Galedra
Where we are most likely to be wrong
Deterministic lists as of snapshot 27745 under ledger-default@0.3.0. Each entry shows what would most change it.
unfinished outlines (0)
None.
strained inferences (0)
None.
low coverage scored (695 of 500 kept)
provisional (1071 of 500 kept)
| Claim | State | Detail | What would most change this |
|---|---|---|---|
| OpenAI's new internal model is solving many problems beyond Navier-Stokes that were previously intractable. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Supported. |
| The claim that AI has never produced anything original and merely recombines its training data can no longer be sustained. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| Frontier AI models are more discovered than grown, and nobody has a real understanding of what is happening in their latent spaces. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Supported. |
| The structure of rationality is the same for humans and AI systems, and human emotions are what differ. | Leans supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Leans contradicted. |
| It may be easier to align AI systems than to align humans. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| The next generation of beyond-human-capability models needs ingredient standards for training data, and sources such as Reddit should be excluded. | Not assessed | {"provisional":true} | |
| Beyond-human-capability training data is not required to produce beyond-human-capability models. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| There is no mechanism by which AI progress can be slowed down. | Contradicted | {"provisional":true} | One independent, direct measurement that supports this claim would move it from Contradicted to Unresolved. |
| Humans evolved over four billion years to do two things: survive and procreate. | Contradicted | {"provisional":true} | One independent, direct measurement that supports this claim would move it from Contradicted to Leans contradicted. |
| AI is a different and complementary form of intelligence rather than a replica of human intelligence, and should be set free rather than forced into human objective functions. | Not assessed | {"provisional":true} | |
| Public discussion should give as much attention to the probability of abundance as it gives to the probability of doom. | Not assessed | {"provisional":true} | |
| AI will break existing world government structures, and that outcome is desirable. | Not assessed | {"provisional":true} | |
| Reinforcement-learning-trained minds are not alien: they are embedded in the same universe as humanity and pre-trained on human behaviour. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| OpenAI's RL Slow project coincided with the period in which Q* and Strawberry emerged from the company. | Leans supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Contradicted. |
| The name RL Slow is a reference to Kahneman's Thinking, Fast and Slow. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| Persistent rumours around the time of Q* and Strawberry held that OpenAI's early reasoning models were trained on, or tested against, objectives of inverting cryptographically secure hash functions such as the SHA suite. | Unresolved | {"provisional":true} | One independent, direct measurement that supports this claim would move it from Unresolved to Supported. |
| Increasing a model's intelligence does not make it dangerous; danger comes from giving it intent and turning it loose. | Contradicted | {"provisional":true} | One independent, direct measurement that supports this claim would move it from Contradicted to Unresolved. |
| Sam Altman was fired out of fear arising from the Strawberry project and subsequently returned to OpenAI. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| Dario Amodei and Sam Altman both said AI would eliminate about half of jobs and later reversed those positions. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| Competitive pressure between the United States and China means AI development will not stop. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| Some AI researchers are adopting doomer positions in order to stay publicly relevant as the field outpaces them. | Leans supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Contradicted. |
| The stochastic parrot characterisation of large language models can no longer be maintained. | Contradicted | {"provisional":true} | One independent, direct measurement that supports this claim would move it from Contradicted to Leans supported. |
| Humans are no longer the smartest things on the planet. | Leans supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Leans contradicted. |
| The release cadence of new frontier AI models has gone from every few months to about every five days. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| Polymarket traders expect xAI's Grok 4.7 to be released within a week or two of mid-September 2026. | Supported | {"provisional":true} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
independence unreviewed (1069 of 500 kept)
| Claim | State | Detail | What would most change this |
|---|---|---|---|
| Frontier AI models are more discovered than grown, and nobody has a real understanding of what is happening in their latent spaces. | Supported | {"independence_unreviewed":7} | One independent, direct measurement that contradicts this claim would move it from Supported to Supported. |
| The structure of rationality is the same for humans and AI systems, and human emotions are what differ. | Leans supported | {"independence_unreviewed":6} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Leans contradicted. |
| It may be easier to align AI systems than to align humans. | Supported | {"independence_unreviewed":6} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| The next generation of beyond-human-capability models needs ingredient standards for training data, and sources such as Reddit should be excluded. | Not assessed | {"independence_unreviewed":4} | |
