Whoops, that discussion is at https://mastodon.de/@ErikUden/114122514403706066

A morning discussion about a company that's creating LLM generated newsletters, and this comment about adding bots to a Skype channel is making me think about bot-free spaces, how to make sure that people aren't violating the social contract of creating and enforcing them. (Also, people must read way way slower than me to make Gmail's "summarize this email" remotely interesting. I, once again, do not get it.)

With the news that _Google is expanding the "AI Overviews" in search mode_ (https://blog.google/products/search/ai-mode-search/), via _Pivot To AI which mentions that you can remove all facts from your search for an extra $20/month_ (https://pivot-to-ai.com/2025/03/07/oops-all-ai-google-replaces-search-with-the-ai-overview/), it's worth pointing out this example of Google uncritically stating joke content as though it's real. _Wikipedia explains the joke_ (https://en.wikipedia.org/wiki/Wild_haggis).

This entire thread, from the top, but the observation that LLMs are most useful for applications where truth is irrelevant and lying is (more?) effective for achieving the goals is landing this morning. https://thepit.social/@peter/114121763629051621