Quantum Measurement Problem
Why does a measurement always show a definite result although quantum theory allows superpositions? Where does quantum behavior end?
Open in the interactive tree →The Schrödinger equation predicts superpositions ('cat both dead and alive'), yet we never see half-states. Interpretations such as Copenhagen, many worlds and Bohmian mechanics make the same predictions, while collapse models (GRW, CSL) change the theory and can be tested. No experiment has decided between the options so far.
As of October 2026
October 2026: no consensus. Experiments push the boundary upward: Vienna interferometers let molecules of about 2,000 atoms (more than 25,000 mass units, 2019) interfere, and a Nature paper of January 2026 did the same with sodium nanoparticles of 5,000 to 10,000 atoms and more than 170,000 mass units, setting the strictest limit so far on generic modifications of quantum mechanics. Underground tests at Gran Sasso excluded the parameter-free Diósi-Penrose gravity collapse model and parts of the CSL parameter space, but collapse models with other parameters survive.
What is missing
- Experiments with much larger objects (microgram scale, 10^9 mass units and more) held in superposition
- Shielding against decoherence: vacuum, cold, vibrations
- A decisive prediction that separates the interpretations
- Understanding the role of gravity and of the observer
Becomes possible once solved
- A definite answer on whether quantum mechanics holds at all scales
- Better quantum computers and sensors through understanding decoherence
- Possibly a bridge to a theory of quantum gravity
Open steps
- Superposition of far heavier objects Low AI leverageInterfere particles far heavier than today's record of about 170,000 mass units, to test where quantum behavior may end.
- Shielding against decoherence Medium AI leverageCut vibration, thermal and gas-collision noise enough to keep large objects coherent long enough for interference.
- Collapse-model parameter space Medium AI leverageClose or confirm the remaining parameters of CSL and gravity-related collapse models against all data from interferometers and underground tests.
- Predictions that split interpretations Low AI leverageDerive an experimentally testable difference between interpretations (many worlds, Bohmian, collapse), which today make identical predictions.
Where AI could help
Low AI leverage. The open question is which interpretation is right; that needs larger superposition experiments and theory, not more computation.
- Design optical and optomechanical experiments that put heavier objects into superposition
- Scan collapse-model parameter space against all existing data (Gran Sasso, Vienna interferometers)
- Model decoherence and noise sources to specify vacuum, cooling and vibration shielding
Shown so far
- In October 2022 (published in Quantum in 2023) the PyTheus AI framework discovered 100 distinct designs for photonic quantum experiments, such as entangled states, gates and communication protocols. source
Prerequisites
- Quantum Mechanics1925
- Bell Test of Entanglement1964Bell tests rule out local hidden variables and sharpen the measurement debate
- Quantum Tunneling in a Circuit2025