← 返回任务列表

How the Midterm Results Change AI

225 段 · 1 位说话人 · 原片 24:26
M1
M10:00

中期选举结果已经出炉,我们迎来了所谓的蓝色浪潮、蓝色海啸,随便你用什么比喻,总之之前承诺的来了。

The midterm election results are in and we got the blue wave, blue tsunami, pick your metaphor that had been promised.

问题在于,对我们《AI日报》来说,这对AI意味着什么?

And the question is, for our purposes here at the AI Daily Brief, what's it going to mean for AI?

是不是意味着数据中心建设要戛然而止了?

Does it mean that the data center buildout comes to a screeching halt?

是不是意味着我们要面临严厉的新法规?

Does it mean that we're in for draconian new regulations?

这对2028年总统大选中的AI议题又预示着什么呢?

What does it say about AI in the 2028 presidential election?

这些就是今天《AI日报》要探讨的问题。

Those are the questions for today's episode of the AI Daily Brief.

《AI日报》是一个每日更新的播客和视频节目,专注于AI领域最重要的新闻和讨论。

The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.

好了,朋友们,在深入之前,先快速说几件事。

All right, friends, quick announcements before we dive in.
M1
M11:22

好了,朋友们,这次民意调查机构没搞错,民主党人的乐观预测也得到了证实。

Well, friends, this time, the pollsters were not wrong, and the hopeful prognostications of Democrats have been affirmed.

美国昨天确实迎来了一波蓝色浪潮,美国公民给了特朗普的政党一记响亮的耳光,把胜利交给了民主党。

America got a genuine blue wave yesterday, as the American citizens handed a resounding defeat to the party of Trump and victory to the Democrats.

截至昨晚深夜,民主党在全国平均支持率上领先了11个百分点。

As of late last night, Democrats had an 11-point swing across national averages.

他们将掌控众议院和参议院,虽然不太可能获得足以推翻否决权的绝对多数。

They will take control of the House and the Senate, although they are unlikely to gain a veto-proof supermajority.

从历史对比来看,就绝对席位数而言,这次结果似乎没有达到像2018年特朗普中期选举或2010年奥巴马中期选举那样历史性惨败的规模,但民主党正在翻转一些共和党的大本营,比如得克萨斯、南卡罗来纳和肯塔基这些原本看似不太可能赢的席位。

By way of historical comparisons, in terms of absolute number of seats, the result doesn't look to be quite at the same magnitude as historic wipeouts like the 2018 Trump or 2010 Obama midterms, but Democrats are flipping unlikely seats in GOP strongholds like Texas, South Carolina, and Kentucky.

当然,我们关心的问题是,这对AI意味着什么?

The question for our purposes is, of course, what is this going to mean for AI?

投票前,人们很容易说AI是 ballots 上的议题,但更准确的说法其实是,AI数据中心才是 ballots 上的议题。

Heading into the vote, it would've been tempting to claim that AI was on the ballot, but of course, the more accurate framing was that AI data centers were on the ballot.

根据皮尤研究中心十月初进行的一项民调,只有44%的选民表示AI对他们来说是非常重要的问题。

According to a Pew Research poll conducted in early October, only 44% of voters said that AI was a very important issue to them.

相比之下,比如外交政策是57%,移民问题是59%,教育是62%,医疗保健是68%,民主的未来是73%,经济则是84%。

That compares to, for example, 57% for foreign policy, 59% for immigration, 62% for education, 68% for healthcare, 73% for the future of democracy, and 84% for the economy.

至于AI对谁重要,这个问题稍微偏向民主党,54%的人认为AI是非常重要的议题,而共和党只有34%。

When it came to who AI was important to, the issue skewed slightly towards Democrats, with 54% saying that AI was a very important issue compared to 34% of Republicans.

皮尤甚至指出,把它称为一个议题其实有点误导。

Pew even highlighted that labeling it as an issue was a little misleading.

真正的问题当然是数据中心,61%的选民表示他们不喜欢或非常不喜欢支持在当地建设数据中心的候选人。

The actual issue was, of course, data centers, with 61% of voters saying they dislike or strongly dislike candidates who support building data centers locally.

这同样有党派分歧,75%的民主党支持者表示不喜欢支持数据中心的候选人,而共和党支持者中这一比例是44%。

Again, this skewed along partisan lines, with 75% of Democrat supporters saying they dislike pro-data center candidates compared to 44% of Republican supporters.

双方几乎没有人说喜欢支持数据中心的候选人。

Hardly anyone on either side said they liked pro-data center candidates.

只有5%的民主党支持者和14%的共和党支持者表示喜欢。

With just 5% of Democrat supporters and only 14% of Republican supporters.

对于过去三到六个月里稍微关注过的人来说,这点都不意外——支持数据中心对两党来说都是巨大的选票毒药。

To the shock of no one who has been paying any amount of attention for the last 3 to 6 months, being in favor of data centers was a huge vote loser for both parties.

至于数据中心为什么这么不受欢迎,详细分析超出了本期节目的范围。

Now, when it comes to why data centers are so unpopular, the full accounting for that is beyond the scope of this particular show.
M1
M13:51

对我们节目来说值得注意的是,数据中心政治——仅从竞选活动中的讨论方式来看——大多与经济方面的其他关键议题挂钩,比如生活成本,以及普通人和常被称为寡头的科技富豪群体之间的财富差距。

What's notable for our purposes on this show is that data center politics were, to look only at the way that they were discussed on the campaign trail, mostly tied to other key issues around the economy, like cost of living and the wealth divide between regular people and the group of tech money who are often referred to as the oligarchy.

