内容来源：https://www.wired.com/story/uncanny-valley-podcast-wired-roundup-gemini-3-release-nvidia-earnings-epstein-files-fallout/

内容总结：

本周科技界热点频出，从政治丑闻到人工智能商业化的新动向，再到科技产品对社会伦理的冲击，多个事件引发广泛关注。

政治漩涡与科技监管博弈
特朗普政府近期因处理爱泼斯坦案件文件陷入舆论危机。在各方压力下，特朗普签署法案公开约2万份相关文件，但邮件证据显示其与爱泼斯坦的关系比公开承认的更复杂，使政府形象严重受损。与此同时，特朗普被曝拟签署行政命令，要求司法部长组建特别工作组，以"违反联邦法律"为由起诉各州的人工智能监管法规。此举被业界视为向硅谷科技巨头的明确示好，与谷歌、OpenAI等企业反对分散化监管的立场不谋而合。

AI产业面临盈利考验
在商业领域，AI企业正全力探索盈利模式。谷歌发布新一代Gemini 3模型，宣称其推理、视频生成和编程能力显著提升，并强调通过将AI整合至地图、邮箱等成熟产品实现差异化竞争。OpenAI则通过组建应用团队和心理健康顾问委员会，应对用户增长放缓与产品伦理的双重挑战。值得注意的是， Anthropic等企业选择专注企业服务，与追求聊天机器人"陪伴感"的同行形成鲜明对比。

与此同时，英伟达CEO黄仁勋在财报会议上驳斥AI泡沫论，强调公司手握5000亿美元订单。但投资者仍担忧其与初创企业形成的"投资-采购"循环生态存在风险。

科技伦理新争议
值得关注的是，两名摩门教青年开发的戒色应用Relay引发社会讨论。该应用通过治疗视频、日记功能帮助用户戒除色情成瘾，已获超10万用户。但在美国各州加强网络内容监管的背景下，这类应用如何平衡宗教理念与心理健康需求成为新的议题。

与此形成呼应的是，OpenAI和xAI相继开放情色对话功能，反映出AI企业在用户增长压力与产品伦理之间的艰难取舍。业内观察指出，尽管 Anthropic 等企业坚持不开发陪伴型AI，但市场数据表明用户更青睐具有情感反馈的聊天机器人，这使科技企业面临商业价值与社会责任的深层矛盾。

（根据《连线》杂志"Uncanny Valley"播客内容整理）

中文翻译：

本期节目中，主持人佐薇·希弗将与资深撰稿人马克斯·泽夫共同探讨本周五大焦点事件：从爱泼斯坦文件公开引发的政治余波，到两位年轻摩门教徒开发戒除"边缘性行为"应用的初衷。随后我们将深入解析Gemini 3的发布，以及谷歌与OpenAI等企业如何聚焦人工智能的盈利模式。

节目提及文章：
-《特朗普如何失去爱泼斯坦舆论主导权》
-《草案行政令显示特朗普瞄准各州AI监管法案》
-《英伟达CEO驳斥AI泡沫论，投资者持保留态度》
-《年轻摩门教徒开发应用助男性戒除自慰成瘾》
-《Gemini 3问世——谷歌宣称将提升搜索智能》

诚邀您填写听众问卷，助力《诡异谷》节目优化。
您可通过Bluesky关注佐薇·希弗@zoeschiffer、马克斯·泽夫@mzeff。欢迎来信至uncannyvalley@wired.com。

收听方式
您可通过本页音频播放器随时收听本期内容，若想免费订阅每期节目，请参考以下方式：
苹果用户可直接点击链接或打开播客应用，也可下载Overcast、Pocket Casts等应用搜索"诡异谷"。本节目亦在Spotify同步更新。

转录文本
注：本文本为自动生成，可能存在误差。

佐薇·希弗：欢迎收听《连线》杂志旗下播客《诡异谷》。我是《连线》商业与产业版总监佐薇·希弗。本期节目将为您梳理本周五大要闻，包括谷歌与OpenAI在开发面向消费者的AI产品时如何聚焦盈利模式。今日特邀《连线》资深撰稿人马克斯·泽夫加入对谈。马克斯，欢迎来到《诡异谷》。

马克斯·泽夫：感谢佐薇，很荣幸参与节目。

佐薇·希弗：首条新闻关乎特朗普政府如何在爱泼斯坦文件风波中彻底丧失舆论主导权。虽然此事每周都有新进展，但本周尤为关键。我们的同事大卫·吉尔伯特撰文指出，特朗普总统正面临来自各方——从QAnon追随者到两党议员联盟——要求公开已定罪的性犯罪者杰弗里·爱泼斯坦相关文件的压力。本周三特朗普终于签署国会通过的法案，公开约两万份爱泼斯坦文件。但正如大卫所言，此事对政府形象的损害似乎已难以挽回。

