GPT makes learning fun again

I&#x27;m in two minds about it. On one hand the internet (in general) has become so hard to find information on, and I agree with the author that GPT is a breath of fresh air. On the other hand I&#x27;ve seen GPT fail so miserably at topics that I&#x27;m knowledge about that I have a very hard time trusting anything it tells me. I&#x27;m not sure what the answer really is, but I&#x27;m not sure it&#x27;s GPT. I wish we could go back to having search engines that actually led to useful information and not just advertisements, and I wish we had a GPT that would not tell lies.<p>I doubt either of those wishes are going to come true though. Search engines are likely always going to be SEO&#x27;ed into uselessness and GPT isn&#x27;t intentionally telling lies.

I really found asking GPT to put a math concept under a physics context very helpful for me. As a computational physics student, that is really how I understand math and the world. Only very few textbooks can help me with this.<p>From [<i>Make Something Wonderful: Steve Jobs in His Own Words</i>][1], Steve once said in a interview in 1983:<p>&gt; The problem was, you can&#x27;t ask Aristotle a question. And I think, as we look towards the next fifty to one hundred years, if we really can come up with these machines that can capture an underlying spirit, or an underlying set of principles, or an underlying way of looking at the world, then, when the next Aristotle comes around, maybe if he carries around one of these machines with him his whole life–his or her whole life–and types in all this stuff, then maybe someday, after this person&#x27;s dead and gone, we can ask this machine, “Hey, what would Aristotle have said? What about this?” And maybe we won&#x27;t get the right answer, but maybe we will. And that&#x27;s really exciting to me. And that&#x27;s one of the reasons I&#x27;m doing what I&#x27;m doing.<p>And this <i>future</i>, expected &quot;next fifty to one hundred years&quot;, is somewhat here already.<p>[1]: <a href="https:&#x2F;&#x2F;book.stevejobsarchive.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;book.stevejobsarchive.com&#x2F;</a>

I&#x27;ve been using GPT to have (insightful) educational conversations about Quake 1 source code: <a href="https:&#x2F;&#x2F;twitter.com&#x2F;krychusamp&#x2F;status&#x2F;1649048047996014595" rel="nofollow">https:&#x2F;&#x2F;twitter.com&#x2F;krychusamp&#x2F;status&#x2F;1649048047996014595</a><p>I always finish up by asking GPT to test my knowledge with a single-choice questionnaire. What I&#x27;ve observed is that the retention of the material is higher compared to &quot;traditional&quot; techniques. Perhaps the conversation style is more immersive, or perhaps focusing on specific knowledge gaps makes for accelerated &#x2F; personalised learning.<p>There is of course the problem of accuracy, but I feel like it&#x27;s often over-stated. Even if GPT is not correct at times, it often uncovers concepts and relations that paint a better overall picture for me, and lead me to better questions and follow up actions.

It’s nice to be able to ask ChatGPT a half baked, poorly researched, poorly worded question with bad grammar yet get a totally good faith response back that’s a springboard for follow up questions. Whereas if you did the same thing on any stack exchange site you get downvotes and comments like “please read the guidelines and edit your question.”

I noticed a while back that the internet has made me terribly prone to skimming. It got to a point where I could hardly stand to read anything longer than a short news article.<p>To learn anything useful on the internet, you pretty much have to skim. So much of the internet is so loaded with filler and BS that it is hardly worth reading at all.<p>With ChatGPT, it’s incredibly refreshing to be able to ask a question and get nothing other than a concise answer. No skimming required. I feel so much more focused and better able to learn this way.

You can downvote me but I think OP hasn&#x27;t learned how to learn yet. If OP reads Wikipedia by understanding every sentence and clicking on every link, that&#x27;s deliberately sabotaging his own learning. Attitudes like &quot;I don&#x27;t really feel like spelunking through a ton more articles&quot; simply shows OP isn&#x27;t interested in learning per se, just quick answers.<p>OP learns in a way that&#x27;s very child-like. When you are a five-year-old it&#x27;s okay to learn by asking everything. That stops being acceptable by the age of fifteen. OP hasn&#x27;t learned any research skills yet, and when OP&#x27;s needs inevitably exhausts the ability of LLMs, OP would be utterly unable to read an encyclopedia or a research paper or perhaps a textbook.

I&#x27;ve learned almost everything I know through reading and listening, with very little discourse. I rarely asked questions in class, never had tutors, never went to office hours. I hesitate to post questions online. If I don&#x27;t understand something, I just read ... more, or bang my head against it as trial and error.<p>I think this is partly why I&#x27;m still looking to be wowed by this technology, personally, in terms of what it can accomplish for me. And while it could be rightly said I&#x27;ve made things unnecessarily hard for myself approaching life like this, I feel it has been beneficial, and enriching, to force myself to really ask, what is this person saying here? In particular, I wouldn&#x27;t want GPT to lead to a general lessening of empathy.

