Back to the episode map

Research Note

Different AI Systems Evidence Matrix

| Question | Scientific prediction system | Multimodal assistant | |---|---|---| | Claimed job | Predict a defined scientific object or relationship | Help a person compl

Aug 4, 20262 min readBy Dalton Anderson

Different AI Systems Evidence Matrix

QuestionScientific prediction systemMultimodal assistant
Claimed jobPredict a defined scientific object or relationshipHelp a person complete an interaction or knowledge task
Core evidenceData boundary, comparator, metric, uncertainty, reproducibility, external validationRepresentative tasks, error classes, latency, privacy, consent, recovery, accepted work
Typical overclaimBenchmark becomes clinical or commercial outcomeDemonstration becomes universal availability or workflow reliability
Domain reviewerScientist with relevant subject and method expertiseQualified task owner plus privacy, security, accessibility, and domain reviewers
Deployment evidencePerformance under intended scientific or regulated conditionsComplete human-system workflow under realistic conditions
Stop conditionUncertain, non-generalizable, or decision-inadequate predictionUnacceptable failure, review burden, data risk, or poor recovery

The shared word "performance" hides incompatible claims. AlphaFold 3's paper evaluates structure-prediction tasks against defined datasets, metrics, and comparators. GPT-4o's launch materials combine model evaluations, product demonstrations, staged rollout, and vendor latency measurements.

The comparison can still be useful if it starts with the decision. A scientific claim needs evidence that the target, test set, metric, comparison, uncertainty, and generalization match the intended scientific use. An assistant claim needs evidence that the actual user can complete the actual task with acceptable error, review, data handling, latency, accessibility, and recovery.

NIST's AI RMF supports context-specific testing, evaluation, verification, and validation. The AlphaFold 3 paper and GPT-4o system card supply system-specific evidence. None of those sources certifies a deployment.

Sources

Follow the evidence.

  1. Google DeepMind about pagedeepmind.google
  2. NIST AI RMF Measure guidanceairc.nist.gov
  3. Google DeepMind AlphaFold 3 launchblog.google
  4. Google DeepMind AlphaFold pagedeepmind.google
  5. Isomorphic Labs company siteisomorphiclabs.com
  6. OpenAI about pageopenai.com
  7. Current GPT-4o API documentationdevelopers.openai.com
  8. GPT-4o system cardcdn.openai.com
  9. FDA machine-learning transparency principlesfda.gov
  10. AlphaFold 3 papernature.com
  11. GPT-4o ChatGPT retirementopenai.com
  12. GPT-4o launchopenai.com
Different AI Systems Evidence Matrix