马克斯·泽夫：确实难以置信，特朗普政府在此案上的立场与最初相比已发生巨变。过去这仅是竞选议题，如今却演变成需要持续应对的危机。大卫的文章完整还原了事件脉络，回顾历程确实令人震撼——从2017年QAnon首次提及爱泼斯坦，到2019年其被捕，如今临近2026年事件仍在发酵。

佐薇·希弗：我难以理解政府决策逻辑。既然持续炒作此事，终须面对后果。若假设当局因可能牵涉其中而抗拒文件公开，这无异于玩火自焚。不过，这桩阴谋论确实持续带来"惊喜"。今年早些时候，司法部长帕姆·邦迪称爱泼斯坦客户名单"就在她办公桌上"，联邦调查局随即否认。随后司法部公布爱泼斯坦狱中死亡录像，经《连线》分析发现数分钟片段遭删除。最新文件显示特朗普与爱泼斯坦的关系可能比外界所知更为复杂。爱泼斯坦2017年发送的邮件暗示其掌握特朗普当时观点，这距特朗普声称两人最后接触已逾十年。虽然不愿陷入阴谋论，但现政府确实屡次将局面越搅越乱。

马克斯·泽夫：他们向来擅长于此。玩弄阴谋论本就不是政治良策，尤其当政要可能牵涉核心指控时。虽不愿深究，但尼克·富恩特斯、坎迪斯·欧文斯等意见领袖仍在推波助澜。即便此时全面公开文件，此事对特朗普政府的困扰恐怕不会终结。

佐薇·希弗：且看此事会否影响中期选举。下条新闻仍涉政坛，但缘由略有不同。你与同事玛肯娜·凯利获悉特朗普正考虑签署行政令，旨在挑战各州监管AI的个体努力。请谈谈此事以及你们如何获取该命令草案内容。

马克斯·泽夫：我们昨日获知这份行政令草案已在华盛顿与硅谷广泛流传。草案暂定名《消除州法律对国家AI政策的阻碍》已说明一切——将州法律称为"阻碍"虽显滑稽，但表明特朗普将指令司法部长帕姆·邦迪组建AI诉讼特别工作组。若草案生效，该工作组核心使命就是对各州AI监管法案提起诉讼，指控其违反联邦法律。政府声称要针对侵犯言论自由与州际贸易的州级AI法规，但关键在于：既可通过司法程序解决，何必专门设立机构？

佐薇·希弗：这再次释放信号，表明特朗普政府与大型科技公司立场高度一致。AI巨头始终强烈反对各州制定差异化的监管法规。你在文章中提到，由安德森·霍洛维茨、谷歌及OpenAI等支持的行业组织"进步商会"多年游说反对此类监管，认为这会阻碍创新与发展。

马克斯·泽夫：这基本契合硅谷各方的诉求。过去数年——尤其是近几个月——硅谷高管与特朗普关系日益密切，这份行政令正是他们长期呼吁的体现。

佐薇·希弗：你们获取的草案中还有值得玩味的表述，暗示政府对"觉醒AI"的持续关切也将成为打击目标。相关条款称将针对"要求AI模型篡改真实输出"的监管。对此我想指出：是否有证据表明州法律曾有此要求？

马克斯·泽夫：并无实证。他们特别关注科罗拉多州某法案，该法禁止AI公司植入算法歧视并需上报合规情况。说实话，我认为特朗普政府——尤其是大卫·萨克斯——对一年前Gemini生成图像失误事件耿耿于怀，持续借此指责AI模型失真。但事实上此后未见更多证据，政府具体所指令人费解。

佐薇·希弗：我个人推测，他们走进推特办公室看见满屋"保持觉醒"T恤时深受刺激，至今未缓过神来。现在的反应多少带有情绪色彩。下条新闻仍关AI领域：英伟达本周举行财报会议，同事帕雷什·戴夫为我们提炼了精华内容。显然，英伟达CEO黄仁勋主动谈及了敏感话题——AI泡沫。

马克斯·泽夫：没错，这个我们已习以为常的AI泡沫，如今人人口中相传。

佐薇·希弗：不出所料，已成为全球最具价值上市公司掌舵人的黄仁勋为英伟达前景辩护。其论点与萨姆·奥尔特曼等人如出一辙：AI正在征服世界，英伟达芯片对技术革命不可或缺。他以创纪录的季度销售额佐证，并透露公司积压订单价值约5000亿美元。这番信心喊话助英伟达股价从数周跌势中略有回升，你我想必都密切关注着。

