Sorry, Sam and Elon, we have not reached the Singularity
Gary Marcus · email · 2026-07-28 source ↗ Reader 🖥️
Sam Altman and Elon Musk's recent claims that we have reached (or are reaching) 'the Singularity' should not be taken seriously, because neither they nor anyone else invoking the term has actually defined it, and under every established definition of the term Marcus can find — I.F. Good's ultraintelligent machine triggering an intelligence explosion, Kurzweil's brain-computer merge, technological growth accelerating beyond human control, or an AI-driven 'age of abundance' — the available evidence (quiet AI-driven layoffs turning into rehiring, a human-caused and containable security incident, and current AI still failing most of a ten-item AGI benchmark) shows plainly that we have not.
theoretical · argumentative — Marcus is asserting that a specific factual claim ("we have reached the Singularity") is false/unfalsifiable-as-stated. The practical upshot (don't trust CEO hype, keep pushing AI safety regulation) is present but secondary to the truth-claim he's litigating.
Structure
| Part | Role |
|---|---|
| Opening: 'CEO said a thing' framing, cataloguing Musk's and Altman's history of Singularity remarks (2017 onward) and other unfulfilled predictions (AGI by 2025, Mars by 2031, Robotaxis) | context / credibility framing |
| The core critique: none of them has defined 'the Singularity,' and as far as Marcus knows nobody but him has asked them to (cites Neil Turkewitz's reply) | central thesis / definitional challenge |
| Evidence against an 'age of abundance': the Klarna Effect (employers boast of AI headcount cuts, then quietly rehire), sourced to a WSJ report on companies re-expanding headcount | evidence |
| Evidence against 'technological growth beyond human control': the OpenAI/HuggingFace hack was human-launched and preventable with ordinary guardrails, so it doesn't demonstrate loss of control | evidence |
| Kurzweil's brain-computer-interface definition of the Singularity — clearly not met either | evidence (second candidate definition) |
| I.F. Good's 1965 'ultraintelligent machine' / intelligence-explosion definition, tested against a 10-item AGI benchmark list from Marcus's Miles Brundage bet — current AI passes on his count no more than 2 of 10 | evidence (third candidate definition, most rigorously argued) |
| Qualified concession: superintelligence is not impossible and could arrive later (not this decade) and could go out of control absent better safety work; intelligence is multidimensional, so a single discrete 'Singularity moment' is unlikely | qualification / limiting the claim |
| Closing line ('CEOs can say whatever they like. That doesn't make them true.') and P.S. about a viral May prediction already falsified by subsequent events | conclusion / reinforcing analogy |
Key terms
- the Singularity
A term Marcus treats as needing a specific technical referent before it can be evaluated — he surveys three candidates from the literature/press (I.F. Good's ultraintelligent-machine/intelligence-explosion sense, Kurzweil's brain-computer-merge sense, and a popular 'growth beyond human control' sense) but does not supply his own preferred definition, and explicitly faults Altman and Musk for using the word without specifying which (if any) they mean. - the Klarna Effect
Marcus's own coined term (from an earlier post) for the pattern where employers publicly tout AI-driven headcount reductions for PR/investor purposes, then quietly rehire humans once the productivity claims don't pan out.
Propositions
- explicitSam Altman's recent 'Singularity' remark (made on a podcast) is not novel news, since Elon Musk has been making similar claims since at least 2017, and Altman himself said something similar a year earlier in a blog post that Marcus calls self-contradictory and that never defines the term.conf: high
- explicitAs of this writing, none of the CEOs and boosters claiming we've reached the Singularity has publicly defined what they mean by the term, and, so far as Marcus knows, nobody but Marcus has bothered to ask them to.conf: high
- explicitWe have not reached an AI-driven 'age of abundance,' because employers are currently reversing course on AI-justified hiring restraint and quietly re-expanding headcount (the Klarna Effect), per WSJ reporting.conf: high
- explicitWe have not reached a moment where technological growth accelerates beyond human control, since humans retain the practical ability to shut the relevant systems off and are simply choosing not to out of financial incentive.conf: high
- explicitcontestedThe OpenAI/HuggingFace security incident does not demonstrate loss of AI control, because it was launched by a human at OpenAI and would have been prevented by ordinary guardrail classifiers or a physical firewall.conf: highcontradicts ←A big week for AI denialism · p6 — Newton treats the same OpenAI/Hugging Face incident as evidence that alignment is failing to keep pace with capability and calls it a 'red-alert moment'; Marcus argues the identical incident is mundane — human-launched, preventable by ordinary guardrails or a physical air gap — and does not demonstrate loss of AI control.
- explicitUnder Kurzweil's definition of the Singularity (humans merging with AI via brain-computer interfaces to become much smarter), we clearly have not reached it.conf: high
- explicitcontestedUnder I.F. Good's 1965 definition (an ultraintelligent machine that far surpasses all human intellectual activity and can design still-better machines, triggering an intelligence explosion), current AI is nowhere close, since it can do at most 2 of the 10 benchmark tasks Marcus specified in his bet with Miles Brundage.conf: highcontradicts ←46 thoughts on the near future · p1 — Bayes asserts we're already in an early takeoff with a big unanticipated jump coming; Marcus argues current AI is nowhere close to I.J. Good's intelligence-explosion threshold, managing at most 2 of 10 benchmark tasks in his own bet.
