WEBVTT - Monologue: How Does This End?

0:00:02.160 --> 0:00:08.039
<v Speaker 1>Zone Media, Hello, and welcome to your weekly Better Offline Monologue. I,

0:00:08.119 --> 0:00:17.959
<v Speaker 1>of course amed Xytron. So, as you'll find out in

0:00:17.960 --> 0:00:21.040
<v Speaker 1>tomorrow's episode, the Future of jennetor of Ai, hinges heavily

0:00:21.040 --> 0:00:23.840
<v Speaker 1>on open Ai, raising tens of billions of dollars, the

0:00:23.920 --> 0:00:27.360
<v Speaker 1>majority of it from SoftBank, a huge Japanese investment firm

0:00:27.520 --> 0:00:29.720
<v Speaker 1>that has to take out billions of dollars of loans

0:00:29.760 --> 0:00:32.879
<v Speaker 1>to fund them, along with their contribution to the Stargate

0:00:32.960 --> 0:00:36.760
<v Speaker 1>data center project. And as I went into in yesterday's episode,

0:00:36.960 --> 0:00:39.600
<v Speaker 1>Microsoft is pulling back from over a gigawatt of data

0:00:39.640 --> 0:00:42.360
<v Speaker 1>center capacity, and it certainly looks like these moves are

0:00:42.360 --> 0:00:45.520
<v Speaker 1>an intentional move to distance themselves from open Ai, cutting

0:00:45.520 --> 0:00:48.360
<v Speaker 1>back on data center expansion just as America's worst company

0:00:48.400 --> 0:00:52.320
<v Speaker 1>needs more of them. Meanwhile, Mark Benioff, CEO of Salesforce,

0:00:52.640 --> 0:00:55.920
<v Speaker 1>he's sounded Neil arm saying a recent CNBC conference that

0:00:55.960 --> 0:00:59.440
<v Speaker 1>he believed hyperscalers were under hypnosis in their aggressive pursuit

0:00:59.480 --> 0:01:02.320
<v Speaker 1>of data center to expansion and training larger and larger

0:01:02.400 --> 0:01:06.720
<v Speaker 1>language models. Benioff believes that and I quote ish it,

0:01:07.080 --> 0:01:10.400
<v Speaker 1>referring to data center expansions and larger language models has

0:01:10.440 --> 0:01:13.840
<v Speaker 1>to be rethought. Exactly what are you doing and why

0:01:13.880 --> 0:01:17.400
<v Speaker 1>are you doing this? That's a bloody good question. Mark.

0:01:18.120 --> 0:01:20.679
<v Speaker 1>To be clear, Mark Benioff has been saying that Salesforce

0:01:20.760 --> 0:01:23.520
<v Speaker 1>was adding some sort of Einstein AI shit for the

0:01:23.520 --> 0:01:25.400
<v Speaker 1>best part of a decade as a means of boosting

0:01:25.400 --> 0:01:29.280
<v Speaker 1>his stock price. So why is Big tech's most effusive

0:01:29.319 --> 0:01:32.320
<v Speaker 1>bullshit as saying this. Do you think it's because things

0:01:32.360 --> 0:01:34.800
<v Speaker 1>are going well? Do you think it's because sales of

0:01:34.840 --> 0:01:39.240
<v Speaker 1>Agent Force and other associated products are doing really well? Look,

0:01:39.680 --> 0:01:42.720
<v Speaker 1>as I've repeatedly said, where is the money in this industry?

0:01:42.920 --> 0:01:46.319
<v Speaker 1>What have these companies actually built with generative AI? Where

0:01:46.360 --> 0:01:48.639
<v Speaker 1>are the products that matter and why do they matter?

0:01:49.360 --> 0:01:52.960
<v Speaker 1>Do you really think chat GPT is revolutionary? Do you

0:01:53.000 --> 0:01:56.480
<v Speaker 1>think any of this is revolutionary? We are two years

0:01:56.520 --> 0:01:59.040
<v Speaker 1>in and I'm still getting dms from people asking me

0:01:59.200 --> 0:02:01.680
<v Speaker 1>what would it take to make you believe that this

0:02:01.720 --> 0:02:04.400
<v Speaker 1>is all the future? And I'm so fucking tired of

0:02:04.400 --> 0:02:08.040
<v Speaker 1>being asked this. The arguments I make are grounded in

0:02:08.160 --> 0:02:11.280
<v Speaker 1>numbers and things that have happened, not just financial details

0:02:11.320 --> 0:02:14.840
<v Speaker 1>and statistics, but in objective evaluations the products in question.