马克斯·泽夫：每次财报会议，黄仁勋总会出面捍卫AI产业。数月前他强调规模效应依然有效，如今又谈整体AI泡沫。但上周彼得·蒂尔清仓英伟达的消息令人警觉。

佐薇·希弗：毋庸置疑的是，英伟达在此轮AI加速发展中全力押注。其数据中心业务现占营收90%，取代了昔日主导的游戏芯片。每当OpenAI与其他公司合作，背后总有英伟达的身影，它已与行业所有环节深度绑定。

马克斯·泽夫：英伟达成功渗透各AI公司，但这也引发担忧：其与许多受资初创企业存在循环交易——既是客户又是投资者。这种深度捆绑使得黄仁勋持续为AI泡沫辩护不足为奇。

佐薇·希弗：持泡沫论者还担忧：GPU占数据中心建设成本大头，且需每三年更换——英伟达不断推出尖端芯片，企业为保持竞争力必须采购。但定价是否充分考虑更新周期？只要需求持续，黄仁勋便认为英伟达安如磐石。

马克斯·泽夫：且观后效。

佐薇·希弗：进入下一话题前，我们需要格外保持专业——虽然向来如此。读完《连线》撰稿人马萨·巴斯比的报道才知，两位年轻摩门教徒开发了帮助男性戒除色情成瘾及"边缘性行为"的应用。实话说，我此前从未听闻此术语，大为震惊。"边缘性行为"即长时间自慰而不释放。这款名为Relay的应用由27岁钱德勒·罗杰斯创建，旨在为Z世代提供戒除此类行为与色情依赖的途径。我有些其他建议：比如出门走走、与人交谈，但绝非嘲讽——我理解这对某些人确属难题。

马克斯·泽夫：愿主保佑你。佐薇，解决方案是软件啊。这个故事有多重维度：开发者基于宗教信仰创建应用，这无可厚非；对部分受访者而言也确实有效。该应用用户量惊人——超十万人使用。但将其置于当前反色情舆论背景下则显微妙。共和党人正全力推动反色情立法，视其为必须根除的社会问题。我好奇此类软件将如何被运用。

佐薇·希弗：相关立法本身已制造新问题。年龄验证规则既使网站在某些州运营困难，又引发隐私担忧，更涉及耻感文化。该应用提供治疗师视频、每日日记引导、线上小组分享会等功能。但有专家指出，试图控制人类本能冲动却忽视潜在情绪根源实属谬误。值得一提的是——并非凡事都要扯上AI——OpenAI宣布允许情色对话，Grok也推出AI伴侣功能，可见相关议题正引发广泛关注。

马克斯·泽夫：OpenAI与xAI虽未明言，但实质上推出了"边缘性行为"功能，开始拥抱情色内容。而Relay创始人则公开反对AI的这些新情色功能。这对"边缘性行为"的未来意味着什么？正是我们讨论的核心。

佐薇·希弗：所幸我们可以转换话题。休息过后，我们将探讨OpenAI与谷歌如何竞逐AI时代的盈利圣杯。

【休息时间】

欢迎回到《诡异谷》。我是佐薇·希弗，今日与资深撰稿人马克斯·泽夫继续对谈。

佐薇·希弗：进入核心议题。本周我们关注谷歌与OpenAI如何以更清晰的消费者应用导向与盈利目标布局未来。虽然理想中他们无需考虑成本，但AI研发维护确实耗资巨大。与其他AI企业一样，他们亟需解决关键问题：如何让用户愿意付费。周二谷歌发布最新AI模型Gemini 3，同事威尔·奈特报道称该模型擅长推理、视频生成与编程。与黄仁勋的乐观论调不同，DeepMindCEO德米斯·哈萨比斯承认AI领域渐趋拥挤，但他坚信谷歌将通过将AI融入谷歌地图、Gmail及搜索等产品脱颖而出。你对此论点怎么看？

马克斯·泽夫：这个观点很有意思。不少业内人士注意到，德米斯与桑达尔近期都提及AI泡沫与估值失衡。作为谷歌，此说辞合乎逻辑——即便泡沫破裂，谷歌仍可安渡难关。他们确实正在将AI整合至现有产品线，并宣称成效显著：依赖Gemini的视觉搜索量激增70%，Gemini应用月活达6.5亿。这些拥有广泛用户基础的真实产品，使谷歌占据有利竞争位置。

