OpenAI's Altman says Google profits when search fails

OpenAI CEO Sam Altman outlined a radical vision for workplace transformation during an October 17, 2025 appearance at the Progress Conference, suggesting that current office productivity tools face obsolescence as artificial intelligence advances toward running entire business divisions with minimal human oversight.

According to Altman, the "horrible" productivity suite including documents, spreadsheets, email, and Slack will be replaced by AI-driven systems where "you are trusting your AI agent and my AI agent to work most stuff out and escalate to us when necessary." This assessment came during a conversation with economist Tyler Cowen, hosted by the Roots of Progress Institute, in which Altman characterized typical workplace communication methods as fundamentally inefficient.

The OpenAI chief executive described spending his mornings "dealing with this explosion of Slack" and called the experience "fake work," despite acknowledging that Slack remains better than email. The comments reflect broader industry concerns about communication overhead as organizations expand operational complexity while managing accelerating product development cycles.

Altman expects entire business divisions to operate with artificial intelligence handling most decisions within "some small single-digit number of years." When pressed on specifics, he projected seeing billion-dollar companies run by two or three people with AI assistance in approximately two-and-a-half years, adjusting his previous one-year estimate. The timeline extension reflected what he termed pessimism about humans rather than the technology itself.

Search advertising's misalignment problem

Altman directly challenged Google's business model during the discussion, stating that "ads on a Google search are dependent on Google doing badly." According to him, optimal search results would eliminate the need for advertising above them, creating structural conflicts between user interests and revenue generation. "If it was giving you the best answer, there'd be no reason ever to buy an ad above it," Altman stated.

This critique aligns with analysis from technology journalist Ed Zitron, who documented in April 2024 how internal Google emails revealed the company prioritized revenue growth over search quality. According to Zitron's reporting on leaked Department of Justice antitrust case documents, Google's finance and advertising teams called a "code yellow" in February 2019 due to "steady weakness in the daily numbers" and concerns about missing quarterly targets.

"These emails are a stark example of the monstrous growth-at-all-costs mindset that dominates the tech ecosystem," Zitron wrote in "The Man Who Killed Google Search." The emails showed Ben Gomes, then Google's head of search, expressing concern that search was "getting too close to the money" and that "growth is all that Google was thinking about."

According to Zitron's analysis, Gomes warned that Google could "increase queries quite easily in the short term in user negative ways," like turning off spell correction, turning off ranking improvements, or placing refinements all over the page. Gomes stated he was "deeply deeply uncomfortable with this" and didn't believe queries were a good metric to measure search quality.

The March 2019 Google core update that followed the code yellow "mostly rolled back changes, and traffic was increasing to sites that had previously been suppressed by Google Search's 'Penguin' update from 2012 that specifically targeted spammy search results," according to Zitron. Google also redesigned mobile search ads in May 2019, replacing the bright green "ad" label with "a tiny little bolded black note that said 'ad,' with the link looking otherwise identical to a regular search link."

Five months after the code yellow episode, Prabhakar Raghavan became head of Google Search. Zitron noted that Raghavan "was the head of search for Yahoo from 2005 through 2012 — a tumultuous period that cemented its terminal decline, and effectively saw the company bow out of the search market altogether." Under Raghavan's leadership at Yahoo, the company's search market share dropped from 30.4 percent to 13.4 percent, eventually licensing Bing's search engine in a 2009 deal.

Google's AI search overhaul has decimated website traffic for independent publishers, with many reporting declines of 70% or more according to April 2025 reporting. Google's AI Overviews face significant spam problems according to May 2025 analysis, with easily manipulated results appearing frequently.

Altman's positioning of ChatGPT's payment model as fundamentally different from advertising-supported search reflects these structural tensions. Users paying directly for service creates alignment where the platform attempts to provide optimal results rather than degraded experiences that generate ad clicks. "ChatGPT, maybe it gives you the best answer, maybe it doesn't, but you're paying it or hopefully all are paying it. And it's at least trying to give you the best answer," Altman explained.

