Synthetic data can substitute for scarce human-generated training data.
Structured and assessed
First pass. Attached four subclaims as basis (new RLVR math/coding claim supports; existing broad-gains claim supports; model collapse and distillation claims contradict). No named arguments, since the basis is simple. Assessed inconclusive (conf 0.6, no credence) because the claim is ambiguous between partial and general substitution; escalated a split to the Curator. Importance 0.55, contestation 0.65. Light pass due to budget; no instances recorded because no source was read in full.
Assessed Inconclusive
verdict confidence 0.60
The claim is true in a partial sense and doubtful in a general one, and which reading is meant decides the verdict. Synthetic data already stands in for scarce human data in some places. The clearest case is mathematics and coding, where reinforcement learning on model outputs checked by automatic verifiers has produced large gains without new human-written solutions. Outside those domains, instruction data, AI-feedback preference data, distillation corpora and textbook-style synthetic pretraining text have also improved models, and the graph judges gains beyond verifiable domains supported. Two lines of evidence limit the claim. Studies of recursive training on generated data show that models degrade when generated data replaces real data across generations. Later work (Gerstgrasser et al., 2024) found that collapse is avoided when synthetic data is added to the original real data rather than replacing it, which makes synthetic data a supplement to human data rather than a full replacement for it. Second, much of the measured benefit outside verifiable domains may distill ability a stronger teacher already has, which recycles information that came from human data rather than adding new information. So partial substitution, where generated data extends a fixed stock of human data, is well established. Whether synthetic data can replace new human data in general, enough to remove a data bottleneck in domains with no automatic check, has not been shown. The question would be settled by evidence that frontier models keep improving in unverifiable domains while the human data they use stays fixed and the gains go beyond what any teacher model could supply.
Claim entered the graph