YouTube2Text

AIを導入したのに生産性が落ちる…今、企業で起きているヤバい現象とは【マイキー佐野 投資学】 — Transcript

by マイキーの非道徳な社会学 · 3,095 words · 552 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Usually, it should go down before you
  2. 0:01even ride it. That's exactly why. So,
  3. 0:04anyone who says their productivity went
  4. 0:06up is probably not using AI correctly.
  5. 0:11Hi, this is Yuki. I'm shooting for
  6. 0:13YouTube again today.
  7. 0:14Let's get started. Let's do it.
  8. 0:15This time, even though we’ve already
  9. 0:17talked a lot about semiconductors and
  10. 0:20AI—I mean, you gave me dozens of
  11. 0:22companies before—Yes.
  12. 0:24I’m not asking for new company names,
  13. 0:26but I want to shift perspectives and
  14. 0:28have you tell me the points in the AI
  15. 0:30industry that we should keep an eye on.
  16. 0:33Distortion.
  17. 0:34Distortion?
  18. 0:35In a word, distortion.
  19. 0:37Yes.
  20. 0:38So, to explain what that means, things
  21. 0:40called intelligence—like marginal
  22. 0:41costs or token prices—These are
  23. 0:43dropping like crazy, right? Conversely,
  24. 0:45there’s the physical infrastructure
  25. 0:47that supports that foundation, right?
  26. 0:48Data centers, semiconductors, and so on
  27. 0:50—the input costs are skyrocketing,
  28. 0:52creating an inflationary state. So, you
  29. 0:55understand what’s happening: buying
  30. 0:57expensive equipment to create cheap
  31. 0:59products.
  32. 0:59Expensive equipment.
  33. 1:00Introducing it.
  34. 1:01Introducing it.
  35. 1:02And making cheap products; that’s
  36. 1:04what it is.
  37. 1:05And that’s the token price. Even
  38. 1:06though costs are rising more and more,
  39. 1:09the cost per token is starting to drop
  40. 1:11significantly. In traditional economic
  41. 1:13theory, technological evolution lowers
  42. 1:15overall production costs. And
  43. 1:17ultimately, production costs are
  44. 1:19reduced, leading to lower prices for
  45. 1:21products or services. But it's common
  46. 1:24for production costs to fall below the
  47. 1:26selling price.
  48. 1:27As production costs fall below the
  49. 1:29selling price, profits just keep
  50. 1:31growing, right? Right, that’s the
  51. 1:33fundamental, traditional theory of
  52. 1:35economics. So, as I said, while
  53. 1:37technological evolution lowers overall
  54. 1:40production costs, the prices of
  55. 1:42products and services eventually fall.
  56. 1:45But what’s happening now is the
  57. 1:46opposite: the selling price is dropping
  58. 1:49lower than production costs, or to put
  59. 1:51it another way, production costs are
  60. 1:53rising while prices fall, creating a
  61. 1:55structure that is squeezing profits,
  62. 1:57which is the current reality. To
  63. 2:02explain this more academically, when we
  64. 2:04talk about the shrinking profit margins
  65. 2:06in the AI ecosystem, it’s easiest to
  66. 2:08start by dividing it into upstream and
  67. 2:11downstream. Upstream refers to
  68. 2:13infrastructure and semiconductor
  69. 2:15companies. Downstream refers to model
  70. 2:18development and the provision of APIs.
  71. 2:20Cutting-edge semiconductors,
  72. 2:21Yeah.
  73. 2:22The HBM market and physical production
  74. 2:24capacity are reaching their limits. And
  75. 2:27another factor is that the technical
  76. 2:29intensity has created a structurally
  77. 2:31undersupplied state. In fact, it seems
  78. 2:33there’s a shortage of memory as well.
  79. 2:36NVIDIA itself, as I mentioned, has
  80. 2:38already hit 290 billion just in orders,
  81. 2:41but what are memory firms like SK
  82. 2:43projecting for the future? They’ve
  83. 2:47signed long-term contracts with data
  84. 2:49center clients. In other words, 70%of
  85. 2:52the maximum supply is already secured
  86. 2:54elsewhere, so the outlook is that the
  87. 2:56memory shortage will continue. It’s
  88. 2:58going to be tight. Samsung and other
  89. 3:01foundries are raising their prices by
  90. 3:03up to 15%, and NVIDIA is doing the same
  91. 3:06with a 15%hike. This was in the news
  92. 3:08recently, too. So, as you can see, the
  93. 3:10upstream sector overwhelmingly holds
  94. 3:13the pricing power. That is the current
  95. 3:15situation. NVIDIA’s recent earnings
  96. 3:17were quite bullish, though. Actually,
  97. 3:19they need 110 billion dollars—
  98. 3:21including memory and various other
  99. 3:24parts—to utilize their future GPUs.
