YouTube2Text

ChatGPT & Google Gemini’s New Features Are Crazy! (Remove Watermarks, History & More) — Transcript

by Rob The AI Guy · 3,078 words · 430 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Chat GPT and Google Gemini just released
  2. 0:01a bunch of brand new updates that change
  3. 0:03everything. Now on Gemini, you can
  4. 0:05actually choose to have watermarks
  5. 0:06turned off, which is pretty crazy. In
  6. 0:08addition to that, Chat GPT got a bunch
  7. 0:10of brand new upgrades that make it way
  8. 0:11better for automating things because it
  9. 0:13has way better memory now and can
  10. 0:15literally see and track everything
  11. 0:17you're doing on your computer. By the
  12. 0:18end of this video, you're going to know
  13. 0:19all these changes are, how to turn them
  14. 0:21on, how to implement them, and how to
  15. 0:23automate things with Gemini and Chat
  16. 0:24GPT. Change number one that I wanted to
  17. 0:26show you is now you could actually
  18. 0:28request to remove watermarks on music,
  19. 0:30videos, and images from directly inside
  20. 0:33of Google Gemini. For example, we could
  21. 0:34say something like create photo of a
  22. 0:37sloth. And now this will go through and
  23. 0:40this will actually create the image.
  24. 0:41What I'm going to show you is right now
  25. 0:43I did not touch that feature. So it's
  26. 0:44not going to remove the watermark. This
  27. 0:46is going to come with a watermark and I
  28. 0:48want to show you exactly what that looks
  29. 0:49like so that you understand exactly what
  30. 0:51this is doing right here. We can see
  31. 0:53that this actually created this image of
  32. 0:54this sloth right here. It actually looks
  33. 0:56incredible. If I come over here and I
  34. 0:59download this right here by clicking on
  35. 1:01download in full size, we could see that
  36. 1:03there's actually going to be a small
  37. 1:05watermark in the right hand side right
  38. 1:07here. In fact, it looks like it's not on
  39. 1:09here, but I asked Gemini and it says the
  40. 1:12watermark is located in the bottom right
  41. 1:13hand corner of the generated image and
  42. 1:15it appears as a small semi-transparent
  43. 1:18white logo. We can't see it because of
  44. 1:20the actual color of this right here, but
  45. 1:23if we come over here and we click on the
  46. 1:24setting right here, we're going to now
  47. 1:26see media watermark right here and we
  48. 1:27can actually turn this off. So now
  49. 1:29watermark will not be included when we
  50. 1:31create an image, when we create a video,
  51. 1:33or when we create music. And even if a
  52. 1:36human can't detect it, guess what can
  53. 1:38detect it? AI detectors. And that's why
  54. 1:40if you've been using Gemini for things
  55. 1:42like social media content or something
  56. 1:44like that, you might be seeing a
  57. 1:45decrease in reach because there's that
  58. 1:47invisible watermark on there that you
  59. 1:49haven't been removing. And then as a
  60. 1:50result, you're not getting the reach
  61. 1:52that you want. But that was just change
  62. 1:54number one. Change number two is now if
  63. 1:56we come over to Gemini Spark right here,
  64. 1:58this can actually browse the internet
  65. 2:00for us, which is pretty awesome. For
  66. 2:01example, I could say something like I
  67. 2:03want to buy a Ferrari SF90. Can you find
  68. 2:07me one for less than 400k? And then what
  69. 2:11this is actually going to do is go
  70. 2:12through and do this. We could see this
  71. 2:14is going to think through this. It's now
  72. 2:16actually doing a Google search and if it
  73. 2:18needed to, it could actually open up my
  74. 2:20browser and browse on my browser through
  75. 2:22the Chrome extension, which is pretty
  76. 2:24incredible. So, this actually goes
  77. 2:26through right here and we could see the
  78. 2:28current market overviews and I could
  79. 2:29find exactly where they are. So, Ferrari
  80. 2:32approval pre-owned is going to be around
  81. 2:34440. That's 100% accurate. If we come
  82. 2:36over here, we could see on CarGurus,
  83. 2:38there are some listed for 390. Cars.com
  84. 2:41has some in the low 400s. And this is
