Hi all, This thread continues to highlight such important insights. I wanted to add another perspective, building on what’s been said.
A key distinction I often emphasize in my own teaching is the relationship between effort and comprehension. It’s true that AI tools can make the process of problem-solving more efficient, but it’s also critical to ensure that learners are actively engaging with the concepts, rather than passively accepting solutions. One possible approach is introducing structured exercises that leverage AI as a learning companion. For instance: Students can use AI tools to draft initial code. They then analyze and critique the AI's output—identifying potential errors, inefficiencies, or areas for improvement. Finally, they optimize the code manually, applying their own knowledge to refine and debug. This strategy shifts the focus from simply generating output to understanding the "why" and "how" behind it. It also aligns with the growing need to teach "meta-skills" for interacting with AI: evaluating its reliability, fine-tuning its outputs, and balancing automation with human creativity. Hao's analogy with meal kits is particularly apt here—AI can provide the "ingredients" or the base recipe, but students still need to understand how to "cook," adapting the process to suit their specific needs or goals. This fosters a deeper understanding of the coding principles and encourages critical thinking. I’m curious if others have tried similar methods or have insights on striking the right balance between leveraging AI tools and preserving hands-on learning opportunities. Best, Federica On Tue, Mar 18, 2025 at 4:58 PM Hao Ye <[email protected]> wrote: > If we only care about the end result (e.g. a function that computes >> something) and not the means to the end (e.g. how you compute something), >> it may be fine to teach an AI heavy methodology. We, as teachers, first >> have to decide what we want a student to learn before deciding the means to >> get them to that objective. We still teach mathematics to students even >> though we've been able to answer most all math questions through the >> college level with Wolfram Alpha for 15 years now and many things can be >> done on a calculator for 50 years now. > > > I am strongly in agreement with the distinction here between learning > programming vs. learning *some* set of skills whose end product is a > correct program. > > And because I like my cooking analogies, I'll add in an additional > metaphor of learning to cook vs. following the instructions in a meal kit. > > I am also reminded about the pedagogical literature findings that people > are bad at evaluating how much they've learned; and one of the common > estimators people use (that is biased) is how *easy* the learning was > (e.g. passively listening to a lecture vs. active learning). > > Best, > -- > Hao Ye > (he/him/his) > [email protected] > > > On Tue, Mar 18, 2025 at 7:31 AM Jason Moore <[email protected]> wrote: > >> I think you have to carefully evaluate what the learning objectives are >> before deciding on AI or not or how much AI, etc. Take the analogy of >> learning an ancient way of making bowls by hand with wood versus learning >> how operate a manufacturing plant's bowl making machine. For the latter, >> you may not even need to know what a carving knife is or how the blade >> interacts with wood. In computer science, we have similar analogies tied to >> learning to code in assembly versus python. I can engineer a solution with >> either language and get something that does the same thing but I need >> different knowledge and understanding to succeed with each language. Of >> course we can teach students "how to code" using AI but they learn >> different things and the focus may mostly be on the product. If we only >> care about the end result (e.g. a function that computes something) and not >> the means to the end (e.g. how you compute something), it may be fine to >> teach an AI heavy methodology. We, as teachers, first have to decide what >> we want a student to learn before deciding the means to get them to that >> objective. We still teach mathematics to students even though we've been >> able to answer most all math questions through the college level with >> Wolfram Alpha for 15 years now and many things can be done on a calculator >> for 50 years now. If we want a student to know how every line in a program >> works, then they probably need to think about and write every line of that >> code. I listened to a recent podcast, the name is escaping me, where the >> caster lamented on students using AI to get the