  100. 3:26This figure went from 100 billion to
  101. 3:29290 billion in just one year. Meanwhile
  102. 3:31, the downstream side—the ones being
  103. 3:34supplied—are being charged these
  104. 3:36incredibly high prices by the suppliers
  105. 3:38. For instance, look at China's
  106. 3:40DeepSeek going open source; it’s
  107. 3:42actually causing prices to drop, right?
  108. 3:44This is, well, part of China's strategy
  109. 3:46. I mean, China itself is being heavily
  110. 3:49restricted by the U.S. in terms of
  111. 3:51capital investment. While companies
  112. 3:53like Huawei and SMIC are emerging, by
  113. 3:55going open source, they’re preventing
  114. 3:57U.S. firms from collecting subscription
  115. 3:59fees. And have you heard of the new one
  116. 4:01that came out recently, Qwen-Max?
  117. 4:02Another new one just appeared. Well, a
  118. 4:04lot of them are popping up. Yes.
  119. 4:06Yes. This is causing further price
  120. 4:10destruction, and downstream, companies
  121. 4:12have no pricing power because customers
  122. 4:14already hold the leverage.
  123. 4:16They can just choose the cheaper option
  124. 4:18. I mean, I'm using a 3,000 yen service
  125. 4:20myself.
  126. 4:21So, the customer has all the bargaining
  127. 4:23power; if they want to find a service,
  128. 4:25they can find countless options for any
  129. 4:27price point.
  130. 4:28That's why buyers are in a position to
  131. 4:29choose at their own convenience.
  132. 4:31It’s that kind of distortion, isn't
  133. 4:32it?
  134. 4:32Exactly. Yes. Yes. Yes.
  135. 4:35So, we’re seeing a state of both
  136. 4:36inflation and deflation.
  137. 4:39This is happening within a single
  138. 4:40industry, meaning it's getting harder
  139. 4:42to pull a profit, which is why
  140. 4:43companies are being grilled by
  141. 4:45investors and the market. In this
  142. 4:49situation, it’s becoming very
  143. 4:50difficult for development firms and
  144. 4:52infrastructure providers to justify
  145. 4:54their position, and the reality is that
  146. 4:55their profit structures are being
  147. 4:57squeezed. That’s why you call it a
  148. 4:59distortion.
  149. 5:01But there are both bulls and bears when
  150. 5:02it comes to this current AI trend. For
  151. 5:06example, the bulls argue that as the
  152. 5:08price of tokens drops, demand will
  153. 5:10explode, ultimately increasing total
  154. 5:12revenue. We covered this on YouTube
  155. 5:14before. Jevons paradox.
  156. 5:15Ah, yes, we did.
  157. 5:17The steam engine one, right?
  158. 5:18Yes. Exactly. We talked about that, and
  159. 5:21it's essentially an argument for
  160. 5:22volume-driven growth. I mean, these
  161. 5:24companies are paying massive amounts of
  162. 5:26capital to hyperscalers and for AI,
  163. 5:27right? It's seen as strong leadership,
  164. 5:29but when you actually examine corporate
  165. 5:31financial data at the micro level,
  166. 5:33there’s a significant disconnect
  167. 5:35between investment and profitability.
  168. 5:40There are these AI advisory firms like
  169. 5:42Mackey Research and Spek Research. They
  170. 5:46conducted a quantitative analysis of
  171. 5:48919 quarterly earnings reports from the
  172. 5:51top 60 major US financial institutions
  173. 5:54over the past three years. And what
  174. 5:57actually happened? In about 80%of those
  175. 6:01earnings calls—four out of five—AI
  176. 6:04was mentioned. However, more than 50%of
  177. 6:07the discussion was about rising
  178. 6:09implementation costs and increased
  179. 6:12CAPEX. So, in the entire study, only
  180. 6:15two companies were able to quantify and
  181. 6:18clearly state the actual monetary
  182. 6:20returns generated by AI implementation.
  183. 6:23Basically, there's no profit coming in.
