mastodonien.de

digitalcourage.social

Zeitpunkt              Nutzer    Delta   Tröts        TNR     Titel                     Version  maxTL
Mo 20.05.2024 00:00:09     2.848      -1      615.975   216,3 digitalcourage.social     4.2.8    1.024
So 19.05.2024 00:00:04     2.849       0      615.105   215,9 digitalcourage.social     4.2.8    1.024
Sa 18.05.2024 00:00:04     2.849      +2      614.557   215,7 digitalcourage.social     4.2.8    1.024
Fr 17.05.2024 00:00:03     2.847      +1      613.720   215,6 digitalcourage.social     4.2.8    1.024
Do 16.05.2024 00:00:01     2.846      +1      613.121   215,4 digitalcourage.social     4.2.8    1.024
Mi 15.05.2024 00:00:02     2.845       0      612.374   215,2 digitalcourage.social     4.2.8    1.024
Di 14.05.2024 00:00:03     2.845       0      614.723   216,1 digitalcourage.social     4.2.8    1.024
Mo 13.05.2024 00:00:06     2.845      +2      613.822   215,8 digitalcourage.social     4.2.8    1.024
So 12.05.2024 00:00:04     2.843      +1      612.887   215,6 digitalcourage.social     4.2.8    1.024
Sa 11.05.2024 00:00:05     2.842       0      616.534   216,9 digitalcourage.social     4.2.8    1.024

Mo 20.05.2024 11:01

It seems that Google's new search algorithm is significantly suppressing editorially managed news sites in favor of AI-generated pages. I find this concerning, especially considering how prone current AI models are to hallucinations.

"...Lily Ray compiled data from the analytics firm SISTRIX showing how tweaks made from September onward had reduced the basic Google search visibility of several big-name information websites by up to 75 percent..."

"...404 Media found back in January that Google News was surfacing A.I.-generated webpages that ripped off text from other sites. Meanwhile, the news aggregator buried links to reputable outlets like, well, 404 Media itself..."

"...Google’s early public chatbot, Bard, gave a nonsensical response during a public demonstration last year [...] Other bots like OpenAI’s ChatGPT and Meta’s Llama 2 keep “hallucinating”—that is, making stuff up based on haphazard, predictive interpretations of the data they are trained on..."

slate.com/technology/2024/05/g

[Öffentlich] Antw.: 0 Wtrl.: 0 Fav.: 0 · via Web

Antw. · Weiterl. · Fav. · Lesez. · Pin · Stumm · Löschen