product, a written essay, >> but not learning how to write. Do we really want to stop teaching people >> how to write because a computer can write things? >> >> Jason >> moorepants.info >> +01 530-601-9791 >> >> >> On Tue, Mar 18, 2025 at 11:25 AM Paola Di Maio <[email protected]> >> wrote: >> >>> Simon, and Toby whose sentiment is echoed- >>> >>> thanks for the discussion points >>> >>> In my personal experience, it is important to use the new tools >>> responsibly and intelligently. >>> if they can help us learn faster and better. Because there is too much >>> to learn, esp for younger generations >>> There is a race going on, and machines can give us some edge >>> >>> I feel that my mission is how to teach how to use AI responsibly* and >>> critically* >>> >>> which in fact is true of any technology, starting from evaluating >>> critically search results of any query >>> >>> So rather than saying do not use AI code generators, I d say >>> >>> a) continue to learn how to code, but use that knowledge >>> to write good prompts for the code generators and >>> b) learn how to evaluate debug, proof, test, implement, integrate, test. >>> evaluate, optimise code >>> *shameless self promotion: I teach a course on responsible use of AI if >>> anyone is interested/able to collaborate with me on offering it >>> >>> By all means we must continue to learn and teach how to code - bur we do >>> not have to do the work ourselves >>> we become supervisors, system integrators etc >>> >>> How to use code generators to produce the code we want/need and how to >>> evaluate the qualify of ai generated code? >>> how to evaluate the output of any technology mediated process,really, is >>> what I think we help learners with >>> >>> I attach three relevant short posts in PDF from my feed >>> *disclaimer A*I supported, let me know if you spot terrible wrongs* >>> >>> for general interest/discussion and it would be great to connect over >>> our shared experiences >>> of using AI tor teaching/learning >>> >>> PDM >>> >>> On Mon, Mar 17, 2025 at 5:26 PM Waldman, Simon via discuss < >>> [email protected]> wrote: >>> >>>> I think the reasoning behind the “don’t use AI when you’re learning” >>>> comment is that there is a risk that people simply use code that they are >>>> given without thinking about it, and therefore don’t build a mental model >>>> of what is going on, don’t learn, and hence are unable to spot mistakes by >>>> the AI or build more advanced things. >>>> >>>> >>>> >>>> In a sense, I think it’s best to think of AI coding assistance as a >>>> more advanced version of “look it up on StackOverflow”. We have all come >>>> across plenty of people who have done all their coding by copy-pasting >>>> snippets from StackOverflow without fully understanding them – as well as >>>> people who have laid the groundwork by learning what they are doing first, >>>> and **then** looked things up to much better effect. I think this is >>>> especially important for a first language – when one is learning >>>> programming as well as a particular syntax. >>>> >>>> >>>> >>>> That’s my 2p; clearly “don’t use AI” is not enforceable and won’t be >>>> adhered to by many, but it’s important that we explain the reason that that >>>> advice is given (and maybe consider moderating it to “limit use of AI” or >>>> similar) so that hopefully most learners will bear it in mind. >>>> >>>> >>>> >>>> -- >>>> >>>> Dr Simon Waldman / [email protected] >>>> >>>> Assistant Professor of Energy Technologies, Heriot-Watt University >>>> >>>> Programme lead for MSc Renewable & Sustainable Energy Transition >>>> >>>> >>>> >>>> *From:* Paola Di Maio <[email protected]> >>>> *Sent:* 16 March 2025 22:46 >>>> *To:* discuss <[email protected]> >>>> *Subject:* Re: [cp-discuss] Re: Feedback Request: Lesson Updates on >>>> Generative AI >>>> >>>> >>>> >>>> >>>> ****************************************************************** >>>> *Caution: This email originated from a sender outside Heriot-Watt >>>> University. Do not follow links or open attachments if you doubt the >>>> authenticity of the sender or the content. * >>>> * ***************************************************************** >>>> >>>> >>>> >>>> Thank you Toby and all for the discussion >>>> >>>> it is an important one, at many levels >>>> >>>> >>>> >>>> I have always detested coding, comparing it to unnecessary bricklaying >>>> vs me being interested in information architectures >>>> >>>> >>>> >>>> When AI generated code became available I felt relief, the finally >>>> humanity has found a way of avoiding coding by hand >>>> >>>> >>>> >>>> My question is: is