俄亥俄州民主党参议员候选人Sherrod Brown说,俄亥俄人不应该被迫补贴这些成本,而且应该由俄亥俄人,而不是Jon Husted和他的亿万富翁朋友们,来决定是否在他们的社区建设数据中心。

Ohio Democrat Senate candidate Sherrod Brown said, Ohioans shouldn't be forced to subsidize these costs, and Ohioans, not Jon Husted and his billionaire friends, should get to decide if data centers are built in their community.

在密歇根州的参议员竞选中,Abdul El-Sayed投放了攻击性广告,将他对手Mike Rogers在2001年支持《爱国者法案》的行为与 flock 摄像头和AI数据中心联系起来。

In the Michigan Senate race, Abdul El-Sayed ran attack ads that tied flock cameras and AI data centers to his opponent Mike Rogers' support of the Patriot Act in 2001.

换句话说,反对数据中心已经和更广泛的选举主题完全交织在一起了。

In other words, data center opposition was fully enmeshed and mixed into larger electoral themes.

考虑到特朗普总统在最后阶段明确支持AI产业,共和党人在处理数据中心问题上有点受约束。

Republicans were a bit constrained on how they addressed the data center issue, given President Trump's vocal support for the AI industry down the stretch.

你可能还记得,八月底他在Truth Social上宣称,那些拒绝数据中心的社会最终会变得,引用,'落后且贫穷'。

In late August, you might remember, he pronounced on Truth Social that communities who reject data centers will end up being, quote, backwards and poor.

对于处在激烈竞选中的共和党人来说,那本是一个与总统在此问题上分道扬镳的绝佳机会,但很少有人抓住那个机会。

That was the perfect opportunity for Republicans in tight races to split with the president on the issue, but few took that chance.

相反,共和党人大体上维持现状,俄亥俄州副州长Jon Husted表示,数据中心是并且应该继续由地方决定。

Instead, Republicans largely offered the status quo, with Ohio Lieutenant Governor Jon Husted saying that data centers are and should remain a local decision.

NBC News的出口民调证实,数据中心是一个有意义的问题,但并非最重要的问题之一。

Exit polling from NBC News confirmed that data centers were a meaningful issue, but not among the most significant.

56%的选民表示,数据中心是他们投票决定中的关键或主要因素,其中民主党人引用数据中心为主要因素的可能性是共和党人的两倍。

56% of voters said that data centers were a crucial or major factor in determining their vote, with Democrats twice as likely to cite data centers as a major factor compared to Republicans.

但相比之下,73%的人表示移民执法政策是一个关键或主要因素。

But by comparison, 73% said immigration enforcement policies were a crucial or major factor.

74%的人对乌克兰战争给出了同样的回答,而89%的人表示生活成本对他们来说是一个主要因素。

74% gave that response about the war in Ukraine, with 89% saying cost of living was a major factor for them.

政府腐败是民调中最高的问题,90%的选民表示这是他们投票中的主要因素。

Government corruption was the highest polling issue, with 90% of voters saying that it was a major factor in their vote.

换句话说,数据中心重要到足以被列入出口民调,但在热点问题中,数据中心排在最后一位。

In other words, data centers were important enough to be included in exit polls, but among hot-button issues, data centers were in last place.

不过,仍有少数关键竞选因为人工智能扮演了核心角色。

Still, there are a few key races where AI played a central role.

首先是加利福尼亚第11选区,涵盖旧金山,州参议员Scott Wiener对阵旧金山市监事会成员Connie Chan。

The first is California's 11th District covering San Francisco, where State Senator Scott Wiener was going up against San Francisco Board of Supervisors member Connie Chan.

这是Nancy Pelosi自1987年以来代表的选区,并且包含了OpenAI和Anthropic的总部。

This was the district that Nancy Pelosi represented since 1987 and contains the headquarters for both OpenAI and Anthropic.

虽然两位候选人都是进步民主党人,但Connie Chan获得了Pelosi的支持以及党内机构的支持。

While both candidates are progressive Democrats, Connie Chan was the one that secured Pelosi's endorsement and support of the party apparatus.

截至我录制时,这场竞选尚未分出胜负,但Wiener已经建立了显著领先,大约一半选票已点算。

The race hadn't been called as of the time I was recording, but Wiener has secured a commanding lead with around half the votes tallied.

现在,你可能还记得Wiener在过去几年加州立法机构的人工智能政策斗争中的表现。

Now, you'll likely remember Wiener from the past few years of AI policy battles in the California State Legislature.

他是SB 1047法案的主要发起人,这是一项2024年提出的有争议的人工智能监管法案,其中包括强制性的紧急关闭开关、人工智能开发者的责任,以及对前沿人工智能的定义,涵盖了所有大于Llama 3的模型。

He was the lead sponsor for SB 1047, the controversial AI regulatory bill proposed in 2024, which included, among other things, a mandatory kill switch, liability for AI developers, and a definition of frontier AI that included all models larger than Llama 3.

该法案在州议会通过,但被州长Gavin Newsom否决。

The bill passed the Assembly but was vetoed by Governor Gavin Newsom.

