial intelligence across education, public policy and scientific research, emphasising that such systems are tools to extend human ability rather than replace human judgement.
From tinkering to teaching
At Stanford’s AI Tinkery, students are given a hands‑on, informal space to experiment with robotics, machine learning and other emerging technologies while being asked to consider privacy, security, copyright and the cognitive effects of using AI. The aim is to build what one researcher calls AI literacy — a balanced understanding of what these systems can do and the trade‑offs that follow.
“We’re not necessarily pro‑AI or for everyone to use AI,”
The quotation above is from a Tinkery lead who said the intention is to let people see what the tools can do so they can make an informed choice about using them. That approach is practical: students in the Tinkery choose between an "AI boost" or a "human boost" at different stages of a project. For example, a student might ask an AI to evaluate an idea, or they might do direct user interviews to gather insight for product design. The distinction encourages reflection on when automation helps and when human contact remains essential.
AI as a partner in science and policy
Beyond education, Stanford researchers are examining how AI augments scientific discovery. Work in an AI for Science laboratory focuses on advancing core AI methods and applying them to problems across domains such as biology and healthcare. In biomedicine, researchers see opportunities for AI to support studies of ageing mechanisms, accelerate molecule design and provide clinicians with more informed decision support.
Crucially, the researchers frame these contributions as enlarging researchers’ capabilities rather than supplanting the scientists who direct investigations. That framing places human oversight, domain expertise and ethical judgement at the centre of any deployment.
Implications for South African science and education
Although the reporting comes from a US university, the themes are directly relevant to South Africa: universities and schools here will face similar choices about how to integrate state‑of‑the‑art models into curricula, research programmes and public‑sector planning. Key questions for policy makers and institutional leaders include:
- How to teach practical AI skills together with critical understanding of limits and harms;
- When to rely on automated evaluations and when to prioritise direct human engagement (for example, with community stakeholders in policy design);
- How to maintain research rigour and accountability when AI systems contribute to hypothesis generation or data analysis.
Adapting to these challenges will require investment in both technical infrastructure and ethics education. It will also call for clear governance frameworks so that AI is used transparently in research and public policy settings.
What to teach about AI
Based on the Stanford approach, an effective AI curriculum mixes practical tinkering with structured reflection. Elements to include are:
- Hands‑on projects where students test AI tools for specific tasks;
- Modules on data privacy, algorithmic bias and intellectual property;
- Exercises that force choices between automated and human methods to surface trade‑offs.
| Domain | Potential AI role | Human oversight needed |
|---|---|---|
| Education | Personalised tutoring, idea generation | Curriculum design, ethical guidance |
| Scientific research | Hypothesis formation, molecule design | Experimental design, interpretation |
| Public policy | Data analysis, scenario modelling | Value judgements, stakeholder engagement |
These categories are illustrative of the discussions reported by the Stanford researchers and should not be read as an exhaustive typology.
Conclusion
The central message from the Stanford work is modest but important: advanced AI can broaden what people and institutions are able to do, but it does not remove the need for human responsibility. As South African universities, government departments and research groups consider adopting such tools, they will need to pair technical capability with instruction in ethics and governance so that benefits are realised while risks are managed.
