“all based on the outcomes from Large Language Models that simply do not do what their progenitors have promised and do not produce ROI or productivity benefits that are in any way measurable.”
Our reading, the sentence that was checked:
LLMs do not do what their makers promised and produce no measurable ROI or productivity benefits.
Assessment: Contradicted Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, The OpenAI Bubble, paragraph 260.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-the-openai-bubble-560a28-a1
We recorded 31 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 2026-07-15. 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)
From the speaker
- Large Language Models
- do not do what their progenitors have promised
- do not produce ROI or productivity benefits
- in any way measurable
Our interpretation
- The ordinary reading is a categorical denial of measurable benefits, rather than a claim that benefits are uncommon or insufficient to justify industry investment.
- For this undated ongoing observation, use April–June 2026, the latest completed calendar quarter before publication. Earlier measurements provide context only.
- A credible measured improvement in productive output can contradict the productivity clause. It does not establish positive financial ROI.
- Productive output per worker is an ordinary productivity indicator, but its adequacy depends on whether changes in hours, quality and downstream work explain the increase.
- The makers, promises and promised deadlines are unspecified. That clause cannot receive a separate definitive assessment.
- No particular company's financial metric is asserted, so metric_scope is null.
Counts as supported if: Credible, sufficiently broad measurements for the target period establish no positive productivity or net financial returns from LLM use, and evaluations of identifiable maker promises establish their failure. Isolated unsuccessful deployments are insufficient to establish the categorical denial.
Counts as contradicted if: At least one credible comparison for the target period measures a positive productivity benefit attributable to LLM use, or a positive financial return after relevant costs. This contradicts the categorical benefits clause without establishing that all promises were fulfilled.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Contradicted
The dated evidence goes against what the claim committed to, under the criteria frozen before grading.
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 categorical denial of any measurable productivity benefit is contradicted under the stated reading. An [original study](https://arxiv.org/html/2607.01904v2) measures higher engineering output associated with LLM-tool adoption in April 2026 and checks several alternative explanations. This establishes a measurable benefit on an ordinary output-per-worker indicator, with substantial qualifications. It does not establish positive financial ROI, universal benefits or fulfillment of unspecified promises. Because the result depends on accepting that productivity indicator, interpretation_determines_result is true.
Limitations
- The decisive evidence is a preprint with nonrandom adoption.
- Engineering output is a partial productivity measure; downstream costs and long-term quality remain incompletely measured.
- The contradiction applies to the categorical productivity denial. The ROI and unspecified-promises clauses remain unsettled.
Evidence (2 items)
Goes against the claim: arXiv; Hao He and colleagues,
The gain grows with time on tool.
Why it matters here: The study reports higher pull-request output per developer in April 2026 and a preferred estimate of 1.46 times output associated with adoption and accumulated use, after calendar-month controls. Checks address extra working days and splitting work, but nonrandom adoption, review burdens and incomplete quality measures limit causal certainty and prevent treating the result as financial ROI.
- Publisher
- arXiv; Hao He and colleagues
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: METR,
AI makes them slower.
Why it matters here: A randomized experiment found slower completion among experienced developers working in familiar repositories. This is material evidence that benefits vary by setting, but its February–June 2025 observations cannot establish the state of productivity in the target quarter.
- Publisher
- METR
- Kind of source
- Independent measurement
- Describes events of
- Public from
What would settle it
Independent replication measuring working hours, useful completed work, downstream maintenance and all relevant costs would strengthen the assessment. Identifying the maker promises would allow that clause to be assessed separately.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, The OpenAI Bubble
all based on the outcomes from Large Language Models that simply do not do what their progenitors have promised and do not produce ROI or productivity benefits that are in any way measurable.
- Where
- paragraph 260
- Published
- 2026-07-15
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-the-openai-bubble-560a28-a1: Current. Outcome: Contradicted. 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.