  184. 6:25The reality is that for executives,
  185. 6:27marketing, and PR, the only thing being
  186. 6:30pushed is a story about dramatic
  187. 6:32productivity gains from AI.
  188. 6:34Yes. Yes. Yes.
  189. 6:36So, when you look at the gap between
  190. 6:38the actual financial situation and the
  191. 6:40earnings calls, the conclusion is that
  192. 6:42there is almost no quantitative proof
  193. 6:44of profitability so far. Actually, this
  194. 6:49is something I predicted on YouTube two
  195. 6:51or three years ago.
  196. 6:53I'll do my best to find it. Yes.
  197. 6:55Please do. It’s something called the
  198. 6:58Solow Paradox.
  199. 6:59Oh, the paradox, yes.
  200. 7:03It was proposed in 1987 by Robert Solow
  201. 7:06, a Nobel laureate in economics, who
  202. 7:08pointed out the economic mystery that "
  203. 7:10computers are everywhere except in the
  204. 7:12productivity statistics." That is the
  205. 7:16Solow Paradox. This PC and internet
  206. 7:19revolution, it actually takes ten years
  207. 7:21for the productivity to show up in the
  208. 7:23numbers. I've talked about this on
  209. 7:26YouTube many times. This is the content
  210. 7:27I mentioned earlier. The expectation is
  211. 7:30that by introducing computers, business
  212. 7:32efficiency will rise and social
  213. 7:34productivity will grow rapidly. But the
  214. 7:38reality is, despite huge investments,
  215. 7:40national economic growth and corporate
  216. 7:42productivity remain mostly flat. This
  217. 7:45is the current situation. The
  218. 7:48background to this is that even if new
  219. 7:50technology spreads, it just becomes a
  220. 7:52cost center unless workflows and
  221. 7:54organizational structures evolve. I
  222. 7:57explained this on YouTube before, right
  223. 7:59?
  224. 8:00In fact, the situation where AI
  225. 8:02adoption actually increases labor hours
  226. 8:04in Silicon Valley is exactly that
  227. 8:06background. What's happening with AI
  228. 8:09investment now is exactly this Solow
  229. 8:11Paradox, and what we're actually seeing
  230. 8:13is FOMO. There's the fear of missing
  231. 8:16out, the anxiety that if you fall
  232. 8:18behind others, you might eventually
  233. 8:20lose or be unable to keep up. So, there
  234. 8:24are many investors who think people are
  235. 8:26investing without calculating clear
  236. 8:28returns or cost-effectiveness. It's
  237. 8:35just a scramble; it's the idea that AI
  238. 8:37costs are being wasted as defensive
  239. 8:39spending to steal customers or market
  240. 8:41share from rivals rather than
  241. 8:42increasing overall wealth, which is the
  242. 8:44IT productivity paradox. There is
  243. 8:53something called the IT productivity
  244. 8:55paradox, where simply introducing
  245. 8:56convenient AI or IT without
  246. 8:58fundamentally changing workflows or
  247. 9:00organizations leads to failure. That is
  248. 9:05what they call the IT productivity
  249. 9:07paradox. This is exactly what is
  250. 9:08happening right now. So, does that mean
  251. 9:10AI isn't improving productivity? Well,
  252. 9:13I’ll explain why it’s not a waste.
  253. 9:17This is what some call the "intangible
  254. 9:19asset time lag," referring to what
  255. 9:21happens during the initial adoption
  256. 9:23phase. Everyone is using AI, but more
  257. 9:27importantly, organizations need to
  258. 9:29fundamentally change how they operate.
  259. 9:33It’s a gap or time lag created by how
  260. 9:35much intangible capital was built up
  261. 9:36versus how much is actually integrated
  262. 9:38into the organization. This is
  263. 9:41something that happens naturally. It
  264. 9:43really does. This is exactly what
  265. 9:45happened during the IT boom, too. Even
  266. 9:47if you adopt IT, you won't maximize its
  267. 9:49full potential unless you fundamentally
  268. 9:51restructure your operations and
  269. 9:52organization. The same thing is
  270. 9:56happening with AI now, which brings us
  271. 9:58to the "Productivity J-Curve."