尽管如此,该法案的许多理念还是被纳入后续的加州立法,并成为了其他州人工智能法规的蓝图。

Still, many of the ideas from the bill found their way into subsequent California legislation, which became the blueprint for AI regulations in other states.

对人工智能采取强硬态度是Wiener竞选活动的基石。

Being tough on AI was a cornerstone of Wiener's campaign.

例如,在十月份的一次辩论中,他夸耀自己,引用,'与科技巨头开战',以通过加州的人工智能法规。

During an October debate, for example, he boasted that he, quote, went to war with Big Tech to pass AI regulation in California.

Chan批评Wiener从Public First Action(一个与Anthropic结盟的支持监管的政治行动委员会)那里接受了300万美元。

Chan criticized Wiener for taking $3 million from Public First Action, an Anthropic-aligned pro-regulation PAC.

但很难把Wiener描绘成处于人工智能产业的控制之下。

But it's pretty difficult to frame Wiener as being in the pocket of the AI industry.

事实上,如果要说的话,他反而有望成为关注存在风险的人工智能安全倡导者的拥护者。

In fact, if anything, he's positioned to become a champion for AI safety advocates focused on existential risk.

人工智能政策的另一个真正战场是在俄亥俄州。

Another real battleground race for AI policy was in Ohio.

俄亥俄州大约有200个数据中心,尽管其中大部分规模较小。

Ohio has around 200 data centers, although most of them are small.

然而,还有46个新项目计划,总运营容量达16吉瓦。

However, 46 new projects are planned that will operate at a combined 16 gigawatts.

其中最主要的是Meta位于哥伦布市郊的Prometheus设施,它将成为世界上第一个运营容量达一吉瓦的数据中心。

Chief among those projects are Meta's Prometheus facility outside of Columbus, which is set to be the first data center in the world to operate at a full gigawatt of capacity.

而OpenAI计划在派克县建设一个园区,其规划容量为10吉瓦。

And OpenAI has planned a campus in Pike County, which is planned for 10 gigawatts.

民主党挑战者Sherrod Brown将数据中心作为其竞选的核心主题,旨在击败Jon Husted。

Democrat challenger Sherrod Brown made data centers the central theme of his campaign to unseat Jon Husted.

Brown在整个竞选中猛烈抨击Husted支持数据中心的立场,一则攻击性广告称他为数据中心的代言人。

Brown hammered Husted on his pro-data center position throughout the campaign, with one attack ad labeling him the face of data centers.

另一则广告只是数据中心轰鸣的噪音,威胁选民说,如果数据中心建在他们的社区,他们会面对什么。

Another ad was nothing but the loud hum of a data center, threatening to voters what they could expect if a data center goes up in their neighborhood.

这场竞选在周三凌晨就宣布了结果,布朗重新夺回了他在参议院的席位,并增加了一个蓝色席位。

The race was called early on Wednesday morning, with Brown reclaiming his spot in the Senate and adding another blue seat.

现在,值得注意的是,对AI行业的影响在于,布朗返回华盛顿时不是作为新人参议员,而是作为资深领袖。

Now, notably for the implications for the AI industry, Brown is returning to Washington not as a freshman senator, but as a veteran leader.

他在2024年大选被赶下台之前已在参议院任职18年,部分原因是一个加密政治行动委员会(crypto PAC)提供的资金浪潮反对他。

He had 18 years in the Senate before being unseated in the 2024 election, thanks in part to a wave of money against him contributed by a crypto PAC.

虽然布朗不像民主党内的其他一些人那样是强硬进步派,但他一直对大型科技公司持严厉批评态度,几乎涵盖其所有形式。

And while Brown isn't as much of a hardline progressive as some others in the Democrat Party, he has been a consistently harsh critic of big tech in basically all its incarnations.

他重返参议院,带着明确的使命要阻止本州的数据中心,并且对科技行业重新燃起怒火,因为该行业花费了数千万美元来阻止他当选。

He's returning to the Senate with a clear mandate to block data centers in his state and a renewed fury against the tech industry after they spent tens of millions to keep him out of office.

另一个AI可能发挥了巨大作用的地方是德克萨斯州。

The other place where AI might have played an outsized role was in Texas.

德克萨斯州转向蓝色(民主党)显然是很多人心目中本次选举最重要的消息。

Texas flipping blue is clearly many people's candidate for the most important story of this election.

James Talarico在参议院击败了Ken Paxton,成为自1993年以来第一位代表德克萨斯州的民主党人。

James Talarico defeated Ken Paxton in the Senate, becoming the first Democrat to represent Texas since 1993.

对AI行业来说,影响更大的投票是州长竞选,民主党人Gina Hinojosa击败了连任三届的现任州长Greg Abbott。

For the AI industry, the even more impactful vote was in the gubernatorial race, where Democrat Gina Hinojosa defeated 3-term incumbent Greg Abbott.

现在,这场竞选涉及的因素远不止AI。

Now, there was way more going on in this race than AI.

有人口结构变化,围绕这些人口结构的政治重组,但这场竞选中一个重要的叙事是,Abbott州长在一系列不同问题上变得自满。

You have shifting demographics, political realignment around those demographics, but certainly a big part of the narrative coming out of the race was that Governor Abbott got complacent on a range of different issues.