0:02:15.160 --> 0:02:18.919
<v Speaker 1>There arefficacy at tasks and the people involved. Yet somehow,

0:02:19.000 --> 0:02:22.720
<v Speaker 1>I and other critics are continually made to justify themselves.

0:02:22.800 --> 0:02:25.560
<v Speaker 1>While Sam Altman of open ai and Dario Amadee of

0:02:25.600 --> 0:02:28.959
<v Speaker 1>Anthropic vaguely suggests that we'll have a conscious autonomous computer

0:02:29.000 --> 0:02:32.399
<v Speaker 1>by the year twenty twenty seven. When I ask how

0:02:32.440 --> 0:02:35.000
<v Speaker 1>open ai survives as it spends nine billion dollars to

0:02:35.040 --> 0:02:37.959
<v Speaker 1>lose five billion dollars, I'm obliquely threatened by Casey Newton

0:02:38.000 --> 0:02:41.560
<v Speaker 1>of Platformer and Hardfork that he's taking detailed notes about

0:02:41.560 --> 0:02:44.040
<v Speaker 1>anyone who believes that open ai might go bankrupt or

0:02:44.120 --> 0:02:47.080
<v Speaker 1>run out of money. When as Recline suggests that AGI

0:02:47.200 --> 0:02:49.280
<v Speaker 1>is about to arrive in a conversation with some sort

0:02:49.280 --> 0:02:52.679
<v Speaker 1>of former Biden administration aicon artist, I'm sent the link

0:02:52.760 --> 0:02:55.600
<v Speaker 1>thirty times with people saying does this mean you're wrong?

0:02:56.400 --> 0:02:58.840
<v Speaker 1>I realize I'm complaining, but I'm justified in doing so.

0:02:59.000 --> 0:03:00.920
<v Speaker 1>Why the fuck do I other critics have to make

0:03:01.000 --> 0:03:05.240
<v Speaker 1>rigorously founded in persuasive arguments while AI companies spout fantastical nonsense.

0:03:05.720 --> 0:03:08.000
<v Speaker 1>Why does Sam Mortman get headlines when he posts about

0:03:08.240 --> 0:03:10.359
<v Speaker 1>and this did just happen? By the way, making an

0:03:10.400 --> 0:03:14.120
<v Speaker 1>AI that can do creative writing, and I wish it

0:03:14.160 --> 0:03:17.480
<v Speaker 1>was just ignorance. People like casing you and a Nezrakline

0:03:17.480 --> 0:03:19.720
<v Speaker 1>aren't stupid, but they're also fully willing to back the

0:03:19.800 --> 0:03:22.040
<v Speaker 1>narratives of powerful people that they want to be friends with.

0:03:22.360 --> 0:03:24.040
<v Speaker 1>They want the rich and powerful to win, and they

0:03:24.080 --> 0:03:26.280
<v Speaker 1>want to be the people that write their narratives and

0:03:26.280 --> 0:03:29.920
<v Speaker 1>get their interviews. And yeah, I'm being petty. These are

0:03:29.960 --> 0:03:32.640
<v Speaker 1>people that ostensibly compete with my work. But people with

0:03:32.720 --> 0:03:36.160
<v Speaker 1>such a large audience have a responsibility two said audience

0:03:36.800 --> 0:03:40.360
<v Speaker 1>not what they wish would come true. And really, I've

0:03:40.360 --> 0:03:44.240
<v Speaker 1>got to ask, how does all of this end? Right now,

0:03:44.240 --> 0:03:47.040
<v Speaker 1>we've got Anthropic, a company allegedly makes one hundred and

0:03:47.080 --> 0:03:49.800
<v Speaker 1>fifty million dollars a month according to the Information, but

0:03:49.880 --> 0:03:53.240
<v Speaker 1>loses over five billion dollars a year or so, reported

0:03:53.280 --> 0:03:55.840
<v Speaker 1>by the Information, And they make a commoditized product one

0:03:55.920 --> 0:03:58.600
<v Speaker 1>very similar to open Aiyes, a company that will also

0:03:58.800 --> 0:04:01.240
<v Speaker 1>likely lose a shit ton of money eleven billion dollars

0:04:01.320 --> 0:04:04.920
<v Speaker 1>or more in twenty twenty five. These companies are dependent

0:04:04.960 --> 0:04:07.560
<v Speaker 1>on receiving billions or tens of billions of dollars a

0:04:07.600 --> 0:04:10.520
<v Speaker 1>year in funding for an indeterminately long period of time.

0:04:10.640 --> 0:04:14.720
<v Speaker 1>For an equally indeterminate goal. I'm being completely objective here.

