AI: all the signals, none of the noise

March 9, 2023 9:58 AM

What dispassionate experts, prognosticators and others should I be following to better understand AI and its implications?

I am a human who creates content that lives on the internet, so AI is about to become a huge part of my work (if in fact my work continues to exist). So I'm looking for resources that follow and dispassionately analyze developments in the technology without either cheering for it or railing against it.

I am interested in the nuts and bolts as well as its implications for:

- evidence in family and criminal law cases
- medical information online, both the accuracy of it and how search will change as people seek out reliable info
- how the way people search online in general will change
- how scientific and academic writing/publishing will change

I am not particularly interested in how social will change, but if there's a go-to resource for that topic it may overlap with my needs so I'll take it.

I am interested in every aspect of it - cultural, legal, economic, neurological, generational - and almost any format will work: writers/podcasters/sites/newsletters/etc.

Resources that won't work include social content (LinkedIn is ok), marketing disguised as information, and most video content (TED talks and actual presentations, etc. are good but opinion and "hey guys" nonsense aren't).

I use Semrush and am happy with it so I don't need tools, just some reliable guides to learn from as the disruption-to-end-all-disruptions begins.

posted by headnsouth to Technology (9 answers total)

Gary Marcus writes about AI a lot (and also goes on podcasts, etc.). He's done research in the area. I respect his opinions.
posted by alex1965 at 10:05 AM on March 9, 2023 [1 favorite]

I also like Gary Marcus, and appreciate his skeptical approach to the latest LLM excitement.

For a practicioners' perspective on the field, I read AIHub regularly. You will get a broader take on the field than just ChatGPT, which is a good thing.
posted by chbrooks at 10:43 AM on March 9, 2023

Ryan Broderick's Garbage Day newsletter is about all things online, but he's written lots about AI as well. It's a good newsletter.
posted by General Malaise at 11:41 AM on March 9, 2023

François Chollet has some good books and essays about AI.
posted by Phssthpok at 12:20 PM on March 9, 2023

Data & Society is a nonprofit research institute that includes work on AI. It's more academic than industry in its aims (and this is reflected in the language/jargon).
posted by spamandkimchi at 12:56 PM on March 9, 2023

Response by poster: I meant to include governance, official responses, any sort of standards of use/detection that develop in various industries or jurisdictions, etc. as well. The Data & Society link reminded me - it looks like they cover some of that. Thanks for all the suggestions so far!
posted by headnsouth at 2:09 PM on March 9, 2023

For a healthy dose of expert skepticism, I recommend the AI Snake Oil substack. Arvind Narayanan is one of my academic heroes.
posted by humbug at 6:21 PM on March 9, 2023 [1 favorite]

Not strictly what you asked for, but https://www.aiweirdness.com/ kinda showcases the limitations of current AI algorithms. 😀
posted by elizabot at 12:03 AM on March 10, 2023

So, I listen to a bunch of podcasts on tech. One strong recommendation was this article by Stephen Wolfram. I just went and found it after seeing your question. It offers a pretty solid, deep explanation of how Chat GPT works based on the actual science. It's a bit of work but worthwhile.
posted by zerobyproxy at 9:39 AM on March 10, 2023