What is machine learning?

[danwaterfield]

Article type

This is a conceptual article (what is...)

The title should be something like: What is AI?

Target audience

Humanistic researchers who've heard of Chat GPT, LLMs, deep learning, machine learning etc but don't understand the differences and difficulties.

What do people reading this article already need to know about, what background knowledge do they need, in terms of technical or domain concepts?
Nothing - the article will explain classical computing, symbolic AI, machine learning, deep learning, neural nets etc in high level terms.

What problem is the person reading this trying to solve in their research? Try to think of a concrete example situation why someone would need to know about/how to do this thing

They need to know what AI "is" and what it isn't; how fancy stats can be useful, and the challenges it poses in terms of training.

Things to cover

How do simple programs work?

Symbolic AI and Machine Learning / neural nets / supervised, unsupervised, reinforcement learning / different architectures / what's an LLM? What's generative AI? / Ethics

Is it related to anything else we've written about?

If you can, point to things we've written about (or are planning) that are useful as prerequisites for this article, or as follow-up.

Anything else?

Anything else we should know? Do you have links to anywhere on social media/blogs of people asking about this topic? Do you have links to relevant/further reading?

Next steps

Would you prefer:

• I want to write the article mostly myself

In either case, the next step is to wait for Dan or Caro to read the proposal and check it's something we want to include, then we'll take a branch and start working on an outline together. Thanks for your contribution 🎉

[caro401]

This seems like it might get really big and complicated - are there any sensible lines to break it down into several shorter, more specific articles @danwaterfield? Or to keep this one as a really broad high-level overview and link out to all the things in more detail in separate articles? Mostly for selfish reasons, because I find reading long things really hard, and the site design isn't amazing for really long prose. No is a completely valid answer, I don't know the topics at all

[danwaterfield]

I was thinking one very broad high level article which then links to increasingly granular articles. Taking notes at the moment to write a ‘hey why do people do data analysis in Python’ article which is similarly high level then drills down. I also think it might be cool to basically show the hierarchy of ‘what you need to know to do the basic stuff in x area’

…

On Wed, 21 Feb 2024 at 11:13, Caro ***@***.***> wrote: This seems like it might get really big and complicated - are there any sensible lines to break it down into several shorter, more specific articles @danwaterfield <https://github.com/danwaterfield>? Or to keep this one as a really broad high-level overview and link out to all the things in more detail in separate articles? Mostly for selfish reasons, because I find reading long things really hard, and the site design isn't amazing for really long prose. No is a completely valid answer, I don't know the topics at all — Reply to this email directly, view it on GitHub <#8 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIVHRYHQ6JLNS4ARZU3ZNF3YUXJERAVCNFSM6AAAAABDNQAKMSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNJWGQZDEOBQGU> . You are receiving this because you were mentioned.Message ID: ***@***.***>