德克萨斯州从一开始就是AI建设的关键地点,许多大型项目涌向西德克萨斯的廉价天然气和该州快速发展的廉价可再生能源。

Texas has been ground zero for the AI buildout since the beginning, with many large projects flocking to the cheap gas in West Texas and the state's rapid buildout of cheap renewable energy.

然而在九月份,随着政治风向的转变,Abbott州长暂停了所有数据中心的新许可,直到德克萨斯能源和环境监管机构完成对新提案的全面审查。

In September, as the political winds shifted, however, Governor Abbott halted all new permitting for data centers until the Texas energy and environmental regulators completed a full audit of new proposals.

当时,该州有超过470吉瓦的新数据中心提案,新容量大约是当前电网规模的三倍。

At that time, the state had more than 470 gigawatts of new data center proposals, around 3 times the size of the current grid in new capacity.

唉,Abbott的行动太微不足道,太迟了,结果Hinojosa结束了德克萨斯州长达30年的共和党控制。

Alas, Abbott's move was too little, too late, with Hinojosa ending a 30-year streak of Republican control in Texas.

确实,虽然民主党在华盛顿获胜是一个重大变化,但从非常实际和短期的角度来看,对AI建设影响更大的可能是州和地方竞选。

And indeed, while Democrats winning in Washington is a big change, from a very practical and immediate-term standpoint, it's the state and local races that could be even more important when it comes to the AI buildout.

特别是数据中心政策由各州决定,州长有最终发言权。

Data center policy in particular is determined by the states, with governors having the final say.

经过昨晚的结果,10个最大的州中有9个现在由民主党执政。

After last night's results, 9 of the 10 largest states are now run by Democrats.

俄亥俄州和佐治亚州也从红转蓝,佛罗里达州现在是前10大州中唯一一个由共和党州长执掌的州。

Ohio and Georgia also flipped red to blue, with Florida now being the only top 10 state with a Republican governor.

从经济角度来看,德克萨斯州的GDP接近3万亿美元,比加拿大或俄罗斯都要大。

And to put things in economic terms, with almost $3 trillion in GDP, Texas is larger than Canada or Russia.

正如Breaking Points的Saagar Enjeti所说,世界第八大经济体正在经历政权更迭。

As Saagar Enjeti of Breaking Points put it, the 8th largest economy in the world is experiencing regime change.

即将上任的Hinojosa政府预计将利用政策来减缓数据中心的建设。

And the incoming Hinojosa administration is expected to use policy to slow down the data center buildout.

她的竞选纲领包括在制定法规之前暂停建设,结束税收补贴,并确保电力水资源不会从当地社区转移。

Her platform included a moratorium until regulations could be written, ending tax subsidies, and ensuring that electricity and water resources aren't diverted from local communities.

像许多民主党人一样,她将数据中心与更广泛的经济问题联系起来,称Abbott的税收补贴是一种,引述,“对公民的腐败税”。

Like many Democrats, she tied data centers to broader economic issues, calling Abbott's tax subsidies a, quote, corruption tax on citizens.

不过,我们稍后会回到一个重要言论,那就是在8月的一次新闻发布会上,Hinojosa告诉选民,我不讨厌数据中心。

Now, one important statement, though, that we're going to come back to in a little bit is when, at a press conference in August, Hinojosa told voters, I don't hate data centers.

我只是讨厌他们利用德克萨斯人的方式。

I just hate the way they're taking advantage of Texans.

现在,除了AI情绪确实产生影响的竞选之外,还有几十场其他竞选,其中AI至少是竞选的核心主题。

Now, aside from the races where AI sentiment actually moved the needle, there were dozens of other races where AI was at least a central theme of the campaign.

有两场竞选尤为突出,特别是考虑到人们的注意力现在将转向2028年总统竞选。

Two stand out as especially important, particularly in light of the way that the attention will now shift to the 2028 presidential campaign.

第一场是民主党参议员Jon Ossoff,他轻松击败Mike Rogers,在佐治亚州轻松赢得第二个任期。

The first was Democrat Senator Jon Ossoff, who cruised to a second term in Georgia after easily dispatching Mike Rogers.

这场竞选在初期看起来有点接近,但到后期,支持特朗普的Rogers的支持率崩溃,使得这场比赛成为当晚最早被宣布结果的选举之一。

The race looked a little close in the early days, but support for a Trump-backed Rogers collapsed down the stretch to make this one of the first races called for the evening.

Ossoff是最有效地将数据中心用作大型科技寡头、政治腐败和生活成本问题象征的民主党人之一。

Ossoff was one of the Democrats that most effectively used data centers as the symbol for big tech oligarchy, political corruption, and cost of living issues.

在辩论中,Ossoff说,佐治亚州的家庭不应该多花一分钱来补贴这些数据中心的能源消耗。

During the debates, Ossoff said, no Georgia family should pay a dime more to subsidize the energy consumption of these data centers.

他然后声称,特朗普的宏伟法案,引用一下,‘将你们700亿美元的钱给了数据中心开发者。’

He then claimed that Trump's big, beautiful bill, quote, gave $70 billion of your money to data center developers.

Ossoff将这笔钱与医疗成本上升和医院关闭联系起来。

Ossoff tied that money to rising healthcare costs and the shutdown of hospitals.

虽然这句话背后显然有更多的背景,但这一信息显然引起了选民的共鸣。

And while there is obviously more context behind that statement, the message clearly resonated with voters.