  85. 2:43the way it goes through and we can
  86. 2:44actually click into all these and we
  87. 2:45could see exactly what they are seeing
  88. 2:48right here. If we come over here, we do
  89. 2:50lowest price, we could see that there
  90. 2:51should be some that are less than 400
  91. 2:54unless they've sold recently. Okay, so
  92. 2:56these first few right here are all going
  93. 2:58to be actually ads right here, but we
  94. 3:01could see 416 is the lowest run right
  95. 3:04here. But, if we come back over here, we
  96. 3:06could come over here and click on
  97. 3:07Ferrari pre-approved and we could see
  98. 3:10that these are going to be a lot more
  99. 3:12expensive right here because obviously
  100. 3:14Ferrari is trying to actually control
  101. 3:17the price on these. So, this is
  102. 3:19incredibly useful for doing research. It
  103. 3:21makes things more accurate and honestly,
  104. 3:23this just goes through and is able to
  105. 3:25give us a much better experience here
  106. 3:26from inside of Spark where it's going to
  107. 3:29be able to use the Chrome extension in
  108. 3:30order to actually get things done. For
  109. 3:32example, I have an automation running
  110. 3:33right here, which is a Gmail sponsor
  111. 3:35email automation where this actually
  112. 3:37goes through and does the work for all
  113. 3:39my different sponsor requests that I
  114. 3:41get. Now, the next change that I wanted
  115. 3:42to show you is if we come back over here
  116. 3:44to chat, we click on the model selector,
  117. 3:46we can now see that we have 3.7 flash.
  118. 3:49In addition to that, if we come into
  119. 3:51Spark right here, this is what is
  120. 3:52actually going to be controlling Spark,
  121. 3:54which is pretty awesome. Now, the reason
  122. 3:56that I would strongly suggest that you
  123. 3:58were using that model above anything
  124. 4:00else is because it is way better at
  125. 4:02agentic tasks. Let me explain. So, if we
  126. 4:05actually come over here and read the
  127. 4:06release notes for Gemini 3.7 Flash, we
  128. 4:08could see that this is way better at
  129. 4:11production code quality. In fact, the
  130. 4:13one thing I want to call out here is how
  131. 4:14much better it is in Claude 3.5 in a lot
  132. 4:16of different areas. Long horizon
  133. 4:18software engineering. This is actually
  134. 4:19the only thing that it gets beat by
  135. 4:21something by which GPT 5.6 Turbo, but if
  136. 4:24we come over here, enterprise workflow
  137. 4:26automation. I want you to think about
  138. 4:27this. Gemini 3.7 Flash is at 30%. Gemini
  139. 4:303.6 was at 17. Claude Sonnet 5 is at 10.
  140. 4:34GPT 5.6 Turbo is at 23. So, it is
  141. 4:37essentially one to two X better, in some
  142. 4:40cases three X better than all the other
  143. 4:42models that this can be compared to. And
  144. 4:44essentially, what this means is it's way
  145. 4:47better from going with a set of
  146. 4:48instructions to figuring out task one,
  147. 4:50task two, task three, task four, and
  148. 4:53completing it incredibly quickly. In
  149. 4:55fact, if we come back over here and we
  150. 4:57remember how quickly this actually did
  151. 4:59this right here, this did this faster
  152. 5:01than I would have been able to go to
  153. 5:03CarGurus, cars.com, Classic Cars JD,
  154. 5:06Bring a Trailer, DuPont, and Ferrari in
  155. 5:08order to figure out the price here. Now,
  156. 5:10the next thing I wanted to show you is
  157. 5:12that if we come over here into my launch
  158. 5:13pad, we open up ChatGPT now on a
  159. 5:16desktop. They got a brand new upgrade
  160. 5:18that literally changes everything about
  161. 5:21ChatGPT. So, now once this actually
  162. 5:23loads right here, we're going to come
  163. 5:24into the settings and we can turn this
  164. 5:26on so that whenever this app right here
  165. 5:28is on on my computer, it's actually
  166. 5:30taking notes of everything that I'm
  167. 5:32doing. So, we come over here, we click
  168. 5:34on settings, and we come into here and
  169. 5:37click right here. We're going to see
  170. 5:38computer history. What this is going to
  171. 5:40allow us to do is actually record. We
  172. 5:42can click on try again right here, and
