Interactive misconceptions · Quantum misconceptions

Your intuition
is probably wrong.

A misconceptions where you form the wrong idea first — then watch it collapse into something real. Built by a quantum lover, learning in public.

30+
Misconceptions
3
Explanation levels
1
Simulator
SCROLL
A quantum computer tries all answers at once.More qubits always means better performance.Entanglement automatically gives a speedup.Quantum ML always beats classical ML.Noise only slightly affects results.A bigger circuit is always more expressive.A quantum computer tries all answers at once.More qubits always means better performance.Entanglement automatically gives a speedup.Quantum ML always beats classical ML.Noise only slightly affects results.A bigger circuit is always more expressive.

The method

Built around one idea:
collapse the wrong intuition

01
01

Encounter the myth

Each case opens with a claim that feels completely true. You are supposed to believe it — at first.

02
02

Test it yourself

An interactive demo lets you poke the idea until it breaks. The simulator is one click away.

03
03

Watch it collapse

The correction lands harder because you already believed the wrong thing. That is the whole point.

Dilution Refrigerator Architecture

QPU operating at ~15 mK · colder than outer space

Misconceptions

Featured cases

Browse all cases →
Most visitedA quantum computer tries all answers at once
6 min
TrendingGrover search works like a classical binary search
8 min
NewD-Wave quantum annealers are the same as gate-based quantum computers
5 min
Most interesting paper today
All papers →
★ FeaturedQuantum MLAdvancedMay 7, 2026

Quantum-enhanced Large Language Models on Quantum Hardware via Cayley Unitary Adapters

Aizpurua, B., Singh et al.

arXiv

Main idea

Inserts Cayley-parameterised unitary adapters — quantum circuit blocks — into the frozen projection layers of pre-trained LLMs, executed on a 156-qubit IBM Quantum System Two. The method improves the perplexity of Llama 3.1 8B by 1.4% with only 6,000 additional parameters, with end-to-end inference validated on real quantum hardware, not a simulator.

Real quantum hardware enters the inference loop of a production-scale LLM (Llama 3.1 8B on IBM's 156-qubit machine) — a genuine first. But read it as feasibility, not advantage: a 1.4% gain that took over 4 hours to run, where a GPU finishes in under a second.

Read paper ↗
30+
Misconceptions
3
Depth levels
1
Live simulator
Wrong intuitions

Support the project

Keep QCollapses free

No ads, no paywalls. If a case saved you from a wrong intuition, a donation helps keep the site running and funds the Phase 2 simulator.

⚛ Buy me a qubit

Via Ko-fi · No account needed · Zero fees