Testing the new Gemini 1.5 Pro Experimental 08-01 model I am impressed with its general capabilities, but frustrated with its inability to manage changing state within an ongoing conversation (such as is natural during a coding exercise). It also has an annoying tendency to gaslight the user with hallucinations about being able to improve -- these seem like naive missteps in the fine tuning by Google and is reminiscent of earlier releases by OpenAI and Anthropic.
1/2
Gute Nachrichten für gesellschaftliche Initiativen, die vom rechten Juristenmob und in Sachsen vom Landesrechnungshof wegen politischer Haltung angegangen werden (Gemeinnützigkeit/ Förderung):
https://www.cellex-stiftung.org/de/news/rechtsgutachten
> Der…