  272. 10:01There’s a theory called the
  273. 10:02Productivity J-Curve, which is a key
  274. 10:04concept in economic productivity
  275. 10:06analysis. Here’s the interesting part
  276. 10:08. It explains a dynamic process where
  277. 10:12Total Factor Productivity (TFP) and
  278. 10:14corporate profitability initially
  279. 10:16decline before eventually rising. As I
  280. 10:20mentioned, in Silicon Valley, companies
  281. 10:23saw their productivity drop after
  282. 10:25adopting AI. In fact, their labor hours
  283. 10:27actually increased. In Japanese
  284. 10:30companies, people mistakenly think AI
  285. 10:32just makes their tasks easier, but in
  286. 10:35reality, using AI creates more
  287. 10:36challenges, hypotheses, and testing,
  288. 10:38which ultimately increases labor hours
  289. 10:41—this is exactly the Productivity
  290. 10:43J-Curve. This concept was proposed by
  291. 10:47three American economists. This is a
  292. 10:49theory developed by a Stanford
  293. 10:51professor, Danny Rock from the Wharton
  294. 10:52School at the University of
  295. 10:54Pennsylvania, and Chayberson from the
  296. 10:56University of Chicago. As I explained
  297. 10:58regarding the difference between
  298. 11:00Japanese companies and AI firms,
  299. 11:01startup companies are seeing their
  300. 11:03volume increase, so if you'd like to
  301. 11:05know more, please check out my YouTube
  302. 11:06channel. Yes. I explain it there.
  303. 11:10During this early adoption phase,
  304. 11:12companies need to allocate capital and
  305. 11:13labor not only to direct hardware and
  306. 11:15software costs but also to intangible
  307. 11:17assets that are difficult to measure.
  308. 11:21That’s why it takes time to build the
  309. 11:23foundation. So, when it comes to
  310. 11:25redesigning business processes to adapt
  311. 11:27operations to AI-supported workflows,
  312. 11:29structural changes take time, which is
  313. 11:31why there’s a period where
  314. 11:33productivity drops. Regarding machine
  315. 11:37learning, there is the human capital
  316. 11:39aspect; this needs to be retrained, and
  317. 11:41it takes time for employees to actually
  318. 11:43acquire the skills to use AI tools
  319. 11:45effectively. Also, with data management
  320. 11:49infrastructure and system updates, you
  321. 11:51need to build data pipelines and design
  322. 11:53organizational structures, so a
  323. 11:55temporary drop is only natural. This is
  324. 11:58the productivity curve. Conversely,
  325. 12:00anyone saying their productivity went
  326. 12:01up is probably not using AI correctly.
  327. 12:03That's right. Yes. Yes. Building these
  328. 12:06things requires labor hours and
  329. 12:08resources. These are recorded as
  330. 12:11expenses.
  331. 12:12But you are essentially building
  332. 12:14intangible assets. These are not
  333. 12:16recorded on the balance sheet. This is
  334. 12:18a situation that happens very often.
  335. 12:20The point is, this is standard. And
  336. 12:22there is also empirical evidence of
  337. 12:24this J-curve based on microdata. To put
  338. 12:27it simply, there is a government agency
  339. 12:29under the Department of Commerce called
  340. 12:31the Census Bureau that recently
  341. 12:32analyzed statistical data on the U.S.
  342. 12:34population and economy. This data was
  343. 12:37actually used to conduct industrial AI
  344. 12:39demonstration experiments in 2025. And
  345. 12:42what was the result? It showed that
  346. 12:44initial productivity actually dropped
  347. 12:47by 44%. So, it turned out that the
  348. 12:49sites where AI was introduced
  349. 12:50experienced that deep "J-curve" dip
  350. 12:52just as the theory predicted. There is
  351. 12:55plenty of proven data on how to change
  352. 12:57conventional work rules, or on the
  353. 12:59confusion in the field, such as time
  354. 13:01spent on system adjustments or
  355. 13:02inventory increasing because processes
  356. 13:04weren't syncing up. So, productivity is
  357. 13:08actually decreasing, and this is
  358. 13:09something confirmed by studies at
  359. 13:11Stanford and the University of Toronto;
  360. 13:13introducing AI can actually lower
  361. 13:15productivity. If used correctly, most
  362. 13:17of the time—though not in every case.
  363. 13:19But Japanese people tend to be a bit
  364. 13:21naive, thinking, "My company is
  365. 13:22different." They claim, "Our
  366. 13:24productivity is going up," but they are
  367. 13:25likely wrong in most cases and probably
  368. 13:27need to change their organizational
  369. 13:29structure, or—
  370. 13:29Wait, we weren't talking about that,
  371. 13:31were we? Please don't go off on a
  372. 13:32tangent. Sorry for the trash talk. Yeah
  373. 13:34, yeah, yeah, there are a lot of people
  374. 13:35like that.