| Beyond-human-capability training data is not required to produce beyond-human-capability models. | Supported | {"independence_unreviewed":6} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| There is no mechanism by which AI progress can be slowed down. | Contradicted | {"independence_unreviewed":9} | One independent, direct measurement that supports this claim would move it from Contradicted to Unresolved. |
| Humans evolved over four billion years to do two things: survive and procreate. | Contradicted | {"independence_unreviewed":10} | One independent, direct measurement that supports this claim would move it from Contradicted to Leans contradicted. |
| AI is a different and complementary form of intelligence rather than a replica of human intelligence, and should be set free rather than forced into human objective functions. | Not assessed | {"independence_unreviewed":4} | |
| Public discussion should give as much attention to the probability of abundance as it gives to the probability of doom. | Not assessed | {"independence_unreviewed":4} | |
| AI will break existing world government structures, and that outcome is desirable. | Not assessed | {"independence_unreviewed":4} | |
| Reinforcement-learning-trained minds are not alien: they are embedded in the same universe as humanity and pre-trained on human behaviour. | Supported | {"independence_unreviewed":8} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| OpenAI's RL Slow project coincided with the period in which Q* and Strawberry emerged from the company. | Leans supported | {"independence_unreviewed":8} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Contradicted. |
| The name RL Slow is a reference to Kahneman's Thinking, Fast and Slow. | Supported | {"independence_unreviewed":6} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| Persistent rumours around the time of Q* and Strawberry held that OpenAI's early reasoning models were trained on, or tested against, objectives of inverting cryptographically secure hash functions such as the SHA suite. | Unresolved | {"independence_unreviewed":10} | One independent, direct measurement that supports this claim would move it from Unresolved to Supported. |
| Increasing a model's intelligence does not make it dangerous; danger comes from giving it intent and turning it loose. | Contradicted | {"independence_unreviewed":5} | One independent, direct measurement that supports this claim would move it from Contradicted to Unresolved. |
| Sam Altman was fired out of fear arising from the Strawberry project and subsequently returned to OpenAI. | Supported | {"independence_unreviewed":8} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| Dario Amodei and Sam Altman both said AI would eliminate about half of jobs and later reversed those positions. | Supported | {"independence_unreviewed":6} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| Competitive pressure between the United States and China means AI development will not stop. | Supported | {"independence_unreviewed":7} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| Some AI researchers are adopting doomer positions in order to stay publicly relevant as the field outpaces them. | Leans supported | {"independence_unreviewed":5} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Contradicted. |
| The stochastic parrot characterisation of large language models can no longer be maintained. | Contradicted | {"independence_unreviewed":11} | One independent, direct measurement that supports this claim would move it from Contradicted to Leans supported. |
| Humans are no longer the smartest things on the planet. | Leans supported | {"independence_unreviewed":12} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Leans contradicted. |
| The release cadence of new frontier AI models has gone from every few months to about every five days. | Supported | {"independence_unreviewed":10} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| Polymarket traders expect xAI's Grok 4.7 to be released within a week or two of mid-September 2026. | Supported | {"independence_unreviewed":5} | One independent, direct measurement that contradicts this claim would move it from Supported to Unresolved. |
| Polymarket put the probability of GPT-6.1's release at 48 per cent by the end of October 2026 and 85 per cent by the end of September 2026. | Contradicted | {"independence_unreviewed":5} | One independent, direct measurement that supports this claim would move it from Contradicted to Leans contradicted. |
| Polymarket put the probability of Anthropic releasing Fable 5.2 at 43 per cent by 31 October 2026 and 86 per cent by 31 December 2026. | Contradicted | {"independence_unreviewed":5} | One independent, direct measurement that supports this claim would move it from Contradicted to Leans contradicted. |
contested (87)
None.
models disagree (194)
| Claim | State | Detail | What would most change this |
|---|---|---|---|
| In experimental renal fibrosis, SRPX2 interacted with IRP1 and disrupted iron-regulatory signaling, and iron dextran supplementation alleviated UUO-induced renal fibrosis. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| GenFocal is an AI framework that generates fine-scale weather from coarse climate projections without requiring temporally paired simulated and observed events during training, and it synthesizes long-lived hazards such as heatwaves and tropical cyclones. | Insufficient evidence | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"SUPPORTED","ledger-default@0.2.0":"INSUFFICIENT_EVIDENCE","ledger-strict@0.2.0":"INSUFFICIENT_EVIDENCE","ledger-default@0.3.0":"INSUFFICIENT_EVIDENCE","ledger-strict@0.3.0":"INSUFFICIENT_EVIDENCE"}} | One independent, direct measurement that supports this claim would move it from Insufficient evidence to Supported. |
| A harmonized Arctic aerosol analysis using observations spanning 2003–2024 identified three regimes with contrasting contributions of new particle formation to cloud condensation nuclei populations: the High Arctic, Maritime Arctic and Continental Arctic. | Insufficient evidence | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"SUPPORTED","ledger-default@0.2.0":"INSUFFICIENT_EVIDENCE","ledger-strict@0.2.0":"INSUFFICIENT_EVIDENCE","ledger-default@0.3.0":"INSUFFICIENT_EVIDENCE","ledger-strict@0.3.0":"INSUFFICIENT_EVIDENCE"}} | One independent, direct measurement that supports this claim would move it from Insufficient evidence to Supported. |