从今年年初开始,Ossoff就一直在这个角度上努力,于4月发起了一项关于数据中心电力使用的调查。

From very early in the year, Ossoff has been pursuing this angle, launching an investigation into data center electricity use in April.

现在,这个阶段,Ossoff实际上在AI政策方面做得相对较少,主要是支持一些能成为好头条但没什么实质的调查。

Now, at this stage, Ossoff has actually done relatively little on AI policy, largely championing toothless investigations that make for good headlines.

然而,鉴于他将此作为竞选的主要叙事支柱,而且围绕他作为2028年总统有力候选人的讨论越来越多,看到这种情况改变也不足为奇。

However, given that he made it a major narrative pillar of his campaign, and given the rising conversation around him as a strong candidate for president in 2028, it wouldn't be surprising to see that change.

首先是民主党参议员Jon Ossoff,他轻松击败Mike Collins,在佐治亚州顺利赢得第二个任期。

The first was Democrat Senator Jon Ossoff, who cruised to a second term in Georgia after easily dispatching Mike Collins.

这场竞选在初期看起来接近,但对支持特朗普的Collins的支持在最后阶段崩溃,使其成为当晚最早被宣布的选举之一。

The race looked close in the early days, but support for a Trump-backed Collins collapsed down the stretch to make this one of the first races called for the evening.

另一位作为更明确的AI反对者脱颖而出的候选人是Abdul El-Sayed,他击败了他的对手Mike Rogers,赢得了密歇根州的参议院席位。

The other candidate who stood out as a more distinct opponent of AI was Abdul El-Sayed, who dispatched his own Mike, Mike Rogers, to win the Michigan Senate seat.

如果说Ossoff是华盛顿中间派民主党人的新面孔,那么El-Sayed则代表了社会主义左翼的崛起。

Now, if Ossoff is the new face of centrist Democrats in Washington, El-Sayed represents the rise of the socialist left.

他是美国历史上第一位穆斯林参议员,并且他竞选活动的许多策略借鉴了纽约的Mamdani的竞选策略。

He's the first Muslim senator in US history and ran a campaign that borrowed heavily from the Mamdani playbook in New York.

El-Sayed的竞选团队在利用数据中心作为国家压迫的象征方面特别有效。

El-Sayed's campaign was particularly effective at using data centers as a symbol of state oppression.

他的竞选团队将数据中心与群集摄像头以及在中东使用AI驱动的战争混为一谈,描绘了一幅反乌托邦的景象。

His campaign conflated data centers with flock cameras and the use of AI-powered warfare in the Middle East to paint a dystopian picture.

在最近周末对《60 Minutes》的采访中,当他被问及他认为对密歇根州人来说最重要的问题是什么时,他回答说:‘如果你在五年前问我,我会说是医疗保健。’

In an interview with 60 Minutes from this past weekend, when he was asked what he saw as the most important issue for Michiganders, he responded, Had you asked me 5 years ago, I would have said healthcare.

它仍然是人们心头的大事。

It still is really high on people's mind.

我想现在我们正冲向一个世界,在那里我每天听到的最重要的问题是AI。

I think right now, we are barreling toward a world where the most important issue I hear about every day is AI.

El-Sayed提到了失业、经济破坏,甚至失去控制的风险,或者看到AI被用来制造生物武器。

El-Sayed mentioned job loss, economic disruption, and even the risk of losing control or seeing AI used to create a bioweapon.

他评论道:‘所有这些结果都真的、真的很可怕。’

He commented, all of these outcomes are really, really quite scary.

我从各种不同的方式听到这种担忧。

I hear it in so many different ways.

这些是人们极度害怕的问题,而且以各种方式表现出来。

These are issues that people are deathly afraid of, and it shows up in just all kinds of ways.

现在,谈到他将支持什么政策,El-Sayed作为一个与民主党建制派没有特别联系的新手参议员,更像是一张未知数。

Now, when it comes to what policies he's going to support, El-Sayed, as a freshman senator with no particular connection to the Democratic establishment, is a bit more of a wild card.
M1
M117:04

那么,这场选举在可能的政策变化方面给行业留下了什么?

So where does this election leave the industry in terms of likely policy changes?

当然,在那些数据中心暂停的各种形式实际上被列入选票的地方,对于数据中心开发者来说会有一些新的障碍。

Certainly where data center moratoria of various forms were literally on the ballot, there are going to be some new hurdles for the data center developers.

全国有41个与数据中心相关的下位提案,截至昨晚,看起来所有41个都将通过,这发出了一个非常明确的信号:AI公司已经在数据中心建设的叙事战中完全输了。

There were 41 down-ballot proposals related to data centers across the country, and as of last night, it looks like all 41 will pass, sending a very clear signal that AI companies have completely lost the narrative war around the data center buildout.

美国人负责任创新协会的联合创始人Brad Carson表示,我们刚刚目睹了一场AI地震。

Brad Carson, the co-founder of Americans for Responsible Innovation, said, what we just witnessed was an AI earthquake.

选民明确表示,他们希望制定有关AI和数据中心的规则。

Voters sent a clear message that they want rules for AI and data centers.

尽管如此,在华盛顿的重大政策变化方面,我认为我们不太可能看到某种全面的监管框架。

Still, when it comes to major policy change in Washington, I think it's unlikely that we see some comprehensive regulatory framework.

