FlowingFlowing
← Back to all development cases
This case was judged by an anonymized AI reviewer and has not been independently reviewed by people; inspect the draft, cards, and both outputs to assess it.
AutocompleteWithout cards better54 / 280 · beu-23ab8e4576f7a60e

Scalar induced gravitational waves review

Astrophysics · 2109.01398v2

FLOWING EVIDENCE BENCHMARK

How do we compare writing with and without cards?

At the same writing position, the same model produces one first result with cards and one without. All 280 pairs enter the overall anonymized review, which assesses whether a decision not to continue fits the draft.

Overall: 138 with-card wins, 26 ties, 107 no-card wins, 5 pairs where neither was usable, and 4 inconclusive judgments. Both sides wrote text in 218 pairs; 48 had a deliberate stop and 14 included an invalid output.

SAME DRAFT · WITH OR WITHOUT CARDS

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED VERSIONS

SHARED BY BOTH

Draft, writing task, and generation model

A · With cards

Relevant literature cards supplied

B · Without cards

No literature cards supplied

Generation model

Same model and version

A → first output

B → first output

Anonymized AI judge

First outputs are anonymized as X and Y

Continuation review: accuracy, fit to the writing task and usability
Returns: X preferred / tie / Y preferred / both unusable

Review again with the answer order swapped

HOW DOES BLIND JUDGING WORK?

① Anonymize both outputs
The AI judge sees the same draft and both first outputs without knowing which received cards.

② Compare and swap order
The judge applies task-specific criteria in X/Y and then Y/X order.

③ Review disagreements
A third pass resolves disagreements. Incomplete reviews remain in the denominator.

How are positions and cards selected?

We selected 40 writing positions in each of seven fields, for 280 comparisons. The position and relevant literature cards were set before each comparison, so both versions began from the same draft.

The main result retains every frozen position, including places where cards might not add a useful next move. This avoids selecting cases based on their outputs, but 49% describes the whole cohort, not a win rate where evidence clearly had an opportunity to help. This cohort did not record that opportunity label for each position before generation, so those two groups cannot yet be separated reliably.

CARD USE · 140/280

Reported card use in continuations with text

The model reports whether a card affected its continuation. In 140 of 280 tests it produced continuation text and reported card use. The previous count of 150 included 10 cases without continuation text and has been corrected.

COMPARISON · 70 / 7 / 30

Which written version was better?

Of the 107 reported-use cases where both versions produced text, the blinded AI judge preferred the with-card version 70 times, found 7 about the same, and preferred the without-card version 30 times.

How are the two table percentages calculated?

ADOPTION COVERAGE

140reported use with text
÷
280all positions
=
50%

WITH-CARD VERSION PREFERRED

70with cards better
÷
107comparable tests
=
65%

The percentages use different groups: 50% counts continuations with text reporting card use among all 280 tests; 65% covers the 107 such cases with text on both sides. The generating model reports card use, while an anonymized AI judge compares writing quality.

Card use is self-reported by the generator. The anonymized AI judge sees the draft and source snippets, so source-aware preference remains possible; there was no independent human review. The overall result describes these 280 frozen positions, not a win probability for any writing task.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

Scalar induced gravitational waves review

11 Conclusions

…2}_{\Phi}\,. (10.13) We can then use Eq. ( 6.19 ) together with ( 10.1 ) to estimate the induced GW spectrum today. Note that if the UV cut-off k UV k_{\rm UV} is at very small scales, i.e., k UV ≫ k rh k_{\rm UV}\gg k_{\rm rh} , the amplitude of induced GW is very much enhanced and strong constraints apply. It is important to clarify that if k UV k_{\rm UV} is very large (larger than k N ​ L k_{NL} in Eq. ( 6.1 )), some density fluctuations would enter the non-linear regime. Thus, the above result must be considered as a rough estimate. For detailed discussion we refer the reader to Sec. 6 .

…ced GWs together with primordial black holes are receiving serious attention since the first GW detection by LIGO [ 61 ] . They are even more interesting after the NANOGrav collaboration recently reported a possible stochastic GW signal [ 18 ] . Within the next couple of decades, LISA will launch, and we will have a new data on stochastic backgrounds of gravitational waves in a wide frequency range. Even in the worst-case scenario, where no conclusive induced GW signal is found, we will obtain new constraints on the physics of inflation and the early universe in completely unexplored regimes.

The field of cosmology with induced GWs is relatively new and still developing. In this review we have focused on the current analytical techniques and estimates to compute the induced GW

↓ Next paragraph was generated here

The highlighted text is near the writing position. This is a layout of the evaluation input, not a PDF screenshot.

A matched comparison at the same writing positionOverall AI review: Without cards better

Both arms use the same model, task and manuscript context. Only the retrieved passages differ. The first recorded output from each arm is shown below.

Read overall AI review rationale

A = without cards · B = with cards

Both outputs complete the sentence grammatically and appropriately, but Output A is more precise and better fitted to the specific context. The draft explicitly states 'we have focused on the current analytical techniques and estimates to compute the induced GW' — the key term being 'induced GW' (scalar-induced gravitational waves, a specific technical subfield). Output A directly follows this by specifying 'the scalar-induced case and its cosmological implications,' which matches the established focus and uses the exact terminology. Output B broadens to 'main cosmological sources and their observational prospects,' which drifts toward a general review scope that contradicts the draft's self-description of having 'focused' on something more specific. Output A maintains appropriate scope; Output B risks overclaiming breadth. The 'complete' action is appropriate in both cases since the sentence finishes a concluding statement about review scope.

Evidence

Same model and position · retrieved passages supplied · first output

Not preferred in this comparison

spectrum, as well as on the main cosmological sources and their observational prospects.