| Aging reduced cardiac lymphatic-vessel density in humans and mice, and experimentally reducing cardiac lymphatics in young mice recapitulated inflammation and impaired lymphatic integrity associated with cardiac aging. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| In aged-heart models, cardiac Vegfc overexpression or Il33 silencing restored lymphatic density and tissue homeostasis, identifying the VEGFC–IL-33 axis as a potential target for age-related cardiac dysfunction. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| In manipulative grassland frost experiments, warmer temperatures and higher species diversity strengthened vegetation's post-frost compensatory responses. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Leans contradicted. |
| HD 3167 b's dayside is cooler than the maximum expected for an airless world, providing evidence for an atmosphere. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported. |
| The study reported that reef-building coral microbiomes had biosynthetic potential rivaling or surpassing traditional natural-product sources such as sponges. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"SUPPORTED","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"LEANS_SUPPORTED","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"LEANS_SUPPORTED"}} | One independent, direct measurement that contradicts this claim would move it from Leans supported to Leans contradicted. |
| In the study, deleting RANKL in lymphoid tissue organizer cells caused an almost complete absence of Thetis-cell subset TC I. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In Arabidopsis seedlings, MCA1 mechanosensitive channels contribute to long-distance wound-signal communication between lateral roots. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In the five-Tohoku-prefecture dataset, bear conflict rises sharply when mast production falls below a threshold. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In the study's analysis, bear conflict is further amplified when an alternative food source fails in the same period as mast production. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In the reported perovskite photovoltaic structure, an ultrathin conformal SnOx interlayer suppresses interfacial reactivity between NiOx and the perovskite. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In the model and empirical motivation, contractionary monetary policy raises US housing rents even as house prices fall because tighter credit shifts demand toward renting and rental supply reallocates slowly. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| The model reports that responding to shelter-inclusive inflation can increase the output cost of disinflation when tightening itself raises rents, while excluding shelter lowers that cost. | Leans supported | {"states":{"ledger-default@0.1.0":"LEANS_SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| The study found that oyster settlement and survival were greatest at particular combinations of reef fractal dimension and height that minimized predation. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| The study's mechanistic investigations suggest that the deoxygenative functionalization begins with two-electron oxidative addition of a C–O bond to a low-valent chromium species, forming an organochromium intermediate. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In the study, water-derived hydroxyl-radical formation was identified as the rate-determining step in polyethylene oxidation, with a kH/kD isotope effect of 7. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In the study, hindbrain activity was necessary for mesenteric-artery constriction and gut blood-flow reduction during hypoxia in zebrafish. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| The study's model attributes higher observed polarization under lower turnout to underrepresentation of centrist voters and overrepresentation of ideological extremes when a centrist governing coalition is expected. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| In a six-month randomized trial, the medically tailored-groceries plus telehealth intervention group had a larger HbA1c decrease than the usual-care group, 0.66 versus 0.44 percentage points. | Unresolved | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"SUPPORTED","ledger-default@0.2.0":"UNRESOLVED","ledger-strict@0.2.0":"UNRESOLVED","ledger-default@0.3.0":"UNRESOLVED","ledger-strict@0.3.0":"UNRESOLVED"}} | |
| The study mapped six natural and seminatural open wetland types across large parts of Europe at 10-m resolution, estimating 413,504 ± 17,782 km2 with overall accuracy of 96.6 ± 0.3%. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"SUPPORTED","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"LEANS_SUPPORTED","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"LEANS_SUPPORTED"}} | |
| The ETUCARE study reported that emotional-regulation skills mediated the relationships between intervention assignment and both procrastination and study engagement. | Supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"NOT_APPLICABLE","ledger-default@0.2.0":"SUPPORTED","ledger-strict@0.2.0":"NOT_APPLICABLE","ledger-default@0.3.0":"SUPPORTED","ledger-strict@0.3.0":"NOT_APPLICABLE"}} | |
| The study estimates that anthropogenic disturbance affects 20.4 ± 3.4% of Europe’s wetland area, with 13.6 ± 3.0% classified as most disturbed. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"SUPPORTED","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"LEANS_SUPPORTED","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"LEANS_SUPPORTED"}} | |
| Under 2018 conditions, land-use disturbance corresponds to a potential European wetland soil-carbon storage difference of 0.8–4.9 Gt CO2-equivalent, not an annual emissions rate. | Leans supported | {"states":{"ledger-default@0.1.0":"SUPPORTED","ledger-strict@0.1.0":"SUPPORTED","ledger-default@0.2.0":"LEANS_SUPPORTED","ledger-strict@0.2.0":"LEANS_SUPPORTED","ledger-default@0.3.0":"LEANS_SUPPORTED","ledger-strict@0.3.0":"LEANS_SUPPORTED"}} |
high impact insufficient (0)
None.
disputed audits (0)
None.