“But on the whole, Generative AI has not been delivering great gains. We have seen that now in study after study, MIT, McKinsey, Bain, and many more.”
Our reading, the sentence that was checked:
Overall, generative AI has not been delivering large gains, as many studies (MIT, McKinsey, Bain) show.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Accenture: Then and now, and how it may signify things to come, paragraph 5.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-accenture-then-and-now-and-how-it-a4ae66-a1
We recorded 4 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2026-06-18. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- on the whole (qualifier, stated by the speaker)
From the speaker
- on the whole
- Generative AI
- has not been delivering great gains
- study after study
- MIT, McKinsey, Bain
Our interpretation
- The surrounding discussion makes substantial business-wide productivity gains and realized financial returns the ordinary reading of 'great gains'. Individual task improvements alone do not settle that claim.
- The cumulative wording refers to the record available by June 18, 2026. I use the latest available study waves, retain their actual observation dates, and leave the starting boundary unspecified because the claim supplies none.
- Positive returns need not be large returns. Missing expectations need not mean gains were small.
Counts as supported if: Broad, reasonably representative evidence shows that substantial business-wide gains from generative AI remained uncommon, and the named studies substantively support that conclusion.
Counts as contradicted if: Comparable evidence shows substantial business-wide gains were widespread, or the named studies substantially undermine the claimed overall pattern.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The retrieved evidence supports concern about limited enterprise impact, but does not settle the full generalization. NANDA reports widespread implementation failure in its sample. Reporting on Bain describes savings below expectations across a broader AI and automation population. Wharton's survey reports positive returns among most respondents, without establishing their size. These are different measures and populations. McKinsey's original text could not be verified. The evidence therefore does not securely establish or refute large gains 'on the whole'.
Limitations
- The NANDA document is preliminary and was retrieved from a third-party mirror.
- Several results rely on respondents' reports rather than independently measured financial outcomes.
- A narrower claim about difficulties obtaining enterprise returns would have stronger support.
Evidence (3 items)
Supports the claim: Project NANDA authors; report mirrored by Mirro,
The 95% failure rate
Why it matters here: The January–June research reports limited sustained productivity or profit impact from enterprise implementations. Its interview-based estimates, varying samples, and differing success definitions limit generalization to all businesses.
- Publisher
- Project NANDA authors; report mirrored by Mirro
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Bloomberg News, republished by Insurance Journal,
reductions of 10% or less
Why it matters here: Bain's survey of 951 companies documents savings below expectations. It covers automation and AI more broadly, and neither this bucket nor a missed target establishes that most generative-AI deployments produced small gains.
- Publisher
- Bloomberg News, republished by Insurance Journal
- Kind of source
- News report
- Describes events of
- Public from
Context only: Knowledge at Wharton,
positive returns
Why it matters here: This institutional summary describes a generative-AI survey of more than 800 senior leaders at large US companies. It challenges a blanket no-return reading, but self-reported positive returns do not establish large gains or a population-wide aggregate.
- Publisher
- Knowledge at Wharton
- Kind of source
- Secondary summary
- Describes events of
- Public from
What would settle it
Readable original studies and comparable evidence showing how commonly substantial enterprise-wide generative-AI gains had materialized by June 18, 2026.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Accenture: Then and now, and how it may signify things to come
But on the whole, Generative AI has not been delivering great gains. We have seen that now in study after study, MIT, McKinsey, Bain, and many more.
- Where
- paragraph 5
- Published
- 2026-06-18
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-accenture-then-and-now-and-how-it-a4ae66-a1: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment marcus-accenture-then-and-now-and-how-it-a4ae66-a1.