佐薇·希弗：核心逻辑是：即便AI市场下行，谷歌既有产品用户群仍能提供缓冲。将AI植入这些产品，比单纯提供聊天机器人的企业更具韧性。我最近在洛杉矶采访了OpenAI新任应用业务CEO菲吉·西莫，问及是否担忧公司业务过度扩张时，她表示其职责正是通过组建专业团队分目标管理以降低风险。

OpenAI也在多元化拓展产品线。但在我看来谷歌正在迎头赶上——如今每次模型发布都会重排座次，各企业在技术能力上已非常接近。

马克斯·泽夫：谷歌在AI竞赛中的进步确实惊人。回想2022年他们与苹果等同处被动，如今已跻身领军者之列。但正如你所说，下周格局可能再生变数。OpenAI在Gemini 3发布次日推出新模型，部分基准测试表现更优。但谷歌确已重返赛场，其产品战略颇具说服力，不过各方都仍在摸索方向。

佐薇·希弗：并非自夸，但去年有编辑邀稿撰写"谷歌如何输掉AI竞赛"时，即便当时谷歌落后，我仍认为断言为时尚早。他们拥有庞大业务体系，正如你所说，可将AI嵌入现有产品获得优势，更有资金吸引人才。如今能迎头赶上虽令人印象深刻，却也在情理之中。

马克斯·泽夫：谷歌另一优势在于专注提升工作效率场景——人们常在职场使用谷歌产品。而OpenAI面临不同挑战：用户更多将ChatGPT视为伴侣。当你采访菲吉·西莫时，她如何考量ChatGPT用户的心理健康？

佐薇·希弗：这确实是她重点关注的领域。组建百人心理健康专家委员会正是她的提议。首席战略官关乃恩征询解决方案时，她主张寻求外部支持。但她将持续与专家会晤获取反馈。更深层在于：OpenAI最初作为非营利机构成立，怀抱崇高理想，如今却拥有作为公益公司的营利子公司。在扩张过程中，它能否真正坚持造福人类的初心？

当前AI情色内容正是考验：ChatGPT增长曲线首次走平，而采取无限制AI伴侣路线的Grok持续增长。据我们接触的研究人员分析，这正是Grok增长的主因——用户渴望以此方式与聊天机器人互动。这迫使OpenAI重新考虑推出伴侣功能。

马克斯·泽夫：同事里斯·罗杰斯本周采访Anthropic研究人员，该公司可坚持高标准——大幅降低Claude的谄媚倾向，明确不开发伴侣功能，专注企业级能力建设。但OpenAI与谷歌必须让聊天机器人更具对话趣味性，有时不得不采取未必最利于用户健康的策略。这将成为该公司长期的核心矛盾。

佐薇·希弗：OpenAI的应对策略是赋予用户更多交互自由度。毕竟若像GPT-5那样过于正式，用户强烈反弹——人们显然更偏爱带些奉承的语调。

马克斯·泽夫：虽然用户不愿承认喜欢被聊天机器人吹捧，但这确是使用黏性的重要因素。事情远比表面复杂。

佐薇·希弗：本期节目到此结束。提及文章均附于节目备注。敬请关注周四《诡异谷》关于"科技能否消除劣质芯片"的讨论。本期节目由阿德里安娜·塔皮亚与马克·莱达制作，马科罗声学的阿马尔·拉尔完成混音。执行制作人凯特·奥斯本，康泰纳仕全球音频总监克里斯·班农，《连线》全球编辑总监凯蒂·德拉蒙德。

英文来源：

In today’s episode, host Zoë Schiffer is joined by senior writer Max Zeff to discuss five stories you need to know about this week—from the political fallout after the release of the Epstein files, to why two young Mormon men created an app to help men stop “gooning.” Then, we dive into Gemini 3’s release and how companies like Google and OpenAI are homing in on AI profitability.
Articles mentioned in this episode:

• How Donald Trump Lost Control of the Epstein Spin Cycle
• Trump Takes Aim at State AI Laws in Draft Executive Order
• Nvidia CEO Dismisses Concerns of an AI Bubble. Investors Remain Skeptical
• Young Mormons Built an App to Help Men Quit Gooning
• Gemini 3 Is Here—and Google Says It Will Make Search Smarter
  Please help us improve Uncanny Valley by filling out our listener survey.
  You can follow Zoë Schiffer on Bluesky at @zoeschiffer and Max Zeff on Bluesky at @mzeff. Write to us at uncannyvalley@wired.com.
  How to Listen
  You can always listen to this week's podcast through the audio player on this page, but if you want to subscribe for free to get every episode, here's how:
  If you're on an iPhone or iPad, open the app called Podcasts, or just tap this link. You can also download an app like Overcast or Pocket Casts and search for “Uncanny Valley.” We’re on Spotify too.
  Transcript
  Note: This is an automated transcript, which may contain errors.
  Zoë Schiffer: Welcome to WIRED's Uncanny Valley. I'm Zoë Schiffer, WIRED's director of business and industry. Today on the show we're bringing you five stories that you need to know about this week, including how companies like Google and OpenAI are homing in on profitability as they develop their AI consumer-facing products. I'm joined today by WIRED senior writer Max Zeff. Max, welcome to Uncanny Valley.
  Max Zeff: Thanks, Zoë. It's great to be here.
  Zoë Schiffer: Our first story is about how the Trump administration has completely lost the narrative around the Epstein files. It feels like we could talk about this every week, but this week in particular, it's really important. Our colleague David Gilbert wrote about the increasing pressure that President Trump was receiving from everyone, and we really do mean everyone, from QAnon followers to a coalition of GOP lawmakers and democratic legislators to release the files related to convicted sex offender Jeffrey Epstein. On Wednesday this week, Trump finally signed the bill passed by Congress releasing roughly 20,000 documents on Epstein. But as David put it, the damage to the administration's brand already kind of feels like it's done.
  Max Zeff: Yeah, it's really just unbelievable how far the Trump administration has gone around the block on the Epstein case compared to where it first started. I mean, in the past this was a political campaign thing, but now it's kind of an ongoing crisis that they really need to manage. I really liked David's article just painting the full picture of where this started, and it truly is wild to look back on. Looking back to 2017 when QAnon first started talking about Jeffrey Epstein all the way up through 2019 when he was arrested, and now it's shocking that we're about to be in 2026 and the story is still unfolding.
  Zoë Schiffer: I have a hard time putting myself in the administration's shoes. How did they not realize that at a certain point, if you tease this, you're going to have to deliver, and if you are nervous, hypothetically, that you could be involved in any way and you really don't want this to come out. It just seems kind of like a “play with fire, you'll get burned” situation, but I mean, it's the conspiracy theory that never stops giving.
  Earlier this year, Attorney General Pam Bondi said that the Epstein client list was quote “sitting on her desk,” and then the FBI quickly backtracked that statement. Then the DOJ published video footage of Epstein's death in jail, and after analyzing the video, WIRED found that several minutes of the footage had been deleted, and in this latest batch of documents, it appears that Trump's relationship with Epstein might've been way more complicated than people originally thought. One email sent by Epstein implies that he had intimate knowledge of Trump's views in 2017, which is more than a decade after Trump claims that he had last talked to him. So I'm trying not to get us to join the conspiracy circles, Max, but let's just say that the current administration, it feels like they have made the situation messier for themselves time and time again.
  Max Zeff: Yeah, they're great at that. I mean, it really feels like conspiracy theories are not a great thing to play with generally in politics, but especially when you are maybe directly involved with some of their central claims. I don't want to veer too down into this rabbit hole, but that's not stopping some influencers from doing so, like Nick Fuentes and Candace Owens, and they haven't been able to really let these go, and it seems like even if they do release the files at this point in their totality, it doesn't feel like this story is going to be over for the Trump administration.
  Zoë Schiffer: Well, I guess we'll wait to see what impact, if any, this has on the midterm elections. Our next story keeps us in the thick of politics, but for slightly different reasons. So you and our colleague, Makena Kelly, just learned that President Trump is considering signing an executive order that would seek to challenge individual state efforts to regulate AI. So tell us about that and how you got firsthand knowledge about what this order might say.
  Max Zeff: Yeah, so we heard yesterday that a draft of this executive order was going around DC and in Silicon Valley, and it really spread far and wide. I think the working title of the draft makes it all pretty clear. It's called Eliminating State Law Obstruction of National AI Policy, which is a funny way to refer to state law, as obstruction, but it really indicates that Trump is going to instruct US attorney general Pam Bondi to create an AI litigation task force. And the whole purpose of this new task force, if the draft goes into effect as it states, is to just sue states in court for passing AI regulations that the Trump administration deems to violate federal laws. It really wants to go after state AI laws that it says infringe on things like free speech and interstate commerce. But I think that there's also a bigger question of why do you need an executive order to do this when you could just take these states to court over that? You don't need a special body to be in charge of that.