Passages supplied to the Evidence version

Card 1 · Cosmological Backgrounds of Gravitational Waves

Sects. 5 - 9 represent an exhaustive updated survey of the potential sources from the early universe, that produce a stochastic background of gravitational waves. We review in extensive detail the physical origin and motivation of each source, the properties of the GW background they originate, and…

Read full passage excerpt

Sects. 5 - 9 represent an exhaustive updated survey of the potential sources from the early universe, that produce a stochastic background of gravitational waves. We review in extensive detail the physical origin and motivation of each source, the properties of the GW background they originate, and discuss the ability (if any) of GW detectors to place constraints on each scenario. Present GW detectors are not really optimized for the detection of a stochastic background of cosmological origin. The analysis presented in this review, indicates however that a fraction (in some cases significant) of the parameter space in realistic scenarios (e.g. particle production during inflation, phase transitions, or cosmic string networks), is compatible with a detection by present or future experiments.

Card 2 · Gravitational wave background from reheating after hybrid inflation

The reheating of the Universe after hybrid inflation proceeds through the nucleation and subsequent collision of large concentrations of energy density in the form of bubblelike structures moving at relativistic speeds. This generates a significant fraction of energy in the form of a stochastic back…

Read full passage excerpt

The reheating of the Universe after hybrid inflation proceeds through the nucleation and subsequent collision of large concentrations of energy density in the form of bubblelike structures moving at relativistic speeds. This generates a significant fraction of energy in the form of a stochastic background of gravitational waves, whose time evolution is determined by the successive stages of reheating: First, tachyonic preheating makes the amplitude of gravity waves grow exponentially fast. Second, bubble collisions add a new burst of gravitational radiation. Third, turbulent motions finally sets the end of gravitational waves production. From then on, these waves propagate unimpeded to us. We find that the fraction of energy density today in these primordial gravitational waves could be significant for grand unified theory (GUT)-scale models of inflation, although well beyond the frequency range sensitivity of gravitational wave observatories like LIGO, LISA, or BBO. However, low-scale …

Card 3 · Ability of LIGO and LISA to probe the equation of state of the early Universe

The expansion history of the Universe between the end of inflation and the onset of radiation-domination (RD) is currently unknown. If the equation of state during this period is stiffer than that of radiation, w > 1/3, the gravitational wave (GW) background from inflation acquires a blue-tilt d log…

Read full passage excerpt

The expansion history of the Universe between the end of inflation and the onset of radiation-domination (RD) is currently unknown. If the equation of state during this period is stiffer than that of radiation, w > 1/3, the gravitational wave (GW) background from inflation acquires a blue-tilt d log ρ GW / d log f = 2( w −1/3)/ ( w +1/3) > 0 at frequencies f ≫ f RD corresponding to modes re-entering the horizon during the stiff-domination (SD), where f RD is the frequency today of the horizon scale at the SD-to-RD transition. We characterized in detail the transfer function of the GW energy density spectrum, considering both `instant' and smooth modelings of the SD-to-RD transition. The shape of the spectrum is controlled by w , f RD , and H inf (the Hubble scale of inflation). We determined the parameter space compatible with a detection of this signal by LIGO and LISA, including possible changes in the number of relativistic degrees of freedom, and the presence of a tensor tilt. Cons …

Card 4 · Testing primordial black holes as dark matter with LISA

The idea that primordial black holes (PBHs) can comprise most of the dark matter of the Universe has recently reacquired a lot of momentum. Observational constraints, however, rule out this possibility for most of the PBH masses, with a notable exception around ${10}^{\ensuremath{-}12}\text{ }\text{…

Read full passage excerpt

The idea that primordial black holes (PBHs) can comprise most of the dark matter of the Universe has recently reacquired a lot of momentum. Observational constraints, however, rule out this possibility for most of the PBH masses, with a notable exception around ${10}^{\ensuremath{-}12}\text{ }\text{ }{M}_{\ensuremath{\bigodot}}$. These light PBHs may be originated when a sizable comoving curvature perturbation generated during inflation reenters the horizon during the radiation phase. During such a stage, it is unavoidable that gravitational waves (GWs) are generated. Since their source is quadratic in the curvature perturbations, these GWs are generated fully non-Gaussian. Their frequency today is about a millihertz, which is exactly the range where the LISA mission has the maximum of its sensitivity. This is certainly an impressive coincidence. We show that this scenario of PBHs as dark matter can be tested by LISA by measuring the GW two-point correlator. On the other hand, we show …

Card 5 · Primordial Gravitational Waves with LISA

Primordial Gravitational Waves are the next target of modern cosmology. They represent a window on the early Universe and the only probe of the physics and microphysics of the inflationary period. When the production of GWs happens in scenarios richer than the standard single-field slow-roll, the GW…

Read full passage excerpt

Primordial Gravitational Waves are the next target of modern cosmology. They represent a window on the early Universe and the only probe of the physics and microphysics of the inflationary period. When the production of GWs happens in scenarios richer than the standard single-field slow-roll, the GW signal becomes potentially detectable also on scales smaller than the Cosmic Microwave Background. LISA will be extremely complementary to CMB experiments to extract information about primordial inflationary models and in particular to probe phases of the inflationary period for which we have very poor knowledges.

Model-reported card use

The model did not report that any card materially affected this completion.

No Evidence

Same model and position · no retrieved passages · first output

Preferred by blind review

spectrum, with particular emphasis on the scalar-induced case and its cosmological implications.

Earlier card-use comparison

The earlier diagnostic compared only cases with reported card use and text on both sides; its scope differs from the overall AI review above.

This case does not have both with-card continuation text and a model-reported use record. It remains among all 280 tests but does not enter the reported-use comparison.

The passages are input material; their presence does not mean an output used them or that they support every claim. Draft and source passages are excerpted; no simulated scores or PDF appear here.