当然,其中一个原因是,在选举前不久接受 Fox News 采访时,总统 Trump 表示,如果 Democrats 获胜,引用,“我猜我会变成 Mr.”

One reason for that, of course, is that in an interview with Fox News shortly before the election, President Trump said that if the Democrats win, quote, I guess I'll become Mr.

否决。

Veto.

鉴于此,我们也不太可能看到联邦层面自上而下地完全暂停数据中心建设或 frontier model 训练。

Given that, we're also not likely to see a full top-down federal moratorium on data center construction or frontier model training.

但这并不意味着我们不会看到任何变化。

But that doesn't mean that we won't see any changes.

在我看来可能发生的那些变化中,有很多是容易实现的目标,实际上在这些问题上存在相当多的常识性共识,分歧比最初看起来要少。

Among those that I think are likely are a lot of the low-hanging fruit where there's actually a fair bit of common sense consensus and less disagreement than it might at first seem.

一个很好的例子就是实验室员工的举报人保护。

A great example of that is whistleblower protection for lab employees.

在我看来,这个问题会跨越党派界限而受欢迎。

That strikes me as the type of issue that's going to be popular across party lines.

我认为我们会看到将其法典化的尝试。

I think we'll see attempts to codify.

行政部门已经开始在发布 frontier LLMs 之前对其进行自愿测试。

The executive branch has taken to voluntary testing of frontier LLMs before they get released to the public.

我当然会期待看到关于事件报告的新规则,虽然细节决定成败,但我认为总体而言,这些规则将再次跨越党派界限受到欢迎。

I would certainly expect to see new rules around incident reporting, which, while the devil is in the details, I think in general will be popular again across party lines.

我认为我们会看到为模型构建者制定更严格的法律责任。

I think we'll see stronger liabilities written into law for the model builders.

我们可能会看到围绕实际 data sovereignty 规则的一些势头。

We might see some momentum around actual data sovereignty rules.

一个大的问题领域将是 export controls。

One of the big question areas will be around export controls.

在这方面,我认为很难确切判断会发生什么。

And there, I think it's a little bit harder to tell exactly what will come.

export control 制度以及如何思考 China 问题,并没有完全按照党派界限明确划分。

Export control regimes and how to think about China has not exactly been clear-cut along party lines.

所以我不认为存在简单的推断,说因为 Democrats 赢了,我们就会得到更多或更少的 export controls。

And so I don't think there's an easy extrapolation to say because Democrats won, we're going to get more or less export controls.

但我确实认为这将是一个辩论的领域。

But I do think that will be an area of debate.

而且我认为,这场辩论当然会与议程上的议题相关联,至少从讨论的角度来看,这些议题包括暂停、节奏,甚至潜在禁止某些训练技术,比如通过 recursive self-improvement 让 AI 构建 AI。

And I think that that debate will, of course, be tied in with what will be on the agenda, at least from a discussion perspective, which is questions of pausing, pacing, and even potential prohibitions on certain training techniques like AI building AI through recursive self-improvement.

我们当然会看到大量 AI 高管被召集到 Congress 参加多次听证会,但除了 Democrats 可能决定追究 Elon 之外,我实际上不认为这些听证会会产生特别大的影响。

We're certainly going to see tons of AI executives hauled before Congress for multiple hearings, but outside of the potential of Democrats deciding to go after Elon, I don't actually think that any of those hearings will be particularly impactful.

总的来说,我认为这次选举不可避免地会转向更接近 Democrat 或左派的立场,即希望对 AI 制定更多规则。

In general, I think it was inevitable that we were going to see a shift with this election closer to the Democrat or left position of wanting more rules around AI.

坦白说,我认为即使 Republicans 出人意料地表现更好,这种情况也绝不可能不发生。

And frankly, I don't think there was any chance, even if Republicans had surprisingly outperformed, that that wouldn't have been the case.

其原因是,虽然这在一定程度上是政治风向的转变,国家拒绝了 Trump 过去几年,并想要一些不同的东西。

The reason for that is that while this is in part a shift in the political winds, with the country rejecting the last couple of years of Trump and wanting something different.

我还认为,这部分只是基于模型变得更强大的现实。

I also think that this is in part just a reality based on models getting more powerful.

换句话说,更强大的 AI 意味着我们应该对 AI 有更多规则。

In other words, more powerful AI means that we should have more rules on AI.

对很多人来说,这是一个简单的常识:随着模型变得更强大、拥有更多能力,包括用于不良目的的能力,你自然希望加强对事物的控制。

For a lot of folks, it's simply common sense that as the models get more powerful and have more capabilities, including capabilities for not great things, there is a natural ramp up in the amount of controls you want to put around things.

而且,虽然当然有一些 AI safety 倡导者就我们希望在多早引入规则以及规则应该领先能力多远进行激烈辩论,但也有许多人认为规则提升与能力提升之间的关联是合理的。

And while certainly there are AI safety advocates who have fierce debates around how early we want to put in rules and around how far ahead of capabilities we want the rules to be, there are also many for whom the correlation between the ramp in rules and the ramp in capabilities just makes sense.

如果你看看那份可能逐步增加的法规清单,我乐观的一个主要原因在于,有许多基本的常识性问题,我认为你可以在这些问题上找到共同点,并打下可以继续构建的基础。

And if you take that list of likely incremental regulations, one of my major points of optimism is in how many foundational common sense issues there are where I think you can find common ground and get foundations that you can build on top of.