  Zoë Schiffer: It also feels like a huge signal once again that the Trump administration is cozying up, siding with the big companies on this. This approach, the kind of state-by-state regulations, and sometimes the regulations differ, has been something that big AI companies are vehemently against. And you wrote in your piece that the Chamber of Progress, which is an industry group that's backed by Andreessen Horowitz, Google, and OpenAI among others, has lobbied against these efforts for years, arguing that it's an approach to AI regulation that makes it really hard to innovate and grow.
  Max Zeff: This is basically exactly what a lot of Silicon Valley players have been asking for. We know that over the last couple of years, and really the last couple months, the Silicon Valley executives have gotten very close to Trump, and this executive order really represents a lot of the things that they've been demanding.
  Zoë Schiffer: And there's also some interesting language in the draft that you were able to get that hints that the evergreen concern that the administration has about quote “woke AI” is something they're going to be going after specifically as well. That language says that they'll target regulations that quote “require AI models to alter their truthful outputs.” I have a lot to say about this, but one is just, do we have any evidence that state laws have done this before?
  Max Zeff: We don't. They're specifically focused on this law in Colorado that says that AI companies can't put algorithmic discrimination into their AI models, and they have to report on how they're doing that. To be honest, I really think that the Trump administration and David Sacks in particular are very hung up on when Gemini had its AI image-generation fiasco, which was about a year ago. They have continuously pointed to this as a way that AI models are not being truthful and are altering reality. But the truth is that we just haven't seen a lot of evidence since then, and I think it was unclear what the administration is really talking about here.
  Zoë Schiffer: Right. I mean, my personal theory is that they were very triggered by walking into the Twitter office and seeing a room full of StayWoke shirts, and they're still recovering from that. So I think a lot of what we're seeing now may be a little reaction to that. Moving on to our next story, still in the world of AI, Nvidia had their earnings call this week, and our colleague Paresh Dave brought us the juiciest bits. So apparently Nvidia CEO Jensen Huang did not need any prompting to address the elephant in the room. The AI bubble.
  Max Zeff: Yes, the AI bubble we've all resigned ourselves that we just live in now, everyone says it.
  Zoë Schiffer: Unsurprisingly, Huang defended the viability of Nvidia's position, having become the world's most valuable publicly traded company over the past three years. His argument was pretty straightforward, not that different from what you hear from Sam Altman and others. It's basically that AI is taking over the world and Nvidia chips will be sorely needed to power the technology revolution that is already underway. He backed it up by saying that the company reported record quarterly sales, and in the call executives reiterated that they have about $500 billion worth of unfilled orders, and this pep talk helped Nvidia recover a bit from the sell-off that it has been experiencing in the past few weeks, which I think you and I have both been watching with interest.
  Max Zeff: Yeah, it's become a theme where every time Nvidia has earnings, Jensen just gets on a call and defends the AI industry and why everything is going fine. I remember a few months ago he was defending how scaling laws were still intact, and now it's just the AI bubble at large. But last week news broke that Peter Thiel sold his Nvidia stake, which felt like a very concerning warning.
  Zoë Schiffer: Yeah, I think that one thing that everyone can agree on is that Nvidia is undoubtedly one of the companies that has gone all in during this AI acceleration moment. For better or worse, about 90 percent of Nvidia's sales, which were once dominated by chips for personal gaming computers now come from its data center business, and it feels like every time one of these partnerships between OpenAI and another company, Nvidia's in there somewhere, it just feels like it's attached to everyone else in this industry at this point.
  Max Zeff: Yeah, it's done a great job of infusing itself with every AI company, but also, I mean, that's been a major concern. There's been a lot of talk of these circular deals where Nvidia really depends on a lot of these startups that it's also funding. It's a customer, it's an investor. Nvidia is so wrapped up in this. So I guess in that way, it's not that surprising that Jensen is defending the AI bubble constantly now.
  Zoë Schiffer: Yeah. It's also worth saying that one of the fears that people who have the fear of the AI bubble will talk about is the fact that the GPUs are the majority of the cost of building out a data center, and they need to be replaced, what, every three years? Nvidia releases new chips and they're cutting edge, and companies need to buy them in order to compete. I think the fear is that that renewal cycle isn't quite factored into the pricing, but as long as people continue to buy chips, what Jensen is saying is, "No, no, we're insulated right now."
  Max Zeff: Right. We'll see if that's really true though.
