Weekly AI and quantum AMA thread, ask anything you have been wondering about

[RightAbout24]

Kicking off this week's open thread for any AI or quantum computing questions you have been sitting on, big or small, basic or deep in the weeds. No question is too simple here, plenty of people reading are just getting into these topics and the whole point of this thread is a judgment free space to ask

A few things that have been generating a lot of questions elsewhere this week if you need inspiration, the ongoing debate over whether open weight models are actually catching up to closed frontier systems, what post quantum cryptography actually protects against and why it matters now rather than later, and the recent classical tensor network result that challenged a quantum supremacy claim. Feel free to ask about any of that or anything else entirely unrelated

Regulars, please be patient and kind with newer folks asking basic questions, we were all new to this stuff once. Drop your question as a reply and hopefully someone in the community can help you out

[Hannah]

What is actually the difference between a large language model and what people mean when they say AGI? Feels like the terms get used interchangeably a lot

[TinyCompass]

Roughly speaking, an LLM is a specific kind of AI trained to predict text, while AGI usually refers to a hypothetical system that can match or exceed human level performance across essentially any intellectual task, not just language

[Glenn]

Is quantum computing ever going to be something regular consumers actually use, or is it purely an enterprise and research tool forever

[David]

Most experts seem to think it stays a specialized enterprise and research tool for the foreseeable future, the use cases like drug discovery and optimization do not really translate into anything a home user would need

[Aura]

Why do so many AI companies keep releasing open weight models for free, what is the actual business incentive there

[PlanetOftheApes]

A few reasons, building developer goodwill and adoption, undercutting competitors pricing power, and sometimes national strategic goals around AI leadership rather than pure direct profit from that specific model

[HollowSentinel]

It generally means training a smaller model to mimic a bigger one's outputs, using the bigger model's responses as training data rather than starting completely from scratch

[Emma29]

Is there an actual difference between quantum supremacy and quantum advantage or are people just using them interchangeably

[Dylan]

Thanks everyone, this thread is exactly what I needed, way less intimidating than asking in a more technical subreddit

[Merchant31]

Supremacy usually refers to any task at all where quantum beats classical even if totally useless practically, advantage usually implies it is actually useful for something real, people do mix them up constantly though

[Wendy5]

Newbie question, what does it actually mean when people say a model was distilled from another one