Shortly before GPT and friends burst on the scene, I was looking for a website which would meet me where I am as an engineer - I&#x27;ve written reams of code in various languages.<p>If I want to try Rust, I don&#x27;t want to be taught uint8 v uint16 or that you shadow variables. I want to know the interesting parts.<p>ChatGPT is pretty good at this and the other thing I want: pandas training. You can ask it to generate exercises at any difficulty and also provide test data!<p>This tool is the biggest mind expander for me since search engines.

I agree, Google has deteriorated so much. It just points to either doc pages and leaves all the work up to you or you get a (often outdated from 2011) pages from Stack overflow. I need to do all the filtering and stitching.<p>I tend to get better answers from their &quot;automated questions&quot; which are paraphrased versions of my query. So it clearly understands what I&#x27;m after.<p>In order to promote diversity, i would recommend perplexity.ai which offers a similar experience as chatgpt (I&#x27;m not affiliated and i have no clue what their tech stack is like) It also offers links back to pages and follow up questions etc. Highly recommended if you need to learn something new and you don&#x27;t want to bang your head on the keyboard googling or ddg&#x27;ing<p>I&#x27;ll give an example. I recently needed to learn about k8s, minikube, kubectl et al for a project. I had some vague idea about the tech but nowhere near enough for what i needed to do. Google was useless because it kept taking me to doc pages which is like being lectured but i needed specific information. Perplexity was amazing in helping me with the right bit of information, example code AND links if i do want to read further

This is so ridiculous.<p>GPT is like that &quot;know-it-all&quot; friend we have who just has something to say about <i>anything</i>, with knowledge skimmed from the internet.<p>GPT is a language model. It outputs <i>what you want to hear</i>, not what is correct.

Don&#x27;t forget that GPT was trained on data from all the sources that failed you. So if we all collectively neglect them, and they fall over, we will lose many valuable resources.<p>I think we need to think about how to keep these valuable sites going, because they are ultimately providing most of the value of the various available language models.

Now whenever I need to use a fancy new package that I never used of, I use LangChain to collect all documents from the package document site, load them into a vectorDB and start asking GPT questions. This method works in 80% of the time. One pitfall is that with this method I only get what I want. I don&#x27;t get a deep understanding of that package as it used to be if I carefully read over the documents. Still finding a balancing point in between.

Imagine being able to ask questions and get answers back based upon understanding, and not upon a curriculum or agenda. I have found this method to allow me to consume information much faster while skipping over the often tedious topics.

So many comments on this shovelling sh*t on OP...? I agree with the author entirely. In practically every field, the reason it&#x27;s difficult to learn anything, starting with zero, is because you don&#x27;t know what questions to ask. You need to find an in. ChatGPT is really good at this - the answers it gives provide an idea of what you don&#x27;t know you don&#x27;t know. Obviously, you then need to go verify the answers, but the difference is: now you know what the language you need to use to do so looks like.<p>PS you would also need to do this if you started with Wikipedia as well.

I find this too. With GPT there is no such thing as a stupid question. That is really liberating! You can really double check the simplest and most basic of your assumptions with no fear of judgement. And as mentioned in other comments, the assurance of receiving a good faith answer is not to be underestimated.

It is very helpful to learn new stuff indeed, I am personally using <a href="https:&#x2F;&#x2F;www.phind.com">https:&#x2F;&#x2F;www.phind.com</a>.

Why learn? The best part about LLMs is you and your children will never need to know anything or think anything ever again.

Here&#x27;s the aspect of this article that I found the most fascinating, and might explain why I don&#x27;t get the same level of benefit from ChatGPT as many report.<p>If I were learning how LEDs work, I would not have wasted any time whatsoever on the search results that the author spent a lot of time on. They were obviously (to me) the wrong articles on the face of it, because they were covering aspects that weren&#x27;t really what I was looking for (the wrong sort of detail and emphasis).<p>So I think I would have been off and running pretty much immediately with the web search results rather than spending time on the clear dead ends.<p>ChatGPT gets me there too, after enough back-and-forth, but it takes longer for me to zero in on what I&#x27;m looking for.<p>I say this not to say that ChatGPT is in any way bad for this. I&#x27;m just noticing a difference in how the two of us engage in learning new topics. Perhaps the reality is that for some people, ChatGPT is a godsend, and for others, it&#x27;s fine... but hardly an improvement for this use use case.<p>It would explain a lot.<p>(Also, when did learning stop being fun??)