  173. 5:43now this is going through and this is
  174. 5:45resuming, but we can see what this is
  175. 5:46actually doing is watching everything
  176. 5:49that I'm doing. For example, ChatGPT
  177. 5:50desktop recording workflow, you spent
  178. 5:52this window recording preparing content
  179. 5:54about ChatGPT desktop settings and
  180. 5:56histories features visible activity
  181. 5:58centered on this. This is actually all
  182. 5:59the information from the video that I
  183. 6:01recorded just a few days ago. And if we
  184. 6:03actually come over here and we click on
  185. 6:05reveal in finder right here, this is
  186. 6:07going to actually show us the MD files
  187. 6:09from this exact thing right here. We
  188. 6:11could go through it and see everything
  189. 6:12that ChatGPT actually took note of. Now,
  190. 6:14if you don't want this to actually be
  191. 6:16recording what you were doing, we could
  192. 6:18come over here into permissions and you
  193. 6:19could exclude certain apps, you could
  194. 6:21exclude certain websites. In addition to
  195. 6:23that, you could simply just come over
  196. 6:24here, you could click on pause, it will
  197. 6:26then be paused, and you could click on
  198. 6:28resume. So, whenever you're doing
  199. 6:29something sensitive, you're doing
  200. 6:30something you don't want ChatGPT to know
  201. 6:31about, no big deal. In addition to that,
  202. 6:33the next change that I wanted to show
  203. 6:35you here is now if you were using
  204. 6:37ChatGPT right here, under work, you
  205. 6:40could automate pretty much anything that
  206. 6:41you could think of so long as you have
  207. 6:43the right plugins, you have the right
  208. 6:45MCP, and you have the right prompt. And
  209. 6:47there's actually an AI tool that will
  210. 6:48figure this out for you. It's called
  211. 6:50llmhelper.ai
  212. 6:52and it allows you to get 10 times more
  213. 6:53out of ChatGPT, Claude, and Gemini. If
  214. 6:55we come over here, come into my
  215. 6:57dashboard right here, what we're going
  216. 6:58to see is absolutely insane. Because now
  217. 7:01what you're able to do with this tool is
  218. 7:03come over here into their automations
  219. 7:05builder and this allows you to choose
  220. 7:06which platform you're using. For
  221. 7:08example, ChatGPT, we could come over
  222. 7:10here, we could give it what my role is,
  223. 7:11for example, YouTuber, tasks I want
  224. 7:13automated, we could do script writing,
  225. 7:15finding content ideas, figuring out new
  226. 7:19sponsors, replying to sponsor emails,
  227. 7:23and basically we're just going to go
  228. 7:24through, tell it everything that we want
  229. 7:25automated, give it the tools that we
  230. 7:27use. I'm going to do VidIQ, Google
  231. 7:29Drive,
  232. 7:31and Gmail. And then what this is going
  233. 7:32to do, we click on build our
  234. 7:34automations, this goes through and this
  235. 7:36actually finds all the different
  236. 7:37connectors we should use, all the
  237. 7:39different MCP servers we should use, the
  238. 7:40exact prompts that we should use, and
  239. 7:42even how to fix those prompts if they
  240. 7:45don't work the first time. Check this
  241. 7:47out. So, this tells me exactly what the
  242. 7:49recommended setup is. It goes through
  243. 7:51and actually gives me all of these
  244. 7:53different automations. Gives me the
  245. 7:54exact prompt where I could actually go
  246. 7:56through and adjust anything that I want
  247. 7:59here. In addition to that, if we
  248. 8:01actually come over here, we could see
  249. 8:03everything that this actually allows us
  250. 8:05to input, all the different things that
  251. 8:06we're going to have to do here. Gives us
  252. 8:08instructions on how to actually build
  253. 8:10this out. And this is absolutely
  254. 8:12incredible. We're simply just going to
  255. 8:14copy this prompt. We're going to put in
  256. 8:16the things where we need to insert
  257. 8:17things right here, and then boom, this
  258. 8:19will be able to automate things. And if
  259. 8:20we had trouble with it, we could
  260. 8:22actually come over here and paste this
  261. 8:23repair prompt right here, where this