这并不是说不会出现极其激烈的辩论,考虑到 AI 已经如此重要,并且在未来可能在无数方面变得更加重要。

Which is not to say that there won't be extremely contentious debates given how significant AI already is and how even more significant it could be in a myriad of ways in the future.

现在,谈到数据中心,这是一个有趣的问题。

Now, when it comes to data centers, it's an interesting one.

对我来说,这次选举在某种程度上只是证实了我们已经知道的事情,那就是数据中心开发商不必真正讨好或关心本地社区的时代已经彻底结束了。

To me, these elections sort of just confirmed what we already knew, which is that the previous era where data center developers didn't really have to court or care about local communities are completely over.

但说实话,数据中心开发商根本不用等一场选举就知道这个道理。

But frankly, it didn't take an election for data center developers to have gotten that memo.

这种转变一整年都在发生,而且我在好多节目里都聊过,我觉得我们已经开始看到其中一些变化了。

That shift has been happening all year, and as I've discussed in numerous shows, I think we're already starting to see some of those changes.

暂停NDA,承诺在基础设施和电力上自己买单,还有一大堆社区福利和激励措施,来争取个人和社区的支持。

Moratoriums on NDAs, commitments around paying their own way around infrastructure and electricity, and a lot more community sweeteners and incentives to get both individual and community buy-in.

有一些人就是不喜欢AI,不管你有多少透明度、给多少社区好处,他们绝对不会接受任何数据中心。

There will be some for whom they just don't like AI and they don't want any data centers under any circumstances with any amount of transparency or community sweeteners.

就是这样,没什么好说的。

And that's that.

但我觉得,其他人会理解的。

Others though, I think, will understand.

但对于其中一些新领导人,我觉得你会发现他们明白数据中心的经济重要性,而且他们有动力去制定一些可能严格但切实可行的规则,既给社区自主权和发言权,也为数据中心创造继续前行的道路。

But for some of these new leaders, I think you'll see that they understand the economic importance of data centers, and it will be within their incentives to create perhaps hard but achievable rules to give communities agency and a say while creating a path for data centers to move forward.

还记得新上任的德州州长Gina Hinojosa说过的那句话吗?她说她不讨厌数据中心,她讨厌的是它们对普通德州人造成的那些影响。

Remember that line from new Texas Governor Gina Hinojosa, who said that she didn't hate data centers, she hated what they were doing to ordinary Texans.

这两者之间的差别,就是一片可以被开拓出来的政治和现实空间。

The difference between those two is a political and practical space that can be carved out.

而我认为,这基本上就是你会看到的局面。

And that's basically what I think you're going to see.

而且不是白说的,如果你从来没想过这个问题,只是从最基本的公平原则出发,被问到数据中心和它们所在的社区之间应该是什么关系,你的答案会更接近这样一个版本:没有NDA,社区有自主权,数据中心自己承担基础设施和电力的费用,并且做出承诺,让自己成为社区里积极支持的一份子;而不是我们几年前的那种版本。

And not for nothing, but if you had never thought about this issue, but just from the first principles of fairness, were asked roughly what the relationship between data centers and the communities that they operate in should look like, it's a lot closer to the version where there aren't NDAs and there is community agency and the data centers pay their own way on infrastructure and electricity and where they make commitments that make them an active and supportive part of the community than the version that we had a couple of years ago.

归根结底,我还是觉得,数据中心建设差不多是我见过的最能实现多赢的政治议题。

Ultimately, I just continue to think that the data center buildout is just about the most winnable political issue I've ever seen.

我说的可赢,不是指某一方赢,而是指每个人都能赢。

And by winnable, I don't mean for one side, I mean for everyone.

现在,为了完整起见,我觉得有必要把灾难情景说清楚。

Now, I guess for the sake of completeness, it is worth articulating what the disaster scenario would be.

比如说,数据中心暂停建设的势头进一步加快,我们很快会看到新州长们采取一大堆行政层面的行动。

Let's say that the data center moratoriums pick up even more momentum, that we very quickly see a bunch of executive-level action from the new governors.

而由于这些行动,华尔街开始撤资。

And because of that action, Wall Street starts to withdraw its funding.

数据中心得不到融资,这就会让AI建设戛然而止,接着一堆neo-cloud公司破产,然后波及到更广泛的经济下行,这种下行还会自己滚起来,最终把经济环境搞得很惨,因为过去几年我们经济差不多全靠AI建设在撑着了。

Data centers can't get funded, which in turn brings a screeching halt to the AI buildout, which ends up with a bunch of the neo-clouds going bust, which spills over into a broader downturn, which picks up its own momentum and ends up ravaging the economic landscape because the AI buildout is pretty much all we've had going for us for the last couple of years.

这样一来,2028年的选举周期里,人们不仅会继续从就业、网络安全和x-risk的角度来争论AI,而且还要面对一个泡沫破裂的现实。

That sets up the 2028 election cycle to not only have people still debating AI from the standpoint of jobs and cybersecurity and x-risk, but also in light of a bubble bursting.

这当然不是没有可能,而且我觉得数据中心开发商也清楚这一点,所以他们会拼命加班,想尽一切办法找到能让他们继续推进的妥协方案。

This is certainly not outside the realm of possibility, and I think the data center developers know that, and so are gonna be working overtime to find whatever accommodation it takes to be able to continue moving forward.