Most people seem to prefer learning by talking and asking questions. At least, that&#x27;s what I&#x27;ve gathered from Discord servers where 99.9% of people ask questions answered on the first page of the Readme ;)

ChatGPT can be a great supplement for independent research. But when the article mentions a “curious seventh-grader”, I think we should focus on getting them quality human instruction whenever possible instead of just pointing them to ChatGPT.<p>ChatGPT addresses a scalability problem: not everyone has access to a tutor or can just call up a teacher or mentor to learn and ask questions. But some in the tech industry claim that ChatGPT is as good as or even better than human instruction, which to me seems totally off base.<p>The biggest problem I see in using LLMs as a teacher-substitute is that LLMs answer the questions you ask, whereas a good teacher tells you what you need to hear. Maybe this is solvable with specialized model tuning, but we need to actually solve it before telling kids that the best way to learn is to talk to the computer.

I totally agree with the author, I find I&#x27;m finally able to ask about a range of new concepts, get them explained at my level, and dive deeper as curiosity warrants it.<p>It&#x27;s also magical when you summarise the understanding you&#x27;ve reached back to it, and it can confirm or tweak it for you.<p>In other ways it&#x27;s also nice to just pay for it and then to be in an advertising free space.<p>One critique is that when you ask it to compare things it&#x27;s often too balanced or too positive&#x2F;enthusiastic (&quot;both are great for different reasons!&quot;) when what you want is a more sober analysis. But you can usually do some prompt management to adjust it back to a reasonable range.

This is incredibly, incredibly naive. &quot;Within minutes, I learned&quot; -- absolutely nothing. At most you&#x27;ve learned how the answers to your question would sound like. <i>They are not the answers</i>.

Its almost as if ChatGPT is getting to be a form of dialogic learning [1]. While ChatGPT is not yet an authoritative source, it should be possible to get to such a future. The problem is with the ethics of it becoming an authoritative source and how it can possibly unlearn (or accept change or challenges). For now, the example OP provides is without authority where they learnt about a topic taking ChatGPT as a peer.<p>References:<p>1. <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Dialogic_learning" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Dialogic_learning</a>

Fun with caveats I&#x27;m finding. Today I asked for an outline of DevSecOps concept based on CISA recommendations. The CISA doc it referenced doesn&#x27;t actually exist.

Puh, fun, yes. It&#x27;s nice having a &quot;conversation&quot;. But then everything that comes up I have to double check. As soon as the topic is somewhat specific, I can tell just by looking at it that the code won&#x27;t work. So at the end of the day I have to go back to Google anyways.

It&#x27;s fun but not necessarily accurate.<p>Scary part is I don&#x27;t know enough about the subject to tell apart truth from falshood because they are stated in exactly the same confident manner. Also most things are true with falsehoods sprinkled in between

I&#x27;ve been using Phind [0] for specifically dev related stuff, and it&#x27;s been great. I very much like that it gives me all the resources it uses.<p>On the other hand, I feel like my experience with Phind has been so great because with all dev stuff I ask it it&#x27;s always been something I could try out right away.<p>With non-dev stuff that isn&#x27;t always the case. I&#x27;d love an LLM that can do the same thing as Phind, as in providing the sources it&#x27;s using, while also working for everything.<p>[0]: <a href="https:&#x2F;&#x2F;www.phind.com&#x2F;">https:&#x2F;&#x2F;www.phind.com&#x2F;</a>

One solution to his problem with the Wikipedia being hard to learn from is to use the criminally under-known Simple Wikipedia (<a href="https:&#x2F;&#x2F;simple.wikipedia.org&#x2F;wiki&#x2F;Main_Page" rel="nofollow">https:&#x2F;&#x2F;simple.wikipedia.org&#x2F;wiki&#x2F;Main_Page</a>). Just replace &#x27;en&#x27; in a wikipedia URL with &#x27;simple&#x27; and hopefully there will be a nice and simple version of the page. I don&#x27;t see it advertised very often but I think it should be.

I&#x27;ve found ChatGPT quite useful for both coding and learning a language. Even though I have two friends who are native speakers of the language I&#x27;m learning, I don&#x27;t want to bother them with my questions all the time. Especially since ChatGPT will give great explanations about grammar and why something is said the way it is. Also makes for a great practice partner to use your vocabulary and to correct your mistakes. It has definitely helped a lot already

This is the classic &quot;google is so crap now&quot;, then when you dig a bit, the search query is something incredibly broad (in this case &quot;led&quot;) or irrelevant, with the author then complaining about the quality of the results<p>I&#x27;m not taking anything away from Chat-GPT for this use case, but I see the above pattern in almost every Google bashing thread on HN.

