YouTube2Text

wk06 3 iat111 — Transcript

by ivizlab · 2,140 words · 303 segments · language en · Watch on YouTube

Full transcript

  1. 0:01Steve DePaulo, IIT 111, week six. We're
  2. 0:04going to look at vibe coding and um
  3. 0:08how you could be using it, but also what
  4. 0:10it's doing in the world today. Um so
  5. 0:13that's describing what you want and the
  6. 0:16AI writes the code and then you refine
  7. 0:18it. You could see any of these examples
  8. 0:21or something that I'll show you today
  9. 0:24rapidly. Now, there are people in the
  10. 0:26class who are probably doing very
  11. 0:27significant vibe coding already. Um but
  12. 0:31this is an intro class, so I want to
  13. 0:32make sure everyone gets this at the
  14. 0:35right level. I'll be building this up
  15. 0:37over the course of the class showing you
  16. 0:40other examples as well.
  17. 0:42Um so vibe coding is building software
  18. 0:44by describing what you want in plain
  19. 0:46language and letting AI write the code.
  20. 0:49You get that result, you run it,
  21. 0:51sometimes it has errors, you got to put
  22. 0:52those errors back in, it helps you to
  23. 0:54fix them. Now, you're up and running, so
  24. 0:57you could ask for changes or new
  25. 0:58features. Uh without reading the code or
  26. 1:02you could learn about coding by looking
  27. 1:04at the code. This is the same as
  28. 1:06everything else we've talked about this
  29. 1:07week, which is the better details and
  30. 1:09approaches will give you the the best
  31. 1:12results possible.
  32. 1:15So you describe like I want to I want to
  33. 1:17build a solar system, I want it to
  34. 1:19rotate. You keep it simple at first, it
  35. 1:21writes the code, you look at it, you
  36. 1:23make sure it's working, and then think
  37. 1:25what else would I want, and you slowly
  38. 1:27build it up over time. So you don't need
  39. 1:29to know code, but you have to have a
  40. 1:31clear idea. And that gets to a bigger
  41. 1:34issue that I will spend more time on,
  42. 1:36which is oh, I'm not a programmer, but
  43. 1:39I'm a smart person, and if I could think
  44. 1:41of a good idea for my website or
  45. 1:43business or an app, I should be able to
  46. 1:46do it. And more and more I'm calling
  47. 1:49this the last mile, that a final that
  48. 1:52that it can handle. If you could
  49. 1:54describe something um where maybe you
  50. 1:57have the the last mile ability, that
  51. 2:01specialist step between your idea and
  52. 2:03the finished result, like writing a
  53. 2:05screenplay cuz you have a great movie
  54. 2:07idea or something to do with business or
  55. 2:10something that would normally cost a lot
  56. 2:12of money but there's a way you can write
  57. 2:14quickly in AI and in this case we're
  58. 2:16talking about writing code that it's
  59. 2:19here to help you do that. Now, I'm of
  60. 2:21the belief this last mile might be one
  61. 2:24of the big
  62. 2:25tools. You've seen me ethically talking
  63. 2:27about all the problems with AI and the
  64. 2:29dangers, but this could really help
  65. 2:31humanity. I call it bringing out the
  66. 2:33Einsteins, that's my
  67. 2:35quote, which is
  68. 2:37people all over the world who are
  69. 2:40impoverished or never have access to
  70. 2:42high school or college or tools or
  71. 2:46computers.
  72. 2:48Um
  73. 2:48Maybe they're just brilliant and they
  74. 2:50could have helped the world if we only
  75. 2:52gave them the ability to do that. I'm
  76. 2:55hoping with this last mile idea, it's
  77. 2:58like just explain your idea and we'll we
  78. 3:01can help you through it. So, am I
  79. 3:04dreaming here? Is this really a
  80. 3:05possibility?
  81. 3:07So, gifted people getting stopped by
  82. 3:09money or schooling or access have now
  83. 3:12access to this AI specialist. Wait a
  84. 3:14minute, Steve. AI costs a lot of money
  85. 3:16and blah blah blah. Well, in communities
  86. 3:20and My Lab and at SFU, we're trying to
  87. 3:23write open source freely
  88. 3:25AI that we would like to give to the
  89. 3:28world with some of this notion in mind.
  90. 3:31Surely, what you're learning with the
  91. 3:34human in the loop and our fork system,
  92. 3:38where you're conducting your ideas, all
  93. 3:40this in an open source free tool would
  94. 3:44move to this space. So, let's keep an
  95. 3:46eye out in that. You're doing the last
  96. 3:48mile with vibe coding. So, let's go back
  97. 3:51to vibe coding, where you're describing
  98. 3:53the AI is the last mile of writing your
  99. 3:56ideas. And then you're with this
  100. 3:5910,000-ft view view, because you don't
  101. 4:01have to think about every line of code,
  102. 4:03like we used to as programmers, you can
