Emergent Methods

Voir is forecasting a narrow Trump win tomorrow. Check out the research and rationale at the interactive dashboard: Forecast Dashboard: https://lnkd.in/d52hrpnU #presidentialelection #prediction #forecast #predictionmarket #polymarket #predictit #metaculus #manifold #kalshi #trump #kamalaharris #election

Need a last-minute comparison of presidential policies before you head to the polls? We got you covered. Share-link: https://lnkd.in/d75frYwb #asknews #newsplunker #presidentialelection #policies #ainews #knowledgegraph

Analysts asked us, and we listened. At AskNews.app, we already let you "talk to a specific news site" or "talk to objective news" or "talk to positive news". But what about "talk to a geographical region?". Layer that into the world's largest news knowledge graph - holy smokes 🚀 Adding geo-coordinates to a database is no easy feat - imagine you have 100s of millions of news articles in an archive, and you need to find which locations each one talks about and then assign *accurate* geo-coordinates to each one. Next, imagine you need to do it with high-throughput in real-time, and index millions of new coordinates every single day. Our latest feature, geo-coordinates, would be impossible without Qdrant, Pelias Geocoder, Meta (Llama), and Flowdapt.ai. OSS #FTW. This is months of research into quality assurance, scalable engineering, and just plain passion. It is what we do best Emergent Methods. Look out for followup features, including a beautiful real-time geographical tracker in Newsplunker, high performance filtering, documentation, examples and much more. #geocoordinates #latitude #longitude #pelias #qdrant #llama #meta #flowdapt #asknews #newsapi

Human in the loop and quality control isn't going anywhere. Here is a "news article" from Archynewsy that shows that they aren't using any human in the loop or quality control mechanisms before "publication". Instead, we see exactly what their values are since their prompt has leaked in the output. They only care about SEO. ⚠️ Notice how you see nothing regarding journalistic integrity. Their prompt has nothing about "identifying attribution", "reporting evidence", "explicitly pointing out allegations/accusations". While this is a glaringly obvious example - what about when their LLM *does* generate something that has a semblance of journalism? Clearly, it hasn't passed any quality checks or human in the loop - whatever is coming out is hallucination prone. Delete this website from your news diet, now. This is precisely why we at AskNews prioritize journalistic integrity with our human in the loop in-house editor-in-chief and our dedication to trust and transparency protocols. With https://asknews.app, hallucination is minimized and journalistic integrity is maximized. I'm curious to hear your thoughts on this Jon Accarrino, Todd Carter, Pete Pachal Link to garbage article: https://lnkd.in/dGSJK5UE #garbageingarbageout #ainews #archynewsy #llmnews #newsapi

Missed our Qdrant coffee chat about Prototyping, Pivoting, and Perfecting products? Check out the recording here! https://lnkd.in/deHTCXn5 #vectordb #vectorsearch #asknews #newsplunker #artificialintelligence

Let's look beyond CNN and Fox News, shall we? Instead, we introduce hundreds of competing perspectives to help map out how Project 2025 and the Heritage Foundation aim to affect change 🌍 Interactive dashboard: https://lnkd.in/dZ-3n6TX Inform your vote 🗳 #project2025 #heritagefoundation #donaldtrump The Heritage Foundation #competingperspectives #ai4news #llm #graphrag #graphchat #newsplunker #cnn #foxnews #vote

What are you up to this Friday at 1 pm EDT? Nothing? Great! Join us for a coffee as we chat about prototyping products with Qdrant 😎 https://lu.ma/3z3kcdw1 #vectordb #vectorsearch #vectorspace #qdrant Emergent Methods

Anthropic just released a new version of Claude 3.5 Sonnet, which can use your computer to perform tasks. While this sounds cool, my first thought is “What are the safety measures?” The announcement (link: https://lnkd.in/dVYbkPge) contains a section on safety that points out that yes, this setup is prone to safety issues. The docs (link: https://lnkd.in/dUjDVHAd) start out with a Caution, listing important aspects to consider before using it, and Anthropic do state that they have “put in place measures to monitor when Claude is asked to engage in election-related activity, as well as systems for nudging Claude away from activities like generating and posting content on social media, registering web domains, or interacting with government websites.” One of the safety risks they point out is prompt injection, stating “Since Claude can interpret screenshots from computers connected to the internet, it’s possible that it may be exposed to content that includes prompt injection attacks.” Indeed, with nearly all of the Google image results for “baby peacock” being AI-generated (try it yourself to see), it’d not be unthinkable for malicious entities to use adversarial encoding for prompt injection purposes. Then there is also the aspect of allowing Claude to capture what ever is on your screen and sending that to Anthropic’s servers. Anthropic do state that “by default we don’t train our generative AI models on user-submitted data, including any of the screenshots Claude receives.” but that is something I personally will never fully trust coming from a company making a living off of training closed-source models. My second thought, after having seen one of the demo videos (link: https://lnkd.in/dZt55jdd) is “Why not just use an agentic framework?” The execution speed would be magnitudes higher and proper guardrails could much easier be put in place. The drawback would be access to services that don’t have APIs, but with the plethora of alternatives currently available I don’t see how efficient workarounds would not be easy to build. So, will I be allowing Claude 3.5 Sonnet to use my computer? Not in the foreseeable future. But hey, maybe I’m too skeptical. #claud35sonnet #anthropic #AIsafety

Ready to go Newsplunking for energy market forecasts? Check out our interactive news knowledge graph: Dashboard link: https://lnkd.in/dC9JSbqM If you want to gain access to this tool to build your own custom topics and analyses, join us for our live demo in our Discord server on Thursday at 16h00 UTC: Event: https://lnkd.in/dWPSxem2 where we will hand out special attendee-only promo-codes! #osint #ainews #knowledgegraph #energymarket #forecasts #renewables #newsplunker #asknews Emergent Methods Connected Data GraphGeeks

Want to know what it is like to chat with the largest news knowledge graph on the planet? Or one step further - what is it like to generate your own real-time forecasts from the largest news graph on the planet? We are doing a live demo of our forthcoming tool, Newsplunker, in the AskNews Discord server on October 24th at 16h00 UTC (12 pm EST). Attendees will receive a special one-time promo-code for early access to the tool 💎. Link to event: https://lnkd.in/dWPSxem2 #newsplunker #graphrag #knowledgegraph #llm #rag #ainews #asknews