Pundit ClaimsAI claims by Gary Marcus and Ed Zitron, with the evidence

“...maybe? It sure isn't trending that way, nor has it gone down yet.”

How To Argue With An AI Booster, paragraph 238. The full statement is at the source.

Our reading, the sentence that was checked:

The cost of inference is not trending downward and has not gone down yet, though he says it maybe could go down.

Assessment: Contradicted Evidence as of

Where and when Ed Zitron wrote it

Where's Your Ed At, How To Argue With An AI Booster, paragraph 238.

Said
Source observed
live page, full text, public
Evidence as of
the cutoff for this assessment
Assessment recorded
assessment wyea-how-to-argue-with-an-ai-booster-064588-a2

We recorded 35 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 2025-08-25. 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

  • '...maybe?' allows that it could go down in future (qualifier, stated by the speaker)

From the speaker

  • It sure isn't trending that way
  • nor has it gone down yet
  • ...maybe?

Our interpretation

  • The maybe concerns a future possibility; it does not qualify the assertions about costs through publication.
  • Read cost as the full inference expense of performing comparable work, including generated tokens, rather than token price alone.
  • Assess both the claimed absence of a downward trend and the categorical historical assertion that costs had not fallen. A documented historical decline can contradict the latter without establishing the latest application-wide trend.
  • Use the surrounding approximately two-year discussion: August 2023 through publication, with the exact anniversary as an operational boundary.
  • This concerns industry inference service costs rather than a named company's financial accounts; metric_scope is null.

Counts as supported if: Comparable task-cost evidence shows no historical reduction during the period and no downward trend through publication, after including token consumption.

Counts as contradicted if: Credible longitudinal measurements show that full inference costs for comparable tasks or capability levels had already fallen during the period, or establish a downward trend through publication.

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

Epoch's March 12, 2025 analysis found that benchmark evaluation costs at comparable performance levels had declined, accounting for recorded token consumption [e1]. These historical declines contradict the assertion that inference costs had not gone down [e1]. The analysis does not establish the direction of general application costs or frontier-workload spending in August 2025 [e1].

Limitations

  • Fixed-capability benchmark costs do not establish aggregate application spending or costs for increasingly demanding workloads.
  • Benchmark coverage, possible overfitting and evaluation variability limit generalization.

Evidence (1 item)

  • Goes against the claim: Epoch AI,

    evaluation costs have declined

    Why it matters here: Epoch checked evaluation costs using recorded token consumption and found historical declines at comparable benchmark performance levels. This addresses Zitron's objection to token-price-only comparisons and contradicts the assertion that inference costs had not fallen.

    Publisher
    Epoch AI
    Kind of source
    Independent measurement
    Describes events of
    Public from

Where it was said (1)

Said once in the sources this register covers.

  • : Where's Your Ed At, How To Argue With An AI Booster

    ...maybe? It sure isn't trending that way, nor has it gone down yet.

    Published
    2025-08-25
    Source observed
    , live page
    Access
    full text, public

History

Assessment versions

Recorded events

  1. Assessment publishedBy this site

    First assessment published after both reviews and the final-text check passed.

    About assessment wyea-how-to-argue-with-an-ai-booster-064588-a2.