  262. 8:25actually goes through and will walk
  263. 8:27ChatGPT through how to repair it if it's
  264. 8:29broken, it breaks, or something went
  265. 8:32wrong. And the best part about LLM
  266. 8:33Helper is that if you go to the pinned
  267. 8:35comment below, you can try it today for
  268. 8:36free. This tool is an absolute
  269. 8:38no-brainer. It's going to show you
  270. 8:39exactly what kind of things you can
  271. 8:40automate. It's going to give you the
  272. 8:41exact prompts to do that, and it makes
  273. 8:43ChatGPT, Claude, and Gemini so much more
  274. 8:46powerful. So, go to that pinned comment
  275. 8:47below, get started with it today. It's
  276. 8:49personally allowed me to save over 10
  277. 8:51hours a week from tedious tasks that I
  278. 8:53used to do, because now they're
  279. 8:54automated. Now, the next update I wanted
  280. 8:56to show you here is going to be the
  281. 8:57ChatGPT Chrome extension. So, now we can
  282. 8:59see that we can actually open up ChatGPT
  283. 9:01right here, and this is going to sync
  284. 9:02with ChatGPT on my computer and ChatGPT
  285. 9:05on my phone. Let's say that we are on
  286. 9:06this article right here. What we can
  287. 9:08actually do is come over here and say
  288. 9:09something like, "Can you write me a
  289. 9:12script in my usual format for a 45- to
  290. 9:1560-second short on the topic of my
  291. 9:19screen?" We're going to come over here.
  292. 9:21We're going to click on this plus. We
  293. 9:23can scroll down right here, and we can
  294. 9:24see that we can grab this right here.
  295. 9:26Then, this is actually going to go
  296. 9:28through this. This is going to use the
  297. 9:29skill and the memory that it has, how I
  298. 9:31actually write my scripts, and it is
  299. 9:33going to write me a script for this
  300. 9:34right here. So, I'm using your YouTube
  301. 9:36script voice skill to match your
  302. 9:38established short-form structure. Then,
  303. 9:39I'm going to ground the script in the
  304. 9:41article that is currently open on your
  305. 9:43screen. It's then going through It is
  306. 9:44doing that. What this is going to be
  307. 9:46able to do is basically allow us to use
  308. 9:48ChatGPT anywhere that we want because
  309. 9:51now we can literally just come over here
  310. 9:53and do this on anything. In addition to
  311. 9:55that, we can see this literally wrote
  312. 9:57perfect script right here. If we wanted
  313. 9:59to actually have multiple things, we
  314. 10:00could have multiple things open. We
  315. 10:02could have it actually grab files from
  316. 10:03my computer. What we can do here is
  317. 10:05absolutely incredible. In addition to
  318. 10:07that, they also have plan mode, goal,
  319. 10:10and record a skill. If you record a
  320. 10:11skill, this is going to record your
  321. 10:13screen, it's going to record your voice,
  322. 10:15it's going to record what you do,
  323. 10:17decipher that into a skill that this can
  324. 10:19then automate. On plan mode right here,
  325. 10:21you tell it what you want to do, and it
  326. 10:23will actually build a plan around it
  327. 10:25that can then implement. In addition to
  328. 10:27that, with goal, you could set a goal,
  329. 10:29and it will keep pursuing that goal,
  330. 10:31which is pretty crazy. I wouldn't be too
  331. 10:33ambitious like this. Like, don't tell it
  332. 10:34to build you a business doing $10,000 a
  333. 10:36month because it's going to eat all your
  334. 10:38credits, and it probably won't be able
  335. 10:40to do that because obviously AI can't do
  336. 10:43that. Now, the other thing that I did
  337. 10:44want to show you, which I find pretty
  338. 10:46incredible, is that from inside of the
  339. 10:49app now, what we could do here is pretty
  340. 10:50cool. There's a few approvals here. So,
  341. 10:53we have for ask for approval, approve
  342. 10:54for me, full access. If you give it full
  343. 10:57access, ChatGPT has unrestricted access
  344. 10:59to the internet and any file on your
  345. 11:01computer. I would do approve for me, so
  346. 11:03it only asks for actions that are
  347. 11:04detected as potentially unsafe, but it's
  348. 11:07not taking any of its capabilities away.