  103. 4:05kind of make it better. So, this is that
  104. 4:07notion again of orchestrating or
  105. 4:09conducting your human idea. So, in that
  106. 4:12case, let's think, oh, I want to build a
  107. 4:143D solar system, create an interactive
  108. 4:163D solar system, keep it simple, I want
  109. 4:19to see a pause and a zoom and some
  110. 4:21features. So, I put that in, you could
  111. 4:24just put that in a browser, and you
  112. 4:26could see something like this. Like, I
  113. 4:28could change the orbital speed here. I
  114. 4:31Later on, I wanted to see the sun pulse.
  115. 4:35I can turn off the labels, I can turn
  116. 4:37them on. I have a cinematic tour, where
  117. 4:40I decided I can go through the planets
  118. 4:42in this really cool way. So, you can do
  119. 4:44this, and you can just keep building it
  120. 4:46the way you want to.
  121. 4:47So, you it generates the code, it you
  122. 4:51run the finished demo, and you just keep
  123. 4:53building on it. So, that's the way you
  124. 4:55want to do it. Keep it simple. You could
  125. 4:57do this right now in the chat window of
  126. 5:00of any LLM, like Claude or ChatGPT
  127. 5:04or Gemini.
  128. 5:06Um And this is the way you can get a
  129. 5:08quick prototype. Honestly, it's starting
  130. 5:10to make code and tools, and that's could
  131. 5:13build up to a whole folder of things.
  132. 5:16So, what we really move into when you
  133. 5:19could still do it that way, but you
  134. 5:21could also move into what the real Vibe
  135. 5:24coding, which you use some kind of
  136. 5:26system like Claude Code and others. And
  137. 5:29what you do in that case, it's able to
  138. 5:31read files on your computer. So, if
  139. 5:33you're getting to an editor like VS
  140. 5:35Code, which is the one I use, and then
  141. 5:37type in Claude Code, I would first go to
  142. 5:40the folder of my big project, and sit
  143. 5:45there and work through exactly what I
  144. 5:46want to do. In this way, the uh I can be
  145. 5:49a lot smarter because it's looking
  146. 5:51through the code, you can tell it how to
  147. 5:53change it. Similar to everything else
  148. 5:55I've been talking to you about in this
  149. 5:57lecture and others, um
  150. 5:59you would you know, when you get
  151. 6:01somewhere good, you summarize, you ask
  152. 6:03it to summarize every big change, things
  153. 6:05you might want in the future to do it in
  154. 6:07steps. Um and that's what us new quote
  155. 6:11programmers are doing. Here's a science
  156. 6:14one. So, you can just put this prompt um
  157. 6:17in and see if it works for you. I just
  158. 6:20did that.
  159. 6:24And I'll bring that in.
  160. 6:31So, I'll start the simulation and you
  161. 6:32can see I have it moving. I can change
  162. 6:35the infection rate and other things
  163. 6:36about it. Um
  164. 6:39I'm on day 13 here. Sorry, it stops
  165. 6:42every once in a while for me. I can't
  166. 6:43forgot why.
  167. 6:45There you go again.
  168. 6:46Um these numbers will change on the fly
  169. 6:49based on what they were at. I can speed
  170. 6:52it up by five times. I'm going to reset
  171. 6:54it. Let's start the simulation again.
  172. 6:57There we go. It's working better now.
  173. 6:59I'll go to five days. Um I'll train I'll
  174. 7:02up the transmission rate. I'll think
  175. 7:04about recovery time
  176. 7:06uh and other kinds of things and I will
  177. 7:09I will as the days go by
  178. 7:12um see how that works. You can tweak
  179. 7:14this in any way you want. I have another
  180. 7:16one here more in the business space
  181. 7:18about starting a startup. Uh and and
  182. 7:21just really wanting to understand what
  183. 7:24do I charge? How does this work? So, my
  184. 7:26monthly price is $12 a month. Should I
  185. 7:28charge a different rate? So, I'll bring
  186. 7:31that in the window here.
  187. 7:35You see I've done this in Gemini.
  188. 7:37Um
  189. 7:39so, um Um
  190. 7:41I I've started at $12 an hour. Well,
  191. 7:44I've now moved it up to $25 an hour. Oh,
  192. 7:47and then I What is my acquisition cost?
  193. 7:50And you can see these curves. You can
  194. 7:51look over time.
  195. 7:53Uh you can see what your cash might be.
  196. 7:55So, you would set all these things that
  197. 7:57you think make sense. Or maybe you
  198. 7:59already have data from your lab. Um and
  199. 8:02you just continue using this. So, here's
  200. 8:04some other example. You can try that.
  201. 8:06Just put that in. Try that yourself.
  202. 8:08Maybe tweak it with with your data or
  203. 8:12you're thinking of a new company. Uh
  204. 8:15another one might be in the sciences of
  205. 8:17predator prey simulation or putting in