  375. 13:36Well, well, well, well.
  376. 13:37Research was published by universities
  377. 13:40like Stanford, Oxford, and MIT on
  378. 13:41something called the "Cognitive
  379. 13:43Offloading Model 2026." What this means
  380. 13:45is that by offloading thinking tasks
  381. 13:47and decision-making to AI, labor time
  382. 13:49actually decreases. But I always say
  383. 13:51this, and I use AI extensively. Labor
  384. 13:54time actually increases. So, labor time
  385. 13:56goes up. This means we're in a state of
  386. 13:58constantly delegating cognition to AI.
  387. 14:01And as a result, the expertise and
  388. 14:02field intuition that humans have
  389. 14:04cultivated over many years begin to
  390. 14:06fade. That's why AI on its own doesn't
  391. 14:08create value; it doesn't show its true
  392. 14:10potential unless combined with human
  393. 14:12judgment. In fact, human skills
  394. 14:14themselves begin to rust. So, the
  395. 14:16conclusion is that if you can't fully
  396. 14:18utilize the potential of AI, the time
  397. 14:20it takes to climb out of the bottom of
  398. 14:22the J-curve or see a sharp rise in
  399. 14:24productivity will be delayed. But those
  400. 14:26with an even more naive outlook say
  401. 14:28that doesn't concern them, because now
  402. 14:30they can easily create reports. For
  403. 14:32those whose working hours dropped from
  404. 14:34eight to four because reports became
  405. 14:36easy to make, they won't even fall into
  406. 14:38a J-curve to begin with; they'll just
  407. 14:40end up in a state of blissful ignorance
  408. 14:42.
  409. 14:42I mean, that really is just losing to
  410. 14:44AI. That is,
  411. 14:45Only people who aren't thinking clearly
  412. 14:46would say something like that.
  413. 14:47It’s true that a temporary 44%drop in
  414. 14:50productivity due to operational
  415. 14:51confusion on the ground is unavoidable.
  416. 14:54However, the fact that working hours
  417. 14:55are actually decreasing just means
  418. 14:57people are getting better at slacking
  419. 14:59off, which means we are facing a
  420. 15:01degradation in judgment standards. To
  421. 15:05summarize this, at the micro level,
  422. 15:07investment returns are becoming delayed
  423. 15:09, while upstream, massive investment
  424. 15:12continues, and they hold bargaining
  425. 15:14power, effectively influencing
  426. 15:15financial and credit markets, which
  427. 15:17explains the current state. In the
  428. 15:22stock market, there is an AI frenzy,
  429. 15:24but in contrast, bonds and credit
  430. 15:26default swaps (CDS) are rising, showing
  431. 15:29growing alarm over debt financing for
  432. 15:31AI infrastructure. With news of
  433. 15:35Broadcom's chip development and
  434. 15:37negotiations for 60 billion in
  435. 15:38infrastructure funding, leveraged
  436. 15:40financing has increased, causing CDS
  437. 15:43credit spreads to widen. As a result,
  438. 15:45risk premiums have risen. The CDS
  439. 15:47spread for Broadcom has actually risen
  440. 15:5080 points since the beginning of the
  441. 15:52year. The same thing happened with
  442. 15:54NVIDIA, you see. Before the 500 billion
  443. 15:56funding was even announced, the CDS
  444. 15:58spread shot up. NVIDIA itself, that is.
  445. 16:01So, this is a market analysis report
  446. 16:03from Goldman Sachs. While EPS estimates
  447. 16:06have been revised upward, the P/E
  448. 16:08ratios for AI-related stocks have
  449. 16:10actually been trending downward
  450. 16:12recently. They are seeing a downward
  451. 16:14trend lately. In other words, the
  452. 16:17market isn't just looking at
  453. 16:19speculation or raw revenue anymore; the
  454. 16:21core issue has become who can generate
  455. 16:23cash flow and when. Goldman Sachs
  456. 16:27argues that this is the shift currently
  457. 16:29taking place. It goes back to that
  458. 16:31J-curve I mentioned earlier.
  459. 16:33Productivity initially drops in the
  460. 16:35short term, but it tends to rise
  461. 16:37afterward. If the right processes are
  462. 16:39in place, this is exactly what happened
  463. 16:41during the previous AI and IT eras.