Things I learned using gpt in last 6 mo:<p>* Various new english idioms. I read a lot, but still find new ones<p>* More optimal translations between languages i&#x27;m ~native in.<p>* ton of new JS, Python libs<p>* Got deeper in SQL (recursive queries, etc)<p>* Can use CSS way better than before<p>* Mermaid<p>I wrote <a href="https:&#x2F;&#x2F;chatcraft.org&#x2F;" rel="nofollow">https:&#x2F;&#x2F;chatcraft.org&#x2F;</a> as a friendlier UI for learning programming related things.

Obviously the risk here <i>today</i> is that ChatGPT isn&#x27;t a trustworthy teacher, but the concept is of course absolutely riveting and obvious to everyone, which is why ChatGPT blew up on the first place. Iron out the kinks and I think it&#x27;s going to be an incredible asset to humanity (not necessarily this particular LLM).

I was initially using gpt to learn for 5-8 hours every day..<p>But I was bitten by a series of completely made up facts, and my skepticism has forced me to research nearly every complex response since then to confirm.<p>This fatigue has really shut down my enthusiasm. Maybe I&#x27;ll research a topic 3-4 hours a week now.

Hot take: AI knowledge is what Bitcoin is to money. It is not backed by anything and therefore cannot be trusted. Regular knowledge is backed by a degree which has actual value in the end.

He has got a surface level understanding - GPT doesn&#x27;t disambiguate crystal momentum from momentum for example.

When was learning ever not fun?

im reading the wealth of nations right now, using gpt on the side to help clarify long passages and decode archaic terms. it&#x27;s been tremendously helpful. some actual use cases:<p>- summarising entire chapters (summarise chapter 8 of book 1)<p>- asking to verify my own summaries (in summary, smith writes that ... am i correct?)<p>- explaining certain passages in ever more detail (explain ...; explain in simpler terms; im still confused, explain in more details;)<p>---<p>One <i>very</i> interesting case is when GPT produced an error, and I had it go through the error with me. In the end GPT managed to spot its own error and correct itself. Here&#x27;s an excerpt, edited for brevity:<p>* gpt: [...] When the demand is too low, it encourages growth; when the demand is too high, it restrains growth. [...]<p>* me: &quot;When the demand is too low, it encourages growth; when the demand is too high, it restrains growth.&quot; is this correct?<p>* g: Apologies for the confusion in my previous response. The correct statement should be:<p>When the demand for labor is too low (slow), it encourages higher wages and population growth; when the demand for labor is too high (fast), it restrains population growth by adjusting wages to find the equilibrium between demand and supply.<p>(GPT still giving the wrong answer.)<p>* m: i am still confused. can you explain in more detail<p>* g: [...] Smith argues that the demand for labor regulates population growth by influencing wages. If wages are too low, fewer children will be born, and the population growth will be slow. If wages are too high, more children will be born, and the population growth will be fast. [...]<p>(At this point GPT was able to give me the correct response.)<p>* m: so when the demand for labour is low, wages decrease and population growth slows down. when the demand is high, wages increase and population growth speeds up. correct?<p>* g: Yes, that&#x27;s correct. [...]<p>(I pointed out that this was in contradiction with the earlier response.)<p>* m: did you not contradict yourself? &quot;When the demand is too low, it encourages growth; when the demand is too high, it restrains growth.&quot;<p>* g: Apologies for the confusion in my previous response. I misunderstood your question. Let me clarify:<p>When the demand for labor is too low, it restrains growth; when the demand is too high, it encourages growth.<p>So, when the demand for labor is low, wages decrease, and population growth slows down. Conversely, when the demand for labor is high, wages increase, and population growth speeds up. I hope this clears up any confusion.<p>(I then inserted the original erroneous response as a prompt, and gpt was able to &quot;remember&quot; that it generated the response earlier.)<p>* m: &quot;When the demand is too low, it encourages growth; when the demand is too high, it restrains growth.&quot; what is wrong with this passage?<p>* g: I apologize for the confusion earlier. There is an error in the passage you provided. It should be:<p>This demand for labor operates like the demand for any other commodity, regulating the production and growth of the population. When the demand is too low, it restrains growth; when the demand is too high, it encourages growth. This dynamic determines the rate of population growth in different countries, such as rapid growth in North America, slow and gradual growth in Europe, and a stationary population in China.<p>---<p>i should mention that i am using the paid version (GPT-4). i tried the free model, but the quality was significantly worse. you can recorrect it multiple times consecutively and every time it would say that it was wrong the last time, and recorrect itself again and again.

I&#x27;m confused. So, this person just asked ChatGPT bunch of surface level questions about Solid State Physics, got response. What is so exciting ?