  206. 8:20your lab data or doing marketing. So,
  207. 8:22this is live coding. Uh you have your
  208. 8:25ideas. AI codes it. Um
  209. 8:28you If you're doing this with bigger
  210. 8:30projects, you make a project folder. You
  211. 8:33go to that folder and open it up so it
  212. 8:35can see it and understand the code. Um
  213. 8:38and this is the deeper way that many
  214. 8:39many people are doing things now.
  215. 8:42Um you could be making your own website,
  216. 8:44portfolio site. Um
  217. 8:47with with the final projects, a lot of
  218. 8:49people took their their knowledge and
  219. 8:52put that in. So, you start small. You
  220. 8:54describe. You try. You refine. You You
  221. 8:57get, you know, if you're an expert, it's
  222. 8:59going to work better. So, it's better to
  223. 9:00do it in your space. You test it. You
  224. 9:02own it. It is yours. So, this is
  225. 9:04slightly changing the world a bit
  226. 9:07because now we do not have to put up
  227. 9:09with um
  228. 9:11using somebody else's tool. You could
  229. 9:14write the tool. If you're smart and you
  230. 9:15want to write your own tool, that makes
  231. 9:17sense. So, what does this mean for
  232. 9:19companies and programmers?
  233. 9:21Um
  234. 9:22So, you can get your exact idea out.
  235. 9:25Uh it's changing how quick small
  236. 9:28companies could put things together.
  237. 9:30Um There's even a situation where
  238. 9:34it's not obvious that you have to
  239. 9:36um
  240. 9:37deal with companies that maybe you're
  241. 9:38not interested in. Adobe has been
  242. 9:40charging wild rates and it's getting
  243. 9:43quite expensive. In fact, at SFU, we
  244. 9:45couldn't make a deal with them and we're
  245. 9:47we're leaving.
  246. 9:49That um and it turns out a company uh a
  247. 9:52group of people recently uh using a
  248. 9:55program called Rust has um
  249. 9:59understood everything about what
  250. 10:01Photoshop or and Adobe other tools are
  251. 10:04and they have a replacement now called
  252. 10:06Photo Craft. I've used it. It it is
  253. 10:08literally everything that's in
  254. 10:09Photoshop. So, this is it's got all
  255. 10:12these views and people using it. Um
  256. 10:15So, there's a this is a negative as
  257. 10:18well. Is this going to take down every
  258. 10:20company who makes a tool because it's
  259. 10:23it's easy to just copy it and put it on
  260. 10:26the web for free. So, we'll see what
  261. 10:28this ends up being.
  262. 10:30So, if anyone could build their own
  263. 10:32tools, why are compa- why are companies
  264. 10:34really selling? So, there's a lot of
  265. 10:36ethical issues that you could think
  266. 10:38about. Uh however, we're back to if you
  267. 10:41have an idea for something, you just
  268. 10:43don't need that level of money you would
  269. 10:46normally need. You could be just trying
  270. 10:48it at home and building it up. So, think
  271. 10:51about building your own tools like uh uh
  272. 10:53in all these possible spaces. I think I
  273. 10:56showed you um last week some of the
  274. 10:59final projects where people were at
  275. 11:01least doing some of these these these
  276. 11:04interactive dashboards within their
  277. 11:06space. Um and more.
  278. 11:10Um so, the takeaways here is you
  279. 11:12describe, you build, you refine. Uh the
  280. 11:14more of an expert in the space you are,
  281. 11:16the more it's your idea. Uh you could do
  282. 11:19the simple chat to an LM as a prototype,
  283. 11:22but you could do full live coding where
  284. 11:24it looks at the folder of your stuff and
  285. 11:26you build it up and then you put it on
  286. 11:28GitHub so others can use it and you can
  287. 11:30community. This is a lot of what my lab
  288. 11:32does as well, where we give these tools
  289. 11:34away. We work with other community. So,
  290. 11:37it's kind of really changing the space
  291. 11:40uh because of the speed. Okay, that's a
  292. 11:42quick primer
  293. 11:44on vibe coding. Why don't you um
  294. 11:48try putting this one in or some of these
  295. 11:51others in and see how it works for you.
  296. 11:55Uh we'll talk more about it.
  297. 11:57Um I the the one thing I wanted to add
  298. 11:59some of these things you need a
  299. 12:01subscription for. So, I believe to do
  300. 12:03full cloud coding in a folder, you need
  301. 12:05you might need the uh $20 a month
  302. 12:08subscription, but there are free ones
  303. 12:10out there. Thank you.

About this transcript

This page contains the full transcript of wk06 3 iat111 by ivizlab, generated from the public captions YouTube serves with the video. The transcript has 2,140 words across 303 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.