How the media write about AI and tech · over the last 3 days

For every outlet that writes about AI and tech, a language model rates the tone of its articles as positive, neutral or negative. That gives a tone index from 0 to 100, where 50 is balanced. At the top are the outlets that frame the news most positively, at the bottom the most negatively.

#	Outlet	Articles	Tone index	Pos / Neg
1	9to5google.comArticles 50 · Pos / Neg 70% / 8%	50	90Strongly positive	70% / 8%
2	siliconangle.comArticles 42 · Pos / Neg 83% / 14%	42	85Strongly positive	83% / 14%
3	engadget.comArticles 39 · Pos / Neg 77% / 15%	39	83Strongly positive	77% / 15%
4	9to5mac.comArticles 70 · Pos / Neg 69% / 16%	70	81Strongly positive	69% / 16%
5	digitimes.comArticles 56 · Pos / Neg 71% / 18%	56	80Strongly positive	71% / 18%
6	androidauthority.comArticles 115 · Pos / Neg 56% / 16%	115	78Strongly positive	56% / 16%
7	thenextweb.comArticles 38 · Pos / Neg 58% / 26%	38	69Positive	58% / 26%
8	techradar.comArticles 88 · Pos / Neg 56% / 30%	88	65Positive	56% / 30%
9	tomshardware.comArticles 45 · Pos / Neg 60% / 33%	45	64Positive	60% / 33%
10	semafor.comArticles 56 · Pos / Neg 41% / 41%	56	50Neutral	41% / 41%

Showing outlets with ≥38 rated articles. 10 of 52 ranked. How it is calculated →

This is the tone of the articles as a model reads them, not a rating of an outlet's quality, reliability or bias. A high index often only means an outlet writes more about rallies and launches, a low one more about sell-offs, hacks and lawsuits.