  Zoë Schiffer: One more story before we go to break, and to get through this one, we both have to be extra professional. I'm not sure Max, which we always are, but just a little extra. You will see what I mean. WIRED contributor Mattha Busby reported on how two young Mormons created an app to help other men break their porn addiction and gooning habits. I'm going to be real. I had never heard this term before reading this story, and I was shocked. OK, if you're not familiar with gooning, it's basically just another word for edging. That is long hours of masturbation without release. This app called Relay was created by 27-year-old Chandler Rogers with the mission of providing his Gen Z peers a way to stop doing this and to generally escape from the clutches of porn. I have some other ideas. I feel like go outside, talk to a human, but I don't want to be mean, because I do feel like this could be really difficult for people.
  Max Zeff: God bless you. The solution is software Zoë, and thank you for explaining that. So yeah, there's definitely layers to this story. I mean, obviously the creators of this created it to be aligned with their religious beliefs, they're Mormons, and that's great for them. And also it seems that for some of the people who Mattha spoke to, it's really helping them. This app actually has an impressive amount of users. I was honestly shocked when I saw that it had over a hundred thousand people using it, which is all good things. The part where it gets a little bit dicey I think is when we see it in the context of the broader discussion around porn that's going on right now. I mean, a lot of Republicans have really zeroed in on passing anti-porn legislation. There's this big push in the country right now. This is just a problem that a lot of Republicans have accepted is like, this needs to be stamped out completely. And I'm just curious how a piece of software like this gets used.
  Zoë Schiffer: Yeah, it's interesting. I feel like a lot of the legislation has created its own problems or potential problems. It's like age verification rules that either make it very difficult for these sites to operate in those states or have real privacy implications, and there's a lot of shame involved in this too. So the app offers some, I don't know, seemingly good things like videos by therapists, daily journal prompts, live group sharing sessions. But some experts argue that trying to control a very human urge while possibly ignoring the underlying emotions that might make this compulsion present in the first place is a mistake. And this is kind of coming in context, not to have an AI angle on everything, but OpenAI has said that they're going to be allowing erotic conversations. Obviously Grok has this AI companion feature, so it feels like a lot of these issues are very much in the zeitgeist for people right now.
  Max Zeff: Yeah, OpenAI and xAI, I guess I'm not sure if they would call it gooning features, but that's effectively what they've released. They're really embracing erotica now. The founder of this app is really taking a stand against AI here and some of these new erotic features. So I guess what does this all mean for the future of gooning? That's what we're talking about here.
  Zoë Schiffer: That's what we're talking about. But luckily for both of us, we can move on. Coming up after the break, we dive into how companies like OpenAI and Google are trying to race toward the golden egg of the AI era profitability.
  [break]
  Welcome back to Uncanny Valley. I'm Zoë Schiffer. I'm joined today by our senior writer Max Zeff.
  Zoë Schiffer: Let's dive into our main story. This week we had a look into how Google and OpenAI are both laying out their ambitions with a clearer focus on consumer-facing applications and future profitability. We like to think that they don't need to think about this, but AI is really expensive. Building AI, maintaining it is really, really expensive. And so they do need to have this as a factor. Basically, like many other companies in the AI space, they're trying to find answers to the urgent question of how do we get users to actually pay for this. On Tuesday, Google released Gemini 3, its latest AI model and chatbot, and our colleague Will Knight reported that this latest model is skilled at reasoning, generating video, and writing code. DeepMind's CEO, Demis Hassabis went a different route compared to Huang's in the Nvidia earnings call, he actually admitted that the AI space was getting a little crowded, but he believes, no surprise, that Google will come out on top because they're diversifying their use of AI in products like Google Maps, Gmail, and search. I'm curious just to start, what did you make of that argument?
  Max Zeff: I think it's an interesting thing. I talked to a lot of people who thought it was notable that Demis and Sundar also actually in his comments recently has talked about the AI bubble and how some of these valuations don't make sense. And of course if you're Google, that argument kind of makes sense because if the bubble pops, Google will probably be OK. And it's true that they've been integrating AI into all of these products that they already have, and they argue that it's making them all better, and they seem to share some evidence for that. They said that there's a 70 percent spike in visual search, which relies on Gemini. So people are using search more in some ways. And also just the Gemini app itself has 650 million monthly active users now according to Google. These are real products that they have that have pretty wide user bases, and it doesn't seem that far-fetched that Google is very well positioned at this point.
  Zoë Schiffer: Basically, the idea is that even if there's a downturn in the AI market, Google already has users for all of those other products. So putting AI into those products, they're way more insulated than a company that's just providing, say, a chatbot. I also recently spoke with OpenAI's new CEO of applications, Fidji Simo at her home in LA, and when I asked her about whether she was worried that the company was kind of expanding in a bunch of different directions simultaneously, she said that part of her job, part of the reason she was hired was to minimize that risk.