  349. 11:09If you do ask for approval, you might as
  350. 11:11well just be doing the work yourself
  351. 11:12because you're constantly going to have
  352. 11:14to be approving things. In addition to
  353. 11:15that, right here, we could see that we
  354. 11:17could choose where chat can run. We
  355. 11:19could have it run on our computer, which
  356. 11:20is going to allow it to read and edit
  357. 11:22local files, of course, with our
  358. 11:24permission, or in the cloud. It won't be
  359. 11:26able to access local files, but it'll be
  360. 11:28able to run 24/7, we'll be able to
  361. 11:30control it from our computer, from our
  362. 11:33desktop, or from any other device that
  363. 11:35we are logged into ChatGPT, and that is
  364. 11:37pretty cool. In addition to that, under
  365. 11:39settings right here, if we come under
  366. 11:41voice, we're going to see we now have a
  367. 11:42bunch of brand new voice acts. You can
  368. 11:44actually create a voice chat hot key,
  369. 11:47which will allow you to start your voice
  370. 11:48chat from anywhere on the desktop. So,
  371. 11:50ChatGPT is basically always listening.
  372. 11:52You could give it screen context, so you
  373. 11:53could like code X inspect the foreground
  374. 11:56app when you refer to what's on your
  375. 11:57screen. In addition to that, with
  376. 11:59dictation, you could hold the dictate,
  377. 12:01you could toggle dictation as a hot key,
  378. 12:03and you could actually create a journal
  379. 12:06where words or phrases that should be
  380. 12:07recognized will be recognized, and you
  381. 12:09could actually see your last 20
  382. 12:12recordings on this device. In addition
  383. 12:14to that, if we come over here and
  384. 12:15actually click on computer use, if you
  385. 12:17turn all these on, this will let ChatGPT
  386. 12:19control any app on your computer,
  387. 12:21basically control your computer. You do
  388. 12:23the same exact thing with a browser
  389. 12:24right here, where you could let ChatGPT
  390. 12:26control its built-in browser. You could
  391. 12:28manage your history, manage your
  392. 12:30passwords, you could manage your
  393. 12:31location, you could manage all of that
  394. 12:33from directly inside of here. If you
  395. 12:35want to get the most out of ChatGPT, I
  396. 12:37would strongly suggest that you're using
  397. 12:39these different things, you're turning
  398. 12:40these different settings on, because it
  399. 12:41really does make a difference. And then,
  400. 12:43the last thing I wanted to show you is
  401. 12:45that now, from inside of ChatGPT, if you
  402. 12:47come into ChatGPT work, you could do
  403. 12:49@googledrive, and we can actually get
  404. 12:51this to edit or create anything inside
  405. 12:54of a Google Drive, be it a sheet, be it
  406. 12:55a doc, and we could see right here that
  407. 12:57we now don't actually have to leave
  408. 12:59ChatGPT, don't have to change tabs. We
  409. 13:01have our doc right here on the
  410. 13:02right-hand side while also having
  411. 13:05ChatGPT pulled up. So, the two things
  412. 13:07are open simultaneously, and this is
  413. 13:09just a way better experience if you're
  414. 13:11doing anything inside of sheets, inside
  415. 13:14of a Google Drive, inside of Google
  416. 13:15Docs, and this personally has made my
  417. 13:17life way easier, because now I can
  418. 13:19script things, it can reference my
  419. 13:21email, all of that, all from just one
  420. 13:23place without a having multiple tabs
  421. 13:25open, which is how I operated in the
  422. 13:27past. Now, if you enjoyed this video and
  423. 13:29you want to learn about a brand new way
  424. 13:30for you to be able to build voice agents
  425. 13:32that can answer the phone for you, that
  426. 13:33can call people for you, and so much
  427. 13:35more, I would strongly suggest you check
  428. 13:36out this video right here that walks you
  429. 13:38through exactly how that works because I
  430. 13:39just released it and it's crazy.

About this transcript

This page contains the full transcript of ChatGPT & Google Gemini’s New Features Are Crazy! (Remove Watermarks, History & More) by Rob The AI Guy, generated from the public captions YouTube serves with the video. The transcript has 3,078 words across 430 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 translate the transcript into 100+ languages.

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, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.