COMPARISON LAB

AI developer productivity: reconcile the evidence

Reconcile Google's product experiments, the METR RCT, Meta DAT construct validity, and SUSVIBES benchmark validity. Produce an AI adoption memo with explicit confidence bounds.

30-MINUTE PAPER LAB

Read the original

The timer helps pace the session, but never locks reading or changes stages for you.

00:00

active time

Saving is off. Answers remain in memory until this tab closes. Turn saving on

1.1What single AI-tool adoption decision do you want to make from the four studies? Record your prior expectation and the conditions under which the effect may differ.*

05 min

Engineering memo

Up to 250 words: decision, evidence, confidence, limitations, applicability, and next step.

Decision *
Evidence *
Confidence *
Limitations *
Applicability *
Next step *

0 / 250 words

Complete lab

Answer or skip required prompts, review the references, and save a memo.