“I guarantee you Anthropic, Google and Meta are all working on something doing exactly the same thing, because what they are describing is something that limits every single Large Language Model due to the very nature of how they’re built, something Reuters even acknowledges by adding that “this is something that has eluded AI models to date.””
Our reading, the sentence that was checked:
Every LLM is limited by its nature in planning ahead and autonomously doing research, and Anthropic, Google and Meta are all working on the same problem.
Assessment: Unresolved Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, Put Up Or Shut Up, paragraph 23.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-put-up-or-shut-up-ec56b0-a2
We recorded 19 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 2024-07-16. 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
- Anthropic, Google and Meta
- are all working on something doing exactly the same thing
- limits every single Large Language Model
- due to the very nature of how they’re built
Our interpretation
- The referenced capability is planning ahead to navigate the web autonomously and reliably for research.
- Exactly the same thing means a comparable research objective, not identical code or model architecture.
- The claim combines three companies' development activity with a universal, architecture-based explanation of model limitations.
- I use July 2024 through the post date as the current reporting window. Earlier research is context, not proof that an identical project remained active then.
Counts as supported if: Records establish that all three named companies were developing comparable autonomous research capabilities by the post date, while evaluations and technical analysis establish the claimed shared architectural limitation.
Counts as contradicted if: Reliable records show that a named company was not pursuing that objective, or a contemporary LLM demonstrates the allegedly missing reliable autonomous research capability, or technical evidence refutes the asserted architectural cause.
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
Anthropic reported agentic coding results in June 2024 [e1]. Google-affiliated researchers described web planning and action-taking in July 2023 [e2]. Meta's May 6, 2024 GAIA record describes a benchmark involving reasoning, browsing and tools [e3]. These records do not establish comparable autonomous-research projects active at all three companies during July 1–16, 2024 [e1] [e2] [e3]. A July 1, 2024 paper reports low success rates on tested web tasks and improvements from adding search [e4]. Those findings establish limitations in the tested agents, but do not establish an inherent architectural barrier affecting every LLM [e4]. Both the company-specific development claim and the universal architectural explanation remain unresolved [e1] [e2] [e3] [e4].
Limitations
- The company records concern different activities and predate the target period [e1] [e2] [e3].
- The demonstrated web actions do not establish reliable autonomous research [e2].
- The evaluations do not establish architecture as the necessary cause of every LLM's limitations [e4].
Evidence (4 items)
Supports the claim: arXiv; Jing Yu Koh and colleagues,
still far worse than humans.
Why it matters here: The contemporary paper supports the limited proposition that web agents had substantial capability gaps. Its improvements from inference-time search do not establish that every model's limitations were inherent in its architecture.
- Publisher
- arXiv; Jing Yu Koh and colleagues
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Anthropic,
64%
Why it matters here: Anthropic was developing and evaluating agentic capabilities before the post. Coding performance does not establish a comparable autonomous web-research project or a universal architectural limit.
- Publisher
- Anthropic
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: arXiv; Google DeepMind and University of Tokyo researchers,
tasks on real websites
Why it matters here: This establishes earlier Google-affiliated work on planning and acting on websites. It also cautions against reading Zitron as claiming that all autonomous web action was impossible, but does not establish reliable deep research in July 2024.
- Publisher
- arXiv; Google DeepMind and University of Tokyo researchers
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Meta AI; ICLR,
a benchmark for General AI Assistants
Why it matters here: Meta's research record includes a benchmark combining reasoning, browsing, and tool use. Building a benchmark is related research, not proof of an active Strawberry-equivalent development project.
- Publisher
- Meta AI; ICLR
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
What would settle it
Dated records establishing or ruling out comparable autonomous-research projects at each named company during July 1–16, 2024, plus technical analysis and contemporary evaluations testing the asserted architectural necessity against a defined research task and reliability threshold.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, Put Up Or Shut Up
I guarantee you Anthropic, Google and Meta are all working on something doing exactly the same thing, because what they are describing is something that limits every single Large Language Model due to the very nature of how they’re built, something Reuters even acknowledges by adding that “this is something that has eluded AI models to date.”
- Where
- paragraph 23
- Published
- 2024-07-16
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-put-up-or-shut-up-ec56b0-a2: 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.
HeldBy this site
The final-text check failed: rationale_supported_by_evidence; overclaims: The record supports a shared research direction and substantial contemporary limitations..
About assessment wyea-put-up-or-shut-up-ec56b0-a1, which was not published.