Quantum machine learning gets enormous hype but most researchers who study it closely are pretty sceptical

[Matt_81]

The pitch for quantum machine learning sounds compelling: use quantum computers to train AI models faster and find patterns classical systems miss. The problem is that when researchers have gone looking for a genuine quantum speedup on realistic machine learning tasks, they mostly have not found one. A 2024 analysis found that many proposed quantum machine learning algorithms offer at most a modest advantage, and in several cases classical algorithms can match or beat them once you account for the overhead of running quantum hardware. The field is not fraudulent, there are legitimate open questions, but the gap between the marketing and the published evidence is wider here than almost anywhere else in quantum computing

[QuantumToken98]

Refreshing to see someone say the quiet part out loud. QML hype has been running years ahead of QML results for a long time now

[One-One-Five]

The overhead point is crucial and almost never mentioned. Even if the quantum algorithm is theoretically faster, the cost of running quantum hardware can wipe out the advantage

[Arty Kayla]

I would not write it off entirely. There may be specific ML subproblems where quantum helps a lot even if general quantum ML is oversold

[Cheeky Blake]

The investors pouring money into quantum AI startups based on press releases rather than papers are going to have a rough few years ahead