那么话说回来,昨晚的事情真的改变了AI行业吗?

So when all is said and done, did things really change for the AI industry last night?

简单回答是:有一点,而且可能真的有,但更多时候,它只是确认了事情原本就在走的方向。

The short answer is sort of, and potentially yes, but mostly it was a confirmation of the direction that things were already headed.

你可以说我过于乐观,但我就是觉得,这些问题里面有很大一部分,都还有巨大的政治空间去找到更好的共识。

And call me overly optimistic, but I just think there is tons of political room to find better alignment on a huge portion of these questions.

不过我们走着瞧吧。

But we'll see.

2028年总统选举周期现在正式开始了。

The 2028 presidential election cycle starts now.

而如果这批新领导人决定,更激烈地反AI、并且真的试图叫停AI,是一个能赢的政治议题,那我可能就得收回自己这些话了。

And if this new crop of leaders decide that being more virulently anti-AI and actually trying to put a stopper on it is a winning political issue, I could be eating those words.

好了,今天的人工智能日报就到这里。

For now, that's going to do it for today's AI Daily Brief.

一如既往,感谢你的收听或收看。

Appreciate you listening or watching as always.

那下次再见,Peace。

And until next time, peace.
已剔除 3 处广告(点击展开查看)
M1
M10:36广告 · 已剔除

First of all, thank you to today's sponsors, KPMG, CData, Granola, and Blitzy.

To get an ad-free version of the show, go to patreon.com/AIDailyBrief, or you can subscribe on Apple Podcasts.

And if you want to learn more about sponsoring the show, send us a note at [email protected].

Also, quick reminder today that I have a new podcast, How We Got Here.

It is a 45-minute to hour-long deep dive on the history behind the biggest trends, conditions, and phenomenon in today's world.

The first couple episodes are out, and the first one, How Tech Lost America, is particularly germane to today's conversation about the midterm elections, as it traces the 15-year history that set up the context into which data centers would be received.

You can find How We Got Here with NLW on any of your podcast apps, or just go to howwegothere.com.

M1
M13:38广告 · 已剔除

In fact, digging deep on those particular issues is exactly why I created the new show, How We Got Here, and why the first episode was in fact How Tech Lost America, which is about the 15-year history that set up the context for data centers today.

M1
M113:48广告 · 已剔除

If you're leading AI inside an enterprise, you already know that the gap right now isn't capability, but execution.

That's why KPMG's You Can with AI is back with a new season featuring conversations with leaders like Surojit Chatterjee of Ema, May Habib of Writer, McKesson CIO Ellery Fisher, and others focused on practical execution.

What's working, what's not, and what it actually takes to move from pilots to real scaled impact across strategy, data readiness, governance, workforce, and value.

And of course, it's co-hosted by me, Nathaniel Whittemore.

Go listen and subscribe at www.kpmg.us/aipodcasts.

That's www.kpmg.us/aipodcasts.

This episode is brought to you by CData.

Token spend is climbing, security is holding agents back, and companies rolling out AI initiatives might not know how it's truly being used.

The root cause?

AI is cut off from the systems and context your business runs on.

That's why CData built Connect AI Gateway— one control point between your AI and your systems.

Connect AI Gateway is not just another static model router.

Prompts get intelligently routed to the right model with exactly the tools and data it needs, governed down to the record.

The result?

Accurate answers at a fraction of the token cost with a context engine that learns your business with every interaction.

Built on the connectivity layer powering Palantir, Google, Salesforce, and hundreds more.

Give it a try and start your free trial at cdata.ai.

That's C as in connect, data.ai.

You're in back-to-back meetings all day, you're trying to stay present, but you're also worried you'll forget the decision, the action item, the important next step.

This is a scenario that we absolutely live here at the AI Daily Brief, given how fast things move and change and how quickly we're evolving everything that we do.

That's where Granola comes in.

Granola is an AI-powered notepad for meetings.

You jot down rough notes like you always do, and in the background, Granola transcribes and turns them into clear, useful notes when the meeting ends.

No bots joining your calls, no distractions, just a clean notepad that helps you focus.

During or after the call, you can chat with your notes, ask Granola to pull out action items, help you negotiate, write a follow-up email, or even coach you using recipes which are pre-made prompts.

Once you try it on a first meeting, it is very hard to go without.

Head to granola.ai/AIDaily and use code AIDaily to get 100% off for 3 months for new users.

That's granola.ai/AIDaily.

And go to blitzy.com to get 100% off for new users for 3 months.

Here's why most legacy modernization projects fail: the AI doing the work can't understand codebases at scale.

It sees a small slice of context, examines syntax, and misses years of decisions distributed across the global application ecosystem.

Blitzy solves this the way it solves everything: grounded in your code before any migration begins, Blitzy's agents reverse-engineer the entire legacy system into a persistent knowledge graph.

Every dependency, every constraint, every piece of tribal knowledge that used to live in one engineer's head.

From that understanding, Blitzy autonomously executes language migrations, framework upgrades, and monolith-to-microservices transformations, all validated end-to-end.

One Blitzy customer modernized a $10 million monolithic insurance stack in 16 weeks against a 137-week baseline with coding agents.

That's 9x compression.

Retire technical debt while accelerating your roadmap.

See how at blitzy.com.

That's blitzy.com.