"Perplexity user: “Microsoft banned your domain, so can't use it on work devices”."
A post from Aravind Srinivas
Posts
Filter by tag
hallucinationsocial-search-requestPerplexity-AIdissertationcomparing-resultsgenerative-searchaiextensionsjagged-frontierseries/extended-frontierworking-draftprompt engineeringsearch-auditsmicro interactions in searchYou.comspeculative-designGoogle-BardCGTbetter-searchcompetition-in-searchchoice-in-searchgenerative-web-searchknowledge-cutoffsalternative-search-enginesevaluating-results-metasearch-quality-complaintsstanding-queriessearch-engine-optimizationrepository-of-examplesAnthropic-ClaudeCFP-RFPRAGPhinddoubtingadvertising-in-searchseo-for-social-goodOpenAI-ChatGPTautomation-biassituated-practicepublic-interest-technologyspeculative-classroom-exercisesTavilysearch-user-interfacesgenerative-search-APIfresh-searchautocomplete-in-web-searchfalse-premiseTikTokAndievaluating-search-engines+236 more
"Personally I've found Bingchat with search, and GPT4 with Browse to be far more confused at all times than the no-RAG model"
A post from Teknium.
"Product human evals are what matter to us"
"Q: “What is Google doing” A: Selling ads."
A tweet from Aravind Srinivas.
"responsibility to test for real world applicability"
"Scheherazade, you easily make up stories of Egypt or any country you please."
Analysis of AI hallucination behavior by testing a Plato quote completion task across multiple LLM systems (ChatGPT, Claude, Bard, Pi, Perplexity, YouChat).
"Search has become ubiquitous but that does not mean that search has been solved."
"search terms that confirm preconceptions"
Re confirmation-biased queries.
"The different worlds of Google"
von Nordheim et al. (2023)
"the importance of open discussion of these new tools gave me pause"
An excerpt from Davey Alba's article, "Even Google Insiders Are Questioning Bard AI Chatbot’s Usefulness".
"The internal name for Deep Search was "Search Harder""
Brief weblink noting Bing's internal codename for Deep Search was 'Search Harder.'
"The robot is not, in my opinion, a skip."
Weblink to a Twitter thread arguing that physical automation shouldn't be dismissed as just another passing tech trend like crypto or metaverse.
"the text prompt is a poor UI"
This tweet is a reply—from the same author—to the tweet in: very few worthwhile tasks? (weblink). [ highlighting added] \@benedictevans In other words, I think the text prompt is a poor UI , quite sep...
"there's pressure for every LLM to have a live web connection"
Commentary on competitive pressure for LLMs to have live web access rather than static knowledge cutoffs.
"this is well within the model training (it has very likely seen a ton of code for pong)"
A post from Anton Bacaj.
"To Build Our Future, We Must Know Our Past"
Gururaja et al. (2023)
"Today marks the first year anniversary of the launch of perplexity.ai"
Weblink marking Perplexity AI's first anniversary with founder reflections in thread format.
"way more accessible to have a list of experiences to try"
A post from Nick Dobos.
"We don’t have relevancy ratings for search engines"
An excerpt from Sullivan (2017).