0:04:15.120 --> 0:04:17.799
<v Speaker 1>There's nothing that these companies have made that suggests anything

0:04:17.839 --> 0:04:21.080
<v Speaker 1>will change. Every new version of claud sonnet or GPT

0:04:21.240 --> 0:04:23.880
<v Speaker 1>is iterative, and the products we see today are alarmingly

0:04:23.920 --> 0:04:26.000
<v Speaker 1>similar to the ones we saw in the last two years.

0:04:26.520 --> 0:04:29.840
<v Speaker 1>Despite everybody talking about agents, the actual agents that exist

0:04:29.839 --> 0:04:32.240
<v Speaker 1>don't really work, and those that are able to kind

0:04:32.240 --> 0:04:34.440
<v Speaker 1>of complete a task costs thousands of dollars and again

0:04:35.160 --> 0:04:40.440
<v Speaker 1>don't always work. This industry is unprofitable, unsustainable, and does

0:04:40.480 --> 0:04:42.120
<v Speaker 1>not appear to be able to create a product that

0:04:42.160 --> 0:04:44.440
<v Speaker 1>people want to pay for, let alone one that they

0:04:44.480 --> 0:04:47.039
<v Speaker 1>pay enough to put the company making it in the green.

0:04:47.320 --> 0:04:49.640
<v Speaker 1>We're two years in. How do we not have one

0:04:49.760 --> 0:04:54.040
<v Speaker 1>profitable generate AVII company other than Boy is it churing?

0:04:54.080 --> 0:04:57.920
<v Speaker 1>They're a consultancy? It does not count. I do want

0:04:57.960 --> 0:05:02.800
<v Speaker 1>to say, though I'm not cheering the apocalypse, what I've

0:05:02.839 --> 0:05:05.120
<v Speaker 1>been describing for the last year is a group delusion

0:05:05.120 --> 0:05:07.360
<v Speaker 1>where hundreds of billions of dollars got funneled into an

0:05:07.400 --> 0:05:10.760
<v Speaker 1>environmentally and financially destructive distraction from the real problems that

0:05:10.839 --> 0:05:14.560
<v Speaker 1>humanity faces. The longer this goes on means that it

0:05:14.600 --> 0:05:16.560
<v Speaker 1>will be worse for the tech industry because once this

0:05:16.600 --> 0:05:20.160
<v Speaker 1>bubble bursts, it will puncture everything, tens of thousands of

0:05:20.160 --> 0:05:23.279
<v Speaker 1>people laid off, brutal damage under tech valuations, and likely

0:05:23.279 --> 0:05:26.080
<v Speaker 1>a glut of tech talent that depresses wages across the valley.

0:05:27.200 --> 0:05:29.040
<v Speaker 1>What's important to know is that so much of this

0:05:29.040 --> 0:05:32.280
<v Speaker 1>could have been avoided. Microsoft could have chosen not to

0:05:32.400 --> 0:05:35.440
<v Speaker 1>continue sustaining open AI, as could Google and Amazon have

0:05:35.480 --> 0:05:38.160
<v Speaker 1>refused to back anthropic or just not do this nonsense

0:05:38.800 --> 0:05:41.719
<v Speaker 1>so called reporters that Casey Newton and Ezra Klein could

0:05:41.720 --> 0:05:44.760
<v Speaker 1>have made these companies justify themselves rather than operating as

0:05:44.760 --> 0:05:48.039
<v Speaker 1>so called cautious optimists that end up mostly just parroting

0:05:48.040 --> 0:05:51.760
<v Speaker 1>marketing materials. And the larger media could have covered generative

0:05:51.760 --> 0:05:53.800
<v Speaker 1>AI based on what it does, rather than what they're

0:05:53.800 --> 0:05:55.920
<v Speaker 1>told it might do by somebody who has the financial

0:05:55.920 --> 0:06:00.799
<v Speaker 1>incentive to lie. In any case, when this collapses, my words,

0:06:01.440 --> 0:06:04.520
<v Speaker 1>I have been taking very, very detailed notes. I've been

0:06:04.560 --> 0:06:07.640
<v Speaker 1>watching those who have sustained this bullshit narrative and other

0:06:07.680 --> 0:06:12.240
<v Speaker 1>bullshit narratives and cryptocurrency in the metaverse. People willfully misleading

0:06:12.240 --> 0:06:14.640
<v Speaker 1>the public in the process, and when the time is right,

0:06:14.720 --> 0:06:18.040
<v Speaker 1>I will coldly and clinically read you every single time

0:06:18.080 --> 0:06:23.320
<v Speaker 1>they've done so. Anyway, enjoy tomorrow's episode