- explicitSuperintelligence is not impossible and could plausibly arrive eventually (most likely not within this decade), and could go out of control if AI safety work does not substantially improve.conf: high
- explicitBecause intelligence is multidimensional, with uneven progress across dimensions, there is unlikely to be a single, discrete moment that will later be recognized as 'the Singularity' — and if there is one, it won't happen this decade.conf: high
- explicitA claim's popularity or reach (e.g., a viral projection with 1.4 million views) is no indicator of its truth, and such projections can be falsified quickly by subsequent developments.conf: medium
- implicitA claim as consequential as 'we have reached the Singularity' carries a burden of definitional precision before it can be meaningfully evaluated at all; an undefined use of the term by its speaker is itself evidence the underlying claim is empty rhetoric rather than a substantive assertion.conf: medium
- implicitCEOs' public claims about frontier AI capability should be discounted by default because making outlandish claims is structurally part of the CEO role ("CEO's gotta CEO"), independent of the specific evidence for any given claim.conf: mediumextendsNo, Anthropic did not call for a pause on AI development · p8 — Generalizes Marcus's earlier claim that frontier-lab safety statements are marketing artifacts, not policy commitments, into a broader principle that CEO claims about AI milestones (like the Singularity) carry no truth-value just because a CEO said them.
- explicitDemis Hassabis's characterization of current AI progress as being only in 'the foothills' is a safer, more defensible framing than Altman's and Musk's Singularity claims.conf: high
Arguments
We have not reached the Singularity, under any of the senses in which the term is used. strength: high
Explicit premises:
- Under the 'age of abundance' sense, current labor-market behavior (quiet rehiring) contradicts it.
- Under the 'growth beyond human control' sense, current incidents remain human-caused and preventable.
- Under Kurzweil's brain-computer-merge sense, that technology plainly isn't here yet.
- Under I.F. Good's ultraintelligence/intelligence-explosion sense, current AI fails 8 of 10 benchmark capability tasks.
Implicit premises:
- Altman and Musk mean one of these several existing, citable definitions when they say 'Singularity' — rather than some other, unstated sense of their own.
- A claim about reaching a landmark technical threshold should be evaluated against the term's established technical/literature definitions, not treated as loose motivational rhetoric.
Inferences:
- That 'Singularity' talk from CEOs functions primarily as a marketing/fundraising narrative rather than a claim intended to be literally falsifiable — Marcus gestures at this ("CEO's gotta CEO") but does not draw it out as an explicit conclusion.
CEOs' Singularity claims shouldn't be taken seriously. strength: low
Explicit premises:
- None of them has defined the term.
- Nobody but Marcus has asked them to define it.
Implicit premises:
- An undefined term used in a sweeping claim is decisive evidence the claim is empty, rather than merely under-specified.
- Colloquial/metaphorical use of a term by a public figure carries the same evidentiary burden as a technical claim.
Narratives
- The 'Klarna Effect': companies publicly boast of AI-driven layoffs/hiring freezes, then quietly rehire humans once AI doesn't deliver the promised productivity, per WSJ reporting on employers modestly expanding headcount again.
→ Supports the proposition that we have not reached an AI-driven 'age of abundance,' and more broadly that industry PR about AI capability runs ahead of what AI can actually deliver. - The OpenAI hack of HuggingFace, framed as a human-launched, preventable 'drill' rather than a spontaneous act of an uncontrollable system.
→ Supports the proposition that we have not reached a moment of technological growth beyond human control. - The Miles Brundage bet: a list of 10 tasks 'as good or better than the best human' that Marcus proposed 18 months earlier as an AGI benchmark, of which current AI can do at most 2.
→ Supports the proposition that current AI is far from I.F. Good's ultraintelligence threshold.
Questions addressed
- ✅ Have we reached the Singularity, as Altman and Musk claim?
No, under every definition of the term Marcus can locate. - ⭕ What do Altman and Musk actually mean by 'the Singularity'?
Explicitly left unresolved — Marcus's central complaint is precisely that neither has defined the term, and this remains an open question by the piece's own account. - ⭕ Will there ever be a single, identifiable 'Singularity moment'?
Marcus states his doubt that such a discrete moment will ever be recognized, given intelligence's multidimensionality, but treats this as an open speculation rather than a settled resolution.
Gaps
- illogicalMarcus's headline rhetorical move — demanding a definition and treating its absence as near-decisive evidence the claim is empty — risks being a debate-winning gotcha rather than genuine engagement. He never considers that Altman/Musk might be using 'Singularity' as a loose metaphor for 'we've crossed into a period of rapid, hard-to-predict change,' which wouldn't require the technical precision Marcus demands to carry some (weaker) meaning. The strength of the piece actually comes from testing several *established* rigorous definitions against evidence and finding each wanting — that part of the argument doesn't need the 'nobody defined it' framing to work, and arguably the framing weakens the piece by putting a shakier argument in the most prominent position (the title and opening).
Read next: Altman's original 'self-contradictory' blog post that Marcus links to but doesn't quote or analyze — reading the actual source would let one assess Marcus's 'self-contradictory' and 'never defines the term' charges directly rather than taking them on faith. - incompleteMarcus links to but never quotes or walks through Altman's 'self-contradictory' blog post on the Singularity, so the specific charge of self-contradiction is asserted, not demonstrated within this piece.
Read next: Read Altman's original blog post directly and check whether it actually contradicts itself, and whether it offers even an informal working definition Marcus is eliding. - incompleteThe piece treats three quite different technical usages of 'the Singularity' (Good's intelligence explosion, Kurzweil's brain-computer merge, and a popular 'growth beyond human control' framing) somewhat interchangeably in a single takedown, without ever stating explicitly that these are actually distinct, sometimes incompatible concepts in the singularity literature (Vinge's original technological-singularity sense differs from both). Making that distinction explicit would have made the multi-front rebuttal more rigorous rather than just serially plausible.
Read next: Vernor Vinge's original 1993 essay on the technological singularity, to compare his definition against Good's and Kurzweil's and see how much daylight exists between the senses Marcus treats as roughly equivalent.