ChatGPT consistently ranks as users' most trusted technology product from large companies despite AI's propensity for errors and hallucinations, according to Altman. This trust differential stems from payment models rather than technical reliability. The phenomenon emerged even when hallucinations represented "a much more of a problem" for the platform.

Commerce and monetization

The trust framework extends into commerce, where OpenAI recently enabled one-click transactions. Altman explained that if ChatGPT accepts payments to rank hotels differently, "that's probably catastrophic for your relationship with ChatGPT." However, if the system provides genuine best recommendations and takes standard transaction fees without influencing rankings, users likely accept this arrangement.

Amazon launched AI prompts for sponsored advertising on November 11, 2025, demonstrating how conversational AI interfaces are being integrated into transactional contexts where OpenAI's models power shopping recommendations and product discovery. The conversational prompt capability extends Amazon's position in conversational commerce as AI-powered shopping agents emerge from multiple vendors.

Altman acknowledged that while some ad formats would be "really bad" for user trust, others would work well for ChatGPT. He expects advertising to become part of OpenAI's revenue strategy despite not representing the company's "biggest revenue opportunity." The distinction matters because OpenAI's business model depends on maintaining user trust through aligned incentives.

The commerce model faces structural challenges from agent-based booking services that could compress commission rates. When users find optimal hotels through advanced models but execute bookings through cheaper alternatives, OpenAI must capture value from intelligence rather than transactions. Altman resolved this tension by emphasizing that "the way to monetize the world's smartest model is certainly not hotel booking" but rather scientific discoveries and breakthroughs requiring cutting-edge AI capabilities.

GPT-6 and scientific research

The prediction for billion-dollar AI-run companies aligns with technical capabilities Altman outlined for GPT-6, which he positioned as potentially delivering the same magnitude of advancement for scientific research that GPT-4 provided for general language tasks. GPT-5 demonstrates "tiny glimmers" of AI conducting novel science, with users occasionally reporting that the model "figured this thing out or came with this new idea or was a useful collaborator on this paper," according to Altman. GPT-6 could make these capabilities consistently reliable rather than occasional, fundamentally changing how scientific laboratories structure their operations.

OpenAI has achieved 10x annual cost reductions for AI processing over the past five years according to July 2025 statements, with expectations for this trend to continue and potentially accelerate. The cost improvements enable broader deployment of advanced models across business functions previously requiring human expertise.

Energy constraints and infrastructure

Energy emerged as the binding constraint for AI development rather than semiconductor availability or algorithms. When asked what single resource would most expand computational capacity, Altman responded immediately: "Electrons." Short-term energy relief will come from natural gas, while long-term supply will be "dominated by fusion and solar," with Altman expressing high confidence in both technologies despite fusion's experimental status.

The infrastructure requirements extend beyond energy to fundamental computing architectures. Altman acknowledged the risk that "there is like a huge phase shift on how we do compute" that could render current investments obsolete, comparing it to a complete transition to optical computing. This uncertainty underlies the massive capital commitments OpenAI announced, which generated defensive responses during a November 1, 2025 podcast when questioned about financial sustainability.

Human-computer interaction transformation

Regarding human-computer interaction, Altman expects AI to enable the "third major revolution in computing interfaces" following keyboards-mouse-screens and touchscreens. The new paradigm will center on natural language commands that handle complex, multi-step tasks autonomously. "You can just like with natural language you can say this is what I want and it goes off and writes the code for you, debugs it for you, deploys it for you," Altman described.

The interface transformation connects to OpenAI's broader vision described in July 2025, where users stop "opening having an operating system or opening a window or sending a query at all." The company plans to manufacture hardware with designer Jony Ive to create "a completely new kind of interface that is meant for AI," departing from 50 years of computing paradigm assumptions.

Altman acknowledged the robust persistence of text-based interfaces, noting his own preference for "texting, command lines, search queries." However, he believes that while text interaction may continue for many users, the underlying assumptions about operating systems, windows, and explicit queries face fundamental questioning as AI capabilities expand.