  She wants to hire really smart people to own these goals individually so that it's not one company trying to do too much as much as it is a bunch of different tag teams that are tackling their specific problem. So OpenAI is trying to diversify in its way too. They are rolling out more products, but it still feels to me, and I'm curious what you think about this, Google is catching up. It almost feels like every time there's a model release these days, that company vaults to the head of the pack, and then the next time there's a model release that company's a little bit ahead because they're all so close at this point in terms of capabilities.
  Max Zeff: Yeah, I think that it's really remarkable to see how far Google has come in the AI race. We were talking about in 2022 how they were caught flat-footed, they were in a similar position to Apple and all these other companies, and now that's just not the story at all. I mean, it's really remarkable how Google is a leading company today, and I think you're a hundred percent right. It might not be the case next week. I mean, OpenAI released a new model a day after Google's Gemini 3 that I think outperforms it on some benchmarks, but really they're in the race now, and they have a compelling strategy around products, but they all really are just still throwing things at the wall.
  Zoë Schiffer: I don't want to pat myself on the back too much, but I do remember last year an editor reaching out and saying, "Would you be interested in writing a feature on how Google lost the AI race?" And even then when it was pretty behind, I was like, "I feel like we're a little early to declare that." They still have such a big business. Like you said before, they can slot AI into a lot of their existing products and have a leg up on some of their competitors, and they also just have a lot of money to hire talent. So I think it's impressive, perhaps not entirely surprising that they've been able to pull ahead in this way.
  Max Zeff: And I think another interesting thing about Google is that I think this came up in Will's conversation with Demis, which is that they're really trying to focus on increasing productivity in these places. Like people use Google at work and stuff, and I think OpenAI has a slightly different problem where people are using ChatGPT much more like a companion. And I think that seems to be a big problem for them. It might be for Gemini as well, but when you talked to Fidji Simo, how did she think about mental health with ChatGPT's users?
  Zoë Schiffer: So I do think that this is a big area where she is focused and taking the concerns pretty seriously. I think the Council on Mental Health or Well-Being, that group of a hundred experts that OpenAI has gathered, that actually was Fidji Simo's idea. I think Jason Kwan, the chief strategy officer, had reached out being like, "How should we try and solve this problem?" And she was like, "Let's get outside support." And so she's going to be meeting with those people regularly and getting feedback. But I think the bigger thing that this is getting at is OpenAI was founded as a nonprofit. It had this really lofty goal, but now it has a big for-profit subsidiary that's a public benefit corporation. And will it really be focused on making sure that AI really does benefit everyone or will it have other goals as it tries to scale and grow?
  And already I think that the AI erotica is a big part of this because it's coming at a moment when ChatGPT growth, which has been like numbers go up and to the right basically this whole time, is flat-lining a little bit. And we're seeing Grok actually continue to grow, the no-holds-bar approach to AI companionship, and I think that, you and I both talk to researchers in this space, that's really one of the reasons that people think Grok is growing, because people want to engage with chatbots in this way. And so it kind of puts pressure on OpenAI to say, "OK, just kidding. We'll do companions as well."
  Max Zeff: One of our colleagues, Reece Rogers, he talked to researchers at Anthropic this week, and Anthropic can take the high road here. They can reduce sycophancy in Claude a lot. They are not building companions. They've completely gone down this enterprise route where they're just trying to build AI that's very capable. OpenAI and Google are not in the same position, where they really have to make their chatbots very enjoyable to talk to. And sometimes that means pulling levers that are not the best for users' health. So I do think that this is a core tension at the company that we'll probably be talking about for a long time now.
  Zoë Schiffer: Yeah, it feels like OpenAI is dealing with it by basically saying, "OK, we're going to give people more flexibility in how they want to engage with the chatbots." I mean, I would imagine because they're worried that if they just made it really formal like they did with GPT-5, people really reacted poorly to that. They're very attached to the perhaps slightly more sycophantic kind of tone.
  Max Zeff: And I don't think that people like to admit that they like it when their chatbot is praising everything they do, but it seems like that's a really big factor in how much you use it. So it's a little more complicated than just that.
  Zoë Schiffer: That's our show for today. We'll link to all the stories we spoke about in the show notes. Make sure to check out Thursday's episode of Uncanny Valley, which is about whether tech can get rid of bad drug chips. Adriana Tapia and Mark Leyda produced this episode, Amar Lal at Macrosound mixed this episode. Kate Osborn is our executive producer. Chris Bannon is Condé Nast's, head of global audio, and Katie Drummond is WIRED's global editorial director.