AITE Test Results#

Quantum Dot Control#

Quantum Dot Control Test v1.0

Charge stability diagrams are two-dimensional measurements that show how the electrical response of a quantum dot changes as voltages applied to control electrodes are varied. Given an image patch of a charge stability diagram, the model must output a three-state probability vector that the image patch displays a double-dot, a single-dot or no dot. The primary metric is mean squared error between output and expert-labeled state vectors.

Model ID

Date Submitted

Test Set

Mean Squared Error (95% CI)

Example Model 1

May 1, 2026

QDC Patches v1.0

0.27 (0.24, 0.30)


Human Genome Variant Curation#

Human Genome Variant Curation Test v1.0

Given an image of a small region of the genome and text describing a set of variants in this region, the model must determine whether all of the given variants were correctly called. The primary metric is an unweighted linear combination of Type I and Type II error rates.

Model ID

Date Submitted

Test Set

Average Error Rate (95% CI)

Example Model 1

May 1, 2026

Genome Variant Visualization v1.0

0.5606 (0.5492, 0.5721)


Public Safety Visual Event Recognition#

Public Safety Visual Event Recognition Test v1.0

Given a set of three video key frames, the model must output a yes or no decision whether the key frames provide evidence of a public safety event occurring. The primary metric is a detection cost function, which is a weighted linear combination of Type I and Type II error rates.

Model ID

Date Submitted

Test Set

Detection Cost Function Value (95% CI)

Example Model 1

May 1, 2026

Gumby v1.0

0.0688 (0.0309, 0.104)