the AI generated code as good as, better or worse >>>> than humanly written code? >>>> >>>> Having the code written up already means learners must learn how to >>>> implement it and run it correctly >>>> >>>> >>>> >>>> can it be used to learn/teach about coding more productively - ie >>>> engaging learners to Implement debug, test, maniupate, evaluate the ai >>>> generated output >>>> >>>> and how to correct it and improve it, rather than putting their effort >>>> into writing >>>> >>>> >>>> >>>> AI generated code could allow learners to move straight into the next >>>> level of coding, that is implementation >>>> >>>> Thorough understanding of how the syntax and logic of the program >>>> should still be required, but the human intelligence >>>> >>>> so rare and precious can be spared the tedious task of actually writing >>>> it >>>> >>>> >>>> >>>> >>>> >>>> On Mon, Mar 17, 2025 at 6:21 AM Paul Harrison via discuss < >>>> [email protected]> wrote: >>>> >>>> Hi Toby, >>>> >>>> I'm following this ongoing discussion with interest. Great to see this >>>> being added to Carpentries material. >>>> >>>> We recommend that you avoid getting help from generative AI while you >>>> learn to code >>>> >>>> >>>> >>>> I was a bit surprised by this negative conclusion. My feeling would be >>>> that it isn't reasonable to expect people not to use these tools while >>>> learning, and therefore they need to know how to use them safely. And they >>>> do seem quite good at explaining code or suggesting different approaches. >>>> >>>> >>>> >>>> Here's a slide I used in a recent workshop, although I'm far from 100% >>>> happy with it. >>>> >>>> >>>> https://monashdatafluency.github.io/r-progtidy/slides/introduction.html#11 >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> ------------------------------ >>>> >>>> Founded in 1821, Heriot-Watt is a leader in ideas and solutions. With >>>> campuses and students across the entire globe we span the world, delivering >>>> innovation and educational excellence in business, engineering, design and >>>> the physical, social and life sciences. This email is generated from the >>>> Heriot-Watt University Group, which includes: >>>> >>>> 1. Heriot-Watt University, a Scottish charity registered under >>>> number SC000278 >>>> 2. Heriot- Watt Services Limited (Oriam), Scotland's national >>>> performance centre for sport. Heriot-Watt Services Limited is a private >>>> limited company registered is Scotland with registered number SC271030 >>>> and >>>> registered office at Research & Enterprise Services Heriot-Watt >>>> University, >>>> Riccarton, Edinburgh, EH14 4AS. >>>> >>>> The contents (including any attachments) are confidential. If you are >>>> not the intended recipient of this e-mail, any disclosure, copying, >>>> distribution or use of its contents is strictly prohibited, and you should >>>> please notify the sender immediately and then delete it (including any >>>> attachments) from your system. >>>> >>>> This list is for the purpose of general discussion about The > Carpentries including community activities, upcoming events, and > announcements. Some other lists you may also be interested in include > discuss-hpc, discuss-r, and our local groups. Visit > https://carpentries.topicbox.com/groups/ to learn more. All activity on > this and other Carpentries spaces should abide by The Carpentries Code of > Conduct found here: > https://docs.carpentries.org/topic_folders/policies/code-of-conduct.html > *The Carpentries <https://carpentries.topicbox.com/latest>* / discuss / > see discussions <https://carpentries.topicbox.com/groups/discuss> + > participants <https://carpentries.topicbox.com/groups/discuss/members> + > delivery options > <https://carpentries.topicbox.com/groups/discuss/subscription> Permalink > <https://carpentries.topicbox.com/groups/discuss/T032962510fa76492-Mf46ca6aad97428924c20f215> > ------------------------------------------ This list is for the purpose of general discussion about The Carpentries including community activities, upcoming events, and announcements. Some other lists you may also be interested in include discuss-hpc, discuss-r, and our local groups. Visit https://carpentries.topicbox.com/groups/ to learn more. All activity on this and other Carpentries spaces should abide by The Carpentries Code of Conduct found here: https://docs.carpentries.org/topic_folders/policies/code-of-conduct.html The Carpentries: discuss Permalink: https://carpentries.topicbox.com/groups/discuss/T032962510fa76492-M77736c28f511f8ac92a79185 Delivery options: https://carpentries.topicbox.com/groups/discuss/subscription
