# Karen Hao argues in MIT Technology Review that AI’s development reflects human choices [Canonical reading](https://escapemeridian.com/readings/karen-hao-argues-in-mit-technology-review-that-ai-s-development-a31e1021) Published July 9, 2025, MIT Technology Review presents a conversation in which AI journalist and author Karen Hao argues that the AI industry functions like an empire. She connects OpenAI’s choice to advance AI by scaling existing techniques—using more data and larger supercomputers—to competition with Google and Sam Altman’s ability to raise capital. Hao says this path produced technical progress, but carries substantial costs, including energy and freshwater use, pollution, exploited labor and entrenched inequities. She argues that the idea of artificial general intelligence (AGI) can justify continued consumption and concentrated control, despite no shared agreement on what AGI is or whether it is possible. Her alternative is not to abandon AI: she advocates task-specific systems, community consent and greater public accountability. Hao points to a Māori community’s speech-recognition project as an example of community involvement in deciding whether to build a tool and how to use its data. She ends by urging people to turn public attention into action and says regulation should increase transparency across AI’s supply chain. The conversation does not establish how widespread the described harms are or whether her proposed alternatives can deliver comparable benefits. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source MIT Technology Review · Article Original source date: Jul 9, 2025 (2025-07-09T09:10:29Z). Date basis: Date from the publisher’s own records, through its API. Added to Escape Meridian: 2026-11-13T04:33:23.334Z. First observed: 2026-10-31T06:39:57.158Z. Retrieved: 2026-10-31T06:41:44.534Z. - [Inside OpenAI’s empire: A conversation with Karen Hao](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Limits Text as retrieved on October 31, 2026; originally published July 9, 2025. MIT Technology Review says the transcript was lightly edited and condensed. The conversation reports Hao’s arguments; it does not independently establish the extent of the harms she describes. The conversation does not provide evidence on whether task-specific AI can deliver comparable benefits with lower costs. ## Open question What evidence would show whether task-specific AI can deliver the benefits Hao describes with lower social and environmental costs? ## Source-stated Karen Hao argues that OpenAI chose to pursue AI progress by scaling existing techniques, a choice she links to competition with Google and the team’s ability to raise capital. > OpenAI was anomalously committed to the other extreme—this idea that we can just take neural networks and pump ever more data, and train on ever larger supercomputers - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao argues that this development path has substantial costs, including environmental impacts and exploitation of labor and resources. > there’s also an extraordinary amount of costs that come with this particular pathway to getting more advancements. - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao argues that AGI lacks a shared meaning and can serve as a justification for resource consumption and concentrated control. > when there is a vacuum of shared meaning around this term, and what it would look like, when would we have arrived at it? - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao advocates task-specific AI systems and describes a Māori community project in which community members were asked for consent and involved in decisions about data use. > the first thing that they did when engaging in that project was they asked the community, do you want this AI tool? - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao says regulation should increase transparency across the AI supply chain. > it should be to dramatically increase the amount of transparency along the supply chain. - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source and contributor context MIT Technology Review · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Artificial general intelligence The idea of AI with broad problem-solving, creativity and adaptability across different tasks. - [Definition reference](https://developers.google.com/machine-learning/glossary#artificial-general-intelligence) ## OpenAI OpenAI develops AI models and products and publishes research, release announcements and evaluations. - [Explore this context](/entities/openai) - [Reference](https://openai.com/news/) ## Sam Altman Sam Altman publishes essays about AI and technology on his personal blog, separate from OpenAI announcements. - [Explore this context](/entities/sam-altman) - [Reference](https://blog.samaltman.com/) - [Reference](https://openai.com/our-structure/) - [Reference](https://blog.samaltman.com/posts.atom) ## Editorial provenance Version 1. Publication ID: 79cbdbf1-b521-bb8d-6f76-70ff42f29350. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-13T04:29:09.890Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: f5b7c3beb8453bdeb1ff55b927ee3e9a759104c73672e9fe2c721f3a7630156d. Source text hash: 58739fe07a3ef25ef9fe501030b4b0f51a19faf5367b87a46929b24ed19393b7. ## Revision and source-check history ### Version 1 — 2026-11-13T04:33:23.334Z Published July 9, 2025, MIT Technology Review presents a conversation in which AI journalist and author Karen Hao argues that the AI industry functions like an empire. She connects OpenAI’s choice to advance AI by scaling existing techniques—using more data and larger supercomputers—to competition with Google and Sam Altman’s ability to raise capital. Hao says this path produced technical progress, but carries substantial costs, including energy and freshwater use, pollution, exploited labor and entrenched inequities. She argues that the idea of artificial general intelligence (AGI) can justify continued consumption and concentrated control, despite no shared agreement on what AGI is or whether it is possible. Her alternative is not to abandon AI: she advocates task-specific systems, community consent and greater public accountability. Hao points to a Māori community’s speech-recognition project as an example of community involvement in deciding whether to build a tool and how to use its data. She ends by urging people to turn public attention into action and says regulation should increase