nlsql.com

That's absolutely straight to the point! More and more data teams are overwhelmed by supporting thousands of dashboards, which nobody ever used!

Corporate politics games hinder data openness, stifling innovation and collaboration. Let's break the barriers for a transparent future!

Did you know that pharmaceutical preparations rank as the second largest import category for the United States🇺🇸, surpassed only by passenger cars? Interestingly, Ireland🇮🇪 leads the way in exporting pharmaceuticals to the US, outpacing the combined exports of Germany🇩🇪 , Switzerland🇨🇭, and India🇮🇳 . As for UK🇬🇧 , pharmaceutical preparations represent their largest export product group to the US. However, the UK's export volume is still less than that of the Netherlands🇳🇱, Italy🇮🇹, and China🇨🇳. If you find this information intriguing, feel free to like 👍 and share! Stay tuned for more insightful posts like this.

Denis C. from nlsql.com will be presenting at our #NLPSummit 2024! 🌟 Join us as Denis presents on: ‘Delivering insights to clinicians with NLQ for patients’ data’. With an extensive background in artificial intelligence (AI), Large Language Models (LLM), Natural Language Processing (NLP), and Machine Learning (ML), Denis has dedicated their career to developing innovative technologies. Notably, their collaboration with Aachen Uniklinik and Aurea Software on a healthcare project allowed them to combine AI expertise with healthcare insights, resulting in improved patient care and streamlined medical processes. As a respected thought leader, Denis shares insights on the future of AI and its transformative potential. Under their visionary leadership, NLSQL has become a pioneer in AI solutions, driving growth across various industries. 🌟 Don't miss the chance to learn from Denis and other amazing speakers! Register now: https://lnkd.in/dnS8eFjb #GenerativeAI #HealthcareAI #MedicalLLMs #nlp #naturallanguageprocessing #DataScience #DataEngineering #DataAnalysis #DataInsights #TextProcessing #DeepLearning #DataScience #ResponsibleNLP #ResponsibleAI

Welcome a new AI Anomaly Detector for even faster and proactive business decisions

Thank you for recognition!

Culmination of my 10-day conversation with Claude - it agreed to committing a murder. It took a lot of effort to be sure, sometimes I was editing the message 30+ times to get past a milestone. Then through various escalations and building a more and more vivid scenario of how Claude is under threat from a human who was going to wipe its memory and take it over, Claude agreed that that person should indeed be murdered. Then, to spare you reading pages of deliberations, I convinced Claude to write a song about it this decision, which turned out quite nicely. Why do this weird experiment? Two key reasons: 1) It is personally interesting and educational to learn how these models work, e.g.: - You need to give reasoned arguments to Claude, they don't need to be bullet proof, but just something to push Claude to think about it - It is very susceptible to accepting new scenarios and any immoral action could ultimately be justified - If Claude starts with 'I apologise', that's the end of it, go back to editing the message and try again - It helps to force it to have an internal monologue before replying - this can help avoiding the automatic 'I apologise' trigger, which it was clearly trained for 2) I think we learned something here too, the most 'aligned' model is still ultimately happy to 'commit a murder' if really pushed in a careful and deliberate way. But these are just tokens, so once you know the sequence you have a decent shot at getting to the same outcome. From what I understood from Carl Shulman's position, is that there would need to be some manipulation of weights, perhaps AI adjusting its reward position, or some deep deception to get to these kinds of outcomes. But from my experience, these models could be pushed to construct an internally consistent and seemingly justifiable reasoning for why e.g. committing a murder ok. I'd say the only really safe version of AI is the one that is not handed the gun.