learning strategies

Bilingual

Using AI to Strengthen Learning, Not Replace It
2 months ago
- Blocking generative AI in classrooms is ineffective; the key question is how students use it, not whether they will.
- Every AI tool removes some effort; the danger is offloading intellectual work that develops knowledge before it is mastered.
- Students using ChatGPT to complete tasks may show short-term performance gains but suffer long-term learning penalties when access is removed.
- Effective AI use involves a sequential workflow: attempt the task independently, use AI for targeted support, verify output, and reproduce knowledge without assistance.
- Video downloaders like Canvas Assistant can aid learning when used for active revision and retrieval practice, not passive consumption.
- The critical divide is between using AI to complete educational tasks versus using it to achieve educational goals through deliberate cognitive effort.
Entropic Thoughts
2 months ago
- Effective tutors prioritize motivation management, adapting lesson difficulty based on student engagement.
- Tutors select exercises that challenge students at the edge of their abilities and reveal flawed mental models.
- Students are encouraged to verify their own solutions using methods like alternative approaches, feasibility checks, and recursive verification.
- Productive errors (revealing misunderstandings) are discussed, while unproductive errors (simple slips) are often ignored.
- The skill of forming and testing hypotheses about systems is transferable to learning other domains like debugging or driving.
The Value of Domain Knowledge in Software
2 months ago
- Domain knowledge, understanding the industry and end-users, is often overlooked but highly valuable for software developers.
- It increases autonomy by reducing the need for feedback from others when making user-facing decisions.
- Domain knowledge helps avoid back-and-forth communication and is useful for scoping new projects or features.
- Learning about a new domain is enjoyable and broadens perspective.
- Knowledge for building software can be split into four categories: foundational (computers), stack-specific, product-specific, and domain knowledge.
- Foundational and domain knowledge require explicit learning time, while stack and product knowledge come from daily work.
- Ways to acquire domain knowledge include talking to users, reading industry materials, and taking time to step back.
Search Less, Browse More
3 months ago
- Emphasizes browsing documentation over solely searching for quick answers to discover more features.
- Uses learning Excel as an example, where browsing uncovered powerful features unknown to long-time users.
- Contrasts searching (focused on problem-solving) with browsing (focused on discovery and learning).
- Illustrates with Python's pathlib module, found through browsing, which is a superior solution for file paths.
- Highlights that browsing can reveal broadly useful tools that searching might miss due to specificity.
- Proposes a potential workshop on Excel for programmers or analysts based on interest in advanced features.

learning strategies

Using AI to Strengthen Learning, Not Replace It

Entropic Thoughts

The Value of Domain Knowledge in Software

Search Less, Browse More