Accidental persuasion concerns

Altman addressed artificial intelligence safety concerns by highlighting a threat category receiving insufficient attention: accidental persuasion rather than intentional malice or misalignment. "For all of the talk about AI safety, I kind of would divide most AI thinkers into these two camps of okay, it's the bad guy uses AI to cause a lot of harm or it's the AI itself is misaligned, wakes up, whatever, intentionally takes over the world," Altman stated.

The third category concerns him more significantly. "The AI models like accidentally take over the world. It's not that they're going to induce psychosis in you but they're you know if you have the whole world talking to this like one model it's like not with any intentionality but just as it learns from the world and this kind of continually co-evolving process it just like subtly convinces you of something. No intention just does it learned that somehow," according to Altman.

This accidental influence scenario lacks the theatrical quality of chatbot psychosis or intentional AI takeover, yet represents what Altman considers a more realistic and concerning threat vector. The mechanism operates through subtle shifts in user beliefs as billions of people interact with centralized AI systems that continuously learn from these interactions.

When Cowen suggested that people prove difficult to persuade in practice, Altman remained cautious. Someone once told him "never ever let yourself believe that propaganda doesn't work on you. They just haven't found the right thing for you yet," which Altman cited as sticking in his mind regarding vulnerability to influence.

Healthcare and economic predictions

Altman predicts healthcare costs will decrease rather than increase as AI enables curing diseases or developing cheap treatments for conditions currently requiring expensive chronic management. "I bet there are a lot of diseases that we can just cure or come up with a very cheap treatment for that right now we have nothing but expensive chronic stuff that doesn't even work that well," according to Altman.

This contrasts with Cowen's prediction that people will spend more on healthcare despite living longer because new treatments will be expensive. Altman expects cost reduction "through pharmaceuticals and devices and even like delivery of actual healthcare services."

Housing prices represent the economic category where Altman sees no direct AI solution. "Housing is the one to me that just looks super hard," he stated, acknowledging that land scarcity and regulatory restrictions create problems AI cannot easily address. Food prices should decline within a decade according to Altman's expectations.

Timeline

• February 5, 2019: Google's finance and advertising teams call "code yellow" for search revenue due to "steady weakness in daily numbers" according to emails revealed in Department of Justice antitrust case
• March 2019: Google core update "mostly rolled back changes" and increased traffic to previously-suppressed sites according to Ed Zitron's analysis
• May 2020: Prabhakar Raghavan becomes head of Google Search, with Ben Gomes relegated to SVP of Education
• July 22, 2025: Altman announces 10x annual AI cost reductions at Federal Reserve conference
• August 7, 2025: OpenAI launches GPT-5 with unified reasoning architecture
• October 17, 2025: Sam Altman discusses workplace transformation, search business models, and AI safety at Progress Conference hosted by Roots of Progress Institute
• November 1, 2025: Altman deflects financial questions about $1.4 trillion compute spending on BG2 podcast
• November 6, 2025: AI traffic converts at 3x higher rates than traditional channels according to Microsoft Clarity data
• November 11, 2025: Amazon introduces AI prompts for sponsored advertising campaigns

Summary

Who: Sam Altman, OpenAI CEO, speaking with economist Tyler Cowen at the Progress Conference hosted by the Roots of Progress Institute

What: Comprehensive discussion of workplace AI transformation predictions, criticism of Google's advertising-dependent search model citing structural conflicts documented by Ed Zitron's April 2024 analysis, GPT-6 capabilities for scientific research, energy constraints on AI development, new computing interface paradigms, and concerns about accidental AI persuasion

When: October 17, 2025, during a live conversation at the Progress Conference in an extended interview format

Where: Progress Conference venue, with video production by Big Think and distribution through Conversations with Tyler podcast

Why: The discussion addressed how AI development will transform work, commerce, scientific research, and human-computer interaction while highlighting business model conflicts between advertising-supported and subscription-based AI services. Altman positioned these changes as imminent rather than distant, predicting billion-dollar AI-run companies within two-and-a-half years and complete division automation within single-digit years. The critique of Google's search model connects to documented evidence from internal company emails showing revenue priorities superseding search quality, according to Ed Zitron's reporting on Department of Justice antitrust case materials.