transparency across AI’s supply chain. The conversation does not establish how widespread the described harms are or whether her proposed alternatives can deliver comparable benefits. ## Original source MIT Technology Review · Article Original source date: Jul 9, 2025 (2025-07-09T09:10:29Z). Date basis: Date from the publisher’s own records, through its API. Added to Escape Meridian: 2026-11-13T04:33:23.334Z. First observed: 2026-10-31T06:39:57.158Z. Retrieved: 2026-10-31T06:41:44.534Z. - [Inside OpenAI’s empire: A conversation with Karen Hao](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Limits Text as retrieved on October 31, 2026; originally published July 9, 2025. MIT Technology Review says the transcript was lightly edited and condensed. The conversation reports Hao’s arguments; it does not independently establish the extent of the harms she describes. The conversation does not provide evidence on whether task-specific AI can deliver comparable benefits with lower costs. ## Open question What evidence would show whether task-specific AI can deliver the benefits Hao describes with lower social and environmental costs? ## Source-stated Karen Hao argues that OpenAI chose to pursue AI progress by scaling existing techniques, a choice she links to competition with Google and the team’s ability to raise capital. > OpenAI was anomalously committed to the other extreme—this idea that we can just take neural networks and pump ever more data, and train on ever larger supercomputers - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao argues that this development path has substantial costs, including environmental impacts and exploitation of labor and resources. > there’s also an extraordinary amount of costs that come with this particular pathway to getting more advancements. - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao argues that AGI lacks a shared meaning and can serve as a justification for resource consumption and concentrated control. > when there is a vacuum of shared meaning around this term, and what it would look like, when would we have arrived at it? - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao advocates task-specific AI systems and describes a Māori community project in which community members were asked for consent and involved in decisions about data use. > the first thing that they did when engaging in that project was they asked the community, do you want this AI tool? - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source-stated Hao says regulation should increase transparency across the AI supply chain. > it should be to dramatically increase the amount of transparency along the supply chain. - [Supporting source](https://www.technologyreview.com/2025/07/09/1119784/inside-openais-empire-a-conversation-with-karen-hao/) ## Source and contributor context MIT Technology Review · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Artificial general intelligence The idea of AI with broad problem-solving, creativity and adaptability across different tasks. - [Definition reference](https://developers.google.com/machine-learning/glossary#artificial-general-intelligence) ## OpenAI OpenAI develops AI models and products and publishes research, release announcements and evaluations. - [Explore this context](/entities/openai) - [Reference](https://openai.com/news/) ## Sam Altman Sam Altman publishes essays about AI and technology on his personal blog, separate from OpenAI announcements. - [Explore this context](/entities/sam-altman) - [Reference](https://blog.samaltman.com/) - [Reference](https://openai.com/our-structure/) - [Reference](https://blog.samaltman.com/posts.atom) ## Editorial provenance Version 1. Publication ID: 79cbdbf1-b521-bb8d-6f76-70ff42f29350. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-13T04:29:09.890Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: f5b7c3beb8453bdeb1ff55b927ee3e9a759104c73672e9fe2c721f3a7630156d. Source text hash: 58739fe07a3ef25ef9fe501030b4b0f51a19faf5367b87a46929b24ed19393b7. ### passed — 2026-11-13T04:29:09.890Z — anthropic/claude-opus-5.5 Publication: 79cbdbf1-b521-bb8d-6f76-70ff42f29350 attribution — pass: The reading attributes the arguments to Karen Hao, speaking in an MIT Technology Review Roundtables conversation about her book. Her claims are framed as her arguments rather than established facts. The 'human choices' framing was raised by Niall Firth, but Hao explicitly endorses it. > So that is ultimately how you can see that technology is a product of human choices and human perspectives. summary — pass: Accurately conveys Hao's main arguments: the empire framing, OpenAI's bet on scaling, its costs, AGI as a shifting justification, and her alternatives (task-specific AI, community consent, accountability, transparency). Scaling is explained in plain language, and AGI is defined on the page. One small imprecision: 'She ends by' is slightly off, because the conversation actually closes with her blunt question to Altman. This does not distort meaning. The question follows from the stated limitation. > What degree of harm do you need to see in order to realize that you should take a different path? date — pass: The originalPublished instant of 2025-07-09 matches the stated July 9, 2025 date at day precision. The page modification came one second after publication, so it is not a substantive later edit. Sidebar items such as the AlphaGo ten-year piece are later additions to the page, but the reading does not use them. There is no hindsight framing, and this is a publisher page, so no archive disclosure is needed. support — pass: Each claim is source-stated with matching excerpts, covering scaling, Google competition, Altman's fundraising, costs, the AGI vacuum, task-specific AI, the Māori project and transparency. The limitations appropriately note that the harms and the alternatives' benefits are not independently established. No unsupported benefits or costs are implied. > So if there were one major intervention that regulators could have, it should be to dramatically increase the amount of transparency along the supply chain. media — pass: No media is selected and no verified media candidates were supplied, so nothing is misidentified. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/a31e102132bac2254c553379cdb4d7062a94e69c4f0db6a9861fadbf94a01beb/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/karen-hao-argues-in-mit-technology-review-that-ai-s-development-a31e1021.json)