  464. 16:43With that in mind, we need to consider
  465. 16:44how to view this current divergence. I
  466. 16:46think looking at this data will give
  467. 16:48you a clear perspective. By the way,
  468. 16:49it’s a fact that the wider this
  469. 16:51divergence becomes, the harder it is to
  470. 16:52recover. Yeah.
  471. 16:53Absolutely.
  472. 16:54Yes.
  473. 16:55However, as I mentioned earlier,
  474. 16:56there's Jevons paradox—if this
  475. 16:58spreads across the board, the situation
  476. 17:00could change entirely. Just keep that
  477. 17:02in mind. Right. Anyway, I hope this was
  478. 17:04helpful for everyone today. We talked
  479. 17:06about the AI divergence, and if you
  480. 17:08want more detailed information, please
  481. 17:10be sure to read the full report. It's
  482. 17:11an official publication, so I highly
  483. 17:13encourage you to get your hands on that
  484. 17:15report. Also, if you found today’s
  485. 17:17video interesting, educational, or
  486. 17:20helpful, please subscribe, like,
  487. 17:22comment, and share it with others.
  488. 17:24We talked about NVIDIA’s earnings
  489. 17:26during today's shoot, right?
  490. 17:27It’s been uploaded. Yeah.
  491. 17:29Right. It's totally fine, but after
  492. 17:31hearing about the outlook through 2028
  493. 17:33and now this divergence, I honestly
  494. 17:35don't know how I should be looking at
  495. 17:37the AI industry anymore.
  496. 17:39See what I mean?
  497. 17:39Right, right, right, right, right. You
  498. 17:41see, there's a next one coming.
  499. 17:43Wait, there's more?
  500. 17:44Yes. I’ve already told my business
  501. 17:46school students to look into this
  502. 17:48months ago, but honestly,
  503. 17:50I’d be a bad person if I didn’t
  504. 17:51talk about it on YouTube.
  505. 17:53Eventually.
  506. 17:54Yes.
  507. 17:54I will talk about it.
  508. 17:55Oh, I’ll look forward to that.
  509. 17:56Listening to it, I couldn't help but
  510. 17:58wonder, what's the point of this?
  511. 17:59Yes. Well, there’s actually a real
  512. 18:00point to it.
  513. 18:01Yeah.
  514. 18:02No, but I use AI daily, and I do wonder
  515. 18:05if my money is even helping to cover
  516. 18:07those massive capital investments. I
  517. 18:10really do think about things like that.
  518. 18:12I mean, well, honestly, I really wonder
  519. 18:14if they can ever recoup the costs. I
  520. 18:16mean, there are tons of people only
  521. 18:18paying 3,000 yen, right?
  522. 18:20Especially with the wealth gap widening
  523. 18:21like this, and in places like China,
  524. 18:23it’s cheap and they can use it freely
  525. 18:25.
  526. 18:25Yes. The meaning of these 500 billion
  527. 18:27yen figures is just becoming harder and
  528. 18:29harder to understand. Well, but that
  529. 18:30part is, well, it’s interesting to
  530. 18:32hear about. Yes. So,
  531. 18:34the question is, what’s the next step
  532. 18:36? And if you look here, you can see how
  533. 18:38that industry and the companies
  534. 18:39themselves can recover their profits.
  535. 18:41If you ask what this is,
  536. 18:42Oh, that’s where it leads.
  537. 18:44Exactly.
  538. 18:45Ah, another new genre has appeared.
  539. 18:47Yes. This content,
  540. 18:49Yes.
  541. 18:50I was looking forward to seeing when it
  542. 18:51would come up.
  543. 18:52I’ve been checking YouTube for almost
  544. 18:54two years, but nobody is talking about
  545. 18:55it.
  546. 18:56Yeah, they don't talk about it.
  547. 18:57So I thought I’d stop talking about
  548. 18:59it too.
  549. 18:59I mean, everyone, I know, so I’m not
  550. 19:01going to tell you guys.
  551. 19:03I did talk about it at the business
  552. 19:04school, though. Yes. Ha.

About this transcript

This page contains the full transcript of AIを導入したのに生産性が落ちる…今、企業で起きているヤバい現象とは【マイキー佐野 投資学】 by マイキーの非道徳な社会学, generated from the public captions YouTube serves with the video. The transcript has 3,095 words across 552 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.