Full Breakdown
DeepMind Paper Asserts AI Will Never Be Conscious Amid Ongoing AGI Debate
4/27/2026, 9:56:47 PM
DeepMind Paper Asserts AI Will Never Be Conscious
A senior staff scientist at Google DeepMind, Alexander Lerchner, released a paper titled *The Abstraction Fallacy: Why AI Can Simulate But Not Instantiate Consciousness*. The work argues that any artificial system, including large language models, lacks intrinsic meaning and cannot achieve consciousness because it depends on human agents to “alphabetize continuous physics into a finite set of meaningful states.” Lerchner concludes that even a highly capable artificial general intelligence (AGI) would remain a non-sentient tool.
Context: DeepMind’s Public AGI Narrative vs. Academic Skepticism
DeepMind CEO Demis Hassabis has repeatedly described AGI as “10 times the impact of the Industrial Revolution, but happening at 10 times the speed.” Lerchner’s paper directly challenges that narrative, suggesting a hard cap on what AI can accomplish without consciousness. The claim revives arguments that have circulated in philosophy and cognitive science for decades.
Key Contributors and Their Perspectives
- Alexander Lerchner – Senior staff scientist, author of the paper.
- Demis Hassabis – DeepMind CEO, proponent of transformative AGI.
- Johannes Jäger – Evolutionary systems biologist and philosopher, critic of Lerchner’s scholarship.
- Mark Bishop – Professor of cognitive computing, Goldsmiths, University of London, notes the paper’s arguments are long-standing.
- Emily Bender – Professor of linguistics, University of Washington, comments on the paper’s lack of peer review.
- Henry Shevlin – Cambridge academic appointed as DeepMind’s first “Philosopher” (May 2026) to study machine consciousness and AGI readiness.
Timeline of Relevant Events
- March 10 – Lerchner’s paper posted on DeepMind’s site.
- April 20 – Request for comment sent to Google; no response received.
- Post-April 2024 – PDF reissued without DeepMind letterhead; disclaimer moved to the top.
- May 2026 – DeepMind will begin employing Henry Shevlin as a philosopher.
- 2024-2025 – DeepMind hires for a “post-AGI” research scientist position.
Official Statements & Responses
DeepMind’s disclaimer states the paper reflects Lerchner’s personal conclusions and “does not necessarily reflect the official stance, views, or strategic policies of his employer.” Google declined to comment on the matter. Hassabis continues to predict massive societal impact from AGI, while DeepMind’s hiring for post-AGI roles signals ongoing investment in advanced AI research despite the paper’s cautionary stance.
Criticism & Opposition
Jäger argues Lerchner “reinvented the wheel” and is “not well read” in philosophy or biology. Bishop concurs that the arguments “have been presented years and years ago.” Bender notes that corporate-lab papers often bypass traditional peer-review channels. Jäger also highlights a broader “insular” culture within AI research that neglects the biological origins of concepts like “agency” and “intelligence.”
Conflicting Reports & Gaps
The paper’s claim that AI cannot be conscious conflicts with DeepMind’s public AGI optimism and its recruitment of talent for post-AGI work. The removal of DeepMind branding from the PDF and the absence of a formal response from Google leave the institutional position ambiguous. Additionally, the paper’s limited citation of existing literature creates a gap in scholarly grounding.
Verbatim Quotes
- “An LLM doesn't do that. It's just a bunch of patterns on a hard drive. Then it gets prompted and it runs until the task is finished and then it's done. So it doesn't have any intrinsic meaning. Its meaning comes from the way that some human agent externally has defined a meaning.” — Alexander Lerchner
- “I think he [Lerchner] arrived at this conclusion on his own and he's reinvented the wheel and he's not well read, especially in philosophical areas and definitely not in biology,” — Johannes Jäger, Evolutionary Systems Biologist & Philosopher
- “I'm in sympathy with 99 percent of everything that he [Lerchner] says,” — Mark Bishop, Professor of Cognitive Computing, Goldsmiths
- “the development of highly capable Artificial General Intelligence (AGI) does not inherently lead to the creation of a novel moral patient, but rather to the refinement of a highly sophisticated, non-sentient tool.” — Alexander Lerchner (paper)
- “We can imagine many financial and legislative reasons why Google would be sanguine with a conclusion that says computations can't be consciousness,” — Mark Bishop
- “Much of what's happening in this research space right now is you get these paper-shaped objects coming out of the corporate labs,” but that do not go through a proper scientific paper publishing process.” — Emily Bender, Professor of Linguistics, University of Washington
What Lies Ahead
Starting May 2026, Henry Shevlin will explore machine consciousness and AGI readiness within DeepMind, potentially bridging the gap between corporate ambition and philosophical caution. The debate over AI’s capacity for consciousness is expected to continue as DeepMind advances its research agenda.
