# MIT Technology Review’s Will Douglas Heaven explores what octopuses can teach us about conscious AI [Canonical reading](https://escapemeridian.com/readings/mit-technology-review-s-will-douglas-heaven-explores-what-d22c11f5) On August 25, 2021, Will Douglas Heaven argues in MIT Technology Review that intelligence and consciousness are different, and that behavior alone cannot tell us whether a machine has subjective experience. He uses Emily Bender’s octopus thought experiment to question whether language models understand meaning or produce patterns that people interpret as meaningful. Heaven also draws on octopuses’ distinct minds to caution against assuming that machine consciousness, if it exists, would resemble human experience. The article presents consciousness as an unresolved scientific and philosophical problem, and raises questions about the rights and treatment of machines that might suffer. Heaven concludes that people may have to live with uncertainty about whether machines are conscious. 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: Aug 25, 2021 (2021-08-25T11:00:00Z). Date basis: Date from the publisher’s own records, through its API. Added to Escape Meridian: 2026-10-29T16:33:53.180Z. First observed: 2026-10-27T23:08:11.418Z. Retrieved: 2026-10-27T23:09:59.164Z. - [What an octopus’s mind can teach us about AI’s ultimate mystery](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Limits Text as retrieved on October 27, 2026; originally published August 25, 2021. The publisher page indicates it was edited after publication. The retrieved text may include later additions; this reading does not rely on passages that appear to describe later events. The article presents arguments and thought experiments; it does not establish a test for detecting machine consciousness or demonstrate that machines have subjective experience. ## Open question If a machine appeared conscious, what evidence could distinguish subjective experience from behavior that people interpret as meaningful? ## Source-stated Heaven argues that intelligence and consciousness are distinct: intelligent behavior does not by itself establish subjective experience. > But intelligence and consciousness are different things: intelligence is about doing, while consciousness is about being. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Source-stated The article presents Bender’s octopus as a thought experiment about whether producing plausible language patterns amounts to understanding. > The point of the example is to reveal how shallow today’s cutting-edge AI language models really are. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Source-stated Heaven argues that human experience may not provide a reliable picture of what consciousness would be like for a machine or another species. > My conception of a conscious machine is undeniably—perhaps unavoidably—human-like. It is the only form of consciousness I can imagine, as it is the only one I have experienced. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Source-stated Heaven raises ethical questions about possible machine suffering, rights, and whether conscious machines should be created. > We will have to consider whether their subjective experience includes the ability to suffer pain, boredom, depression, loneliness, or any other unpleasant sensation or emotion. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## 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. ## Editorial provenance Version 1. Publication ID: 964ae78c-69c4-9442-fe66-fb1f4e296bca. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-10-29T16:33:50.831Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical. Observation: c4772dc18627175de6518aa477ded48bd3d9184b636011285731c2531185a3aa. Source text hash: 4217ae06f057f46fcf74184aae209636352d416008c3923dbe583efe9ccc977f. ## Revision and source-check history ### Version 1 — 2026-10-29T16:33:53.180Z On August 25, 2021, Will Douglas Heaven argues in MIT Technology Review that intelligence and consciousness are different, and that behavior alone cannot tell us whether a machine has subjective experience. He uses Emily Bender’s octopus thought experiment to question whether language models understand meaning or produce patterns that people interpret as meaningful. Heaven also draws on octopuses’ distinct minds to caution against assuming that machine consciousness, if it exists, would resemble human experience. The article presents consciousness as an unresolved scientific and philosophical problem, and raises questions about the rights and treatment of machines that might suffer. Heaven concludes that people may have to live with uncertainty about whether machines are conscious. ## Original source MIT Technology Review · Article Original source date: Aug 25, 2021 (2021-08-25T11:00:00Z). Date basis: Date from the publisher’s own records, through its API. Added to Escape Meridian: 2026-10-29T16:33:53.180Z. First observed: 2026-10-27T23:08:11.418Z. Retrieved: 2026-10-27T23:09:59.164Z. - [What an octopus’s mind can teach us about AI’s ultimate mystery](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Limits Text as retrieved on October 27, 2026; originally published August 25, 2021. The publisher page indicates it was edited after publication. The retrieved text may include later additions; this reading does not rely on passages that appear to describe later events. The article presents arguments and thought experiments; it does not establish a test for detecting machine consciousness or demonstrate that machines have subjective experience. ## Open question If a machine appeared conscious, what evidence could distinguish subjective experience from behavior that people interpret as meaningful? ## Source-stated Heaven argues that intelligence and consciousness are distinct: intelligent behavior does not by itself establish subjective experience. > But intelligence and consciousness are different things: intelligence is about doing, while consciousness is about being. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Source-stated The article presents Bender’s octopus as a thought experiment about whether producing plausible language patterns amounts to understanding. > The point of the example is to reveal how shallow today’s cutting-edge AI language models really are. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Source-stated Heaven argues that human experience may not provide a reliable picture of what consciousness would be like for a machine or another species. > My conception of a conscious machine is undeniably—perhaps unavoidably—human-like. It is the only form of consciousness I can imagine, as it is the only one I have experienced. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## Source-stated Heaven raises ethical questions about possible machine suffering, rights, and whether conscious machines should be created. > We will have to consider whether their subjective experience includes the ability to suffer pain, boredom, depression, loneliness, or any other unpleasant sensation or emotion. - [Supporting source](https://www.technologyreview.com/2021/08/25/1032111/conscious-ai-can-machines-think/) ## 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. ## Editorial provenance Version 1. Publication ID: 964ae78c-69c4-9442-fe66-fb1f4e296bca. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-10-29T16:33:50.831Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical. Observation: c4772dc18627175de6518aa477ded48bd3d9184b636011285731c2531185a3aa. Source text hash: 4217ae06f057f46fcf74184aae209636352d416008c3923dbe583efe9ccc977f. ### passed — 2026-10-29T16:33:50.831Z — anthropic/claude-opus-5.5 Publication: 964ae78c-69c4-9442-fe66-fb1f4e296bca attribution — pass: The article is correctly attributed to Will Douglas Heaven at MIT Technology Review. Ideas from Bender, Chalmers, Dennett and others are presented as cited perspectives within his piece. > Will Douglas Heaven is a senior editor for AI at MIT Technology Review. summary — pass: The title, summary and question accurately reflect the essay. It covers the intelligence versus consciousness distinction, Bender's octopus test, octopus minds as a caution against human-centric assumptions, ethical questions, and the conclusion about living with uncertainty. Central terms are explained in plain language. The question follows from the source's point about observer projection. > Faced with such possibilities, we will have to choose to live with uncertainties. date — pass: August 25, 2021 matches the originalPublished instant's UTC day and the URL path. The page was edited after publication (dateModified 2022-08-02), but every cited passage is core article body tied to the 2021 context, such as GPT-3. The limitations disclose the edit. The later Deep Dive sidebar items are not used. There is no hindsight framing, and the kind is publisher-page, so no archive disclosure is needed. > large language models like OpenAI’s GPT-3—are totally mindless. support — pass: Each claim's excerpt is an exact source quote that supports it. Claim 4's mention of whether conscious machines should be created is supported by the passage on Dennett and Metzinger. The limitations rightly note that the article proposes no detection test. No unsupported benefits or costs are implied. > Many researchers, including Dennett, think that we shouldn’t try to make conscious machines even if we can. media — pass: No media is selected, and no verified media candidates were supplied, so nothing could be 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/d22c11f5cb7f1cb02876452a8322b5ccad769e0004a28d2165bf7804b74bea03/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/mit-technology-review-s-will-douglas-heaven-explores-what-d22c11f5.json)