The authors see no reason to think that current large language models encode stable representations of the world at all, and say that all these models are literally doing is predicting the next words in sentences.

What happened

How we judged it

Kind of claim

Observation

Deadline

None given. We don’t add one.

Conditions

• we see no reason to think (their words)
• current large language models (their words)

Where we had to interpret

• Preserve the authors' qualified assessment of the evidence; do not replace it with a categorical denial that any internal world-related information exists.
• The next-word statement is presented as the rationale for that assessment. A training objective alone does not specify what intermediate representations a model can learn.
• The passage supplies neither a criterion for stable world representations nor a standard for evidence sufficient to change the authors' assessment.
• A verdict on the combined claim would therefore require adding a consequential standard the speaker did not supply.

We don’t grade claims like this. We only check the quote and who said it.

What would settle it

Clarification of the intended stability criterion and whether the authors mean an objective absence of qualifying evidence or their own assessment of its persuasiveness.