{"id":17931,"date":"2025-12-02T08:15:12","date_gmt":"2025-12-02T16:15:12","guid":{"rendered":"https:\/\/www.scotthyoung.com\/blog\/?p=17931"},"modified":"2025-12-02T09:26:05","modified_gmt":"2025-12-02T17:26:05","slug":"5-strategies-to-learn-better-with-ai-and-traps-to-avoid","status":"publish","type":"post","link":"https:\/\/www.scotthyoung.com\/blog\/2025\/12\/02\/5-strategies-to-learn-better-with-ai-and-traps-to-avoid\/","title":{"rendered":"5 Strategies to Learn Better with AI (and Traps to Avoid)"},"content":{"rendered":"

Recently<\/a>, I shared my predictions about how AI will change the process of learning. In short, I expect that AI will reduce the need for many skills, potentially dumbing down the typical user.<\/p>\n

Just as the invention of writing made us worse at remembering things and the invention of the calculator made us worse at mental math, having a technology that can read emails, write code and do homework for us will invariably make most of us less proficient at doing those things ourselves.<\/p>\n

But I think the average case here will mask a lot of variation. Many people will learn and think less, but some people will learn much more.<\/p>\n

I\u2019ve been experimenting with AI quite a bit in my own research and learning, and the results have been all over the place. For some aspects of learning, AI does a fantastic job, saving me hours of needless effort. But in other cases the results are mediocre, and in some cases they\u2019re downright misleading.<\/p>\n

Here are some of my takeaways in using AI as a tool for accelerating learning:<\/p>\n

1. Choose which<\/em> books to read (but you still need to read them).<\/h2>\n

AI is a great tool for book recommendations. I used it extensively in my recent Foundations<\/a> project, often using ChatGPT and other tools to recommend books based on fairly specific criteria to round out my reading lists.<\/p>\n

\n
\"\"<\/figure>\n<\/div>\n

But while AI can give some great recommendations about what to read, AI summaries of those books aren\u2019t a good substitute for reading the books themselves. Some of this is a problem of verifiability (more on that in a moment), but it\u2019s a problem with summaries in general\u2014you don\u2019t learn ideas deeply by skimming them. It\u2019s only by reading a book in full that you can truly learn and understand the examples, knowledge base and authorial perspective that allows you to use that information to reason about other things.<\/p>\n

AI can help you spot if a book isn\u2019t worth your time so you can read from a tightly curated list that matches your interests, ability level and knowledge gaps.<\/p>\n

2. Source alternative suggestions (but do your own thinking).<\/h2>\n

A common refrain in the responses to my recent essays has been that AI seems<\/em> to be great at things you\u2019re not an expert in, but whenever you are highly competent at a domain, the AI advice seems to be fairly bad. Sort of an AI-version of Gell-Mann amnesia<\/a>.<\/p>\n

For instance, I find AI to be a lousy ghostwriter. Even for little things where I don\u2019t have ethical qualms about using AI, such as generating paragraph-level summaries of lessons I wrote for a course, I\u2019ve found the AI summaries to be inferior and have ended up just writing them myself.<\/p>\n

\u2028Similarly, I\u2019ve found AI to be unhelpful for giving me business advice, generating flashcards or designing curricula for topics I want to learn. Some of that may be a skill issue on my part, or a lack of good prompts, but the fact that I can generally get satisfactory results in other domains makes me think that the problem is simply that my standards are much higher here.<\/p>\n

\n
\"\"<\/figure>\n<\/div>\n

But even if AI isn\u2019t helpful with the things you know best, its breadth is incredible. Often what AI does best is expose you the breadth of knowledge outside your area of expertise, offering suggestions that you may not have heard before.<\/p>\n

When using AI to help me with problem solving, I have found it helpful to ask the AI to suggest alternative ways of approaching the problem that I might not have considered. This often opens solution paths I never would have explored on my own.<\/p>\n

\n
\"\"<\/figure>\n<\/div>\n

3. Ask for verifiable answers when accuracy matters (and fact-check critical information).<\/h2>\n

Everyone makes a big deal out of AI hallucinations. I agree they\u2019re a problem, but all information sources have factual inaccuracies, so the problem isn\u2019t limited to AI. When I was researching my last book<\/a>, I was shocked how often even peer-reviewed papers would cite something inaccurately.<\/p>\n

I think the actual problem with hallucinations isn\u2019t the error rate, but that the pattern of errors is very unhuman. Generally speaking, when you write something that looks like a well-researched essay, the chance of it having factual mistakes is lower than, say, an off-the-cuff remark on a podcast. AI breaks these conventions because it\u2019s just as likely to hallucinate when it is writing in a seemingly meticulous style as when it appears to be riffing.<\/p>\n

There is a simple fix to this problem: ask questions in a way that they can be verified. Examples:<\/p>\n