#AI models based on historical customer behavior struggle in this unprecedented #Covid crisis. How to adjust them? • Deep dive & precise monitoring of the data • Retraining on the latest available data • Adaptation of targets & messages Via KDnuggets kdnuggets.com/2020/04/ai-mac…

Very proud to reveal the history of Neon, N26‘ #chatbot, that uses state of the art #AI and #NLP to serve our customers 24/7 in 5 languages 🤖 Kudos to all the N26ers involved in this initiative and to our partner Rasa! medium.com/insiden26/hist…

Meet Neon, our Customer Service #chatbot: 2 years old, speaks 5 languages and knows all about banking! 🤖💳 #n26 #AI #operations #InsideN26 medium.com/insiden26/hist…

How Philip Sheppard (composer) & Tom Gruber (Siri) created LifeScore to augment #music with #AI by combining & creating unique compositions based on user biometrics, environment & emotions. Many use cases from video games to ambiant music Via WIRED wired.com/story/ai-maest…

7 suggestions to succeed as a Data Science Product Manager, among which: • Understand how & when to use #ML • Adapt your #agile framework to uncertainty & experimentation • Carefully chose your model KPIs & constantly monitor results Via Towards Data Science towardsdatascience.com/on-being-a-dat…

How can state of the art #AI and #NLP help support research on #CoVID19? By providing a smart search engine for scientific papers: covid19.ai2prod.com Kudos to Ahmed Besbes , olivier Auliard and the Capgemini Invent teams 🚀 m.youtube.com/watch?v=iVsteY…

#AIOps is a key success factor for any #datascience and #AI initiative, which includes a state-of-the-art environment and a multidisciplinary team to manage (re)training & deployment, integration, monitoring and scaling of the engines Via Harvard Business Review hbr.org/2020/06/the-du…

#google’s latest attempt to enter the social media space: #Keen, a curation platform, similar to #Pinterest, but relying more on #AI to identify and suggest relevant content Via The Verge theverge.com/2020/6/19/2129…

How can #AI improve new #product development? • Trend discovery • Demand forecast & propensity models • Generative #design • Optimization of offers & bundles • Customer feedback & #insights Via Forbes www-forbes-com.cdn.ampproject.org/c/s/www.forbes…

Great interview of S. Paulenz from N26, explaining how we created Neon, our customer service #chatbot, and how we are managing its content evolutions Big shout out to all the teams involved and of course to Rasa our key partner on this journey 🤖💬🤝 m.youtube.com/watch?v=N_ZGzD…

Comparison #GPT3 vs. rasa.io #chatbot to answer questions about movies: • Rasa: more accurate answers & easier to fine-tune / retrain • GPT-3: good results out-of-the-box for a very large variety of #NLP use cases #Transformers revolution on its way! link.medium.com/vucyVMNBZ8

Are #GAN and generative #AI “just” tools or independent entities? New Massachusetts Institute of Technology (MIT) study stressing the need to distinguish myths from reality, to give more credits and responsibilities to human stakeholders. Via ZDNET www-zdnet-com.cdn.ampproject.org/c/s/www.zdnet.…

New #android feature to improve #CX & reduce waiting time when calling #customerservice, by using #AI to: • Detect hold musics & loops • Alert the user when a representative is on the line • Ask the representative to hold the line if needed Via Google blog.google/products/pixel…

Focus on 4 #deepfake projects using different #AI technologies (dialogue replacement, voice cloning and face swapping), to analyze the new possibilities they create for actors and creatives but also the ethical questions that arise Via MIT Technology Review technologyreview.com/2020/10/09/100…

Researchers from University of Waterloo & Massachusetts Institute of Technology (MIT) invented “LO-shot learning” to build #AI models with very few examples by distillating training data - e.g. hand-written number recognition trained on 10 ex. reaching 94% accuracy on the #MNIST dataset! Via MIT Technology Review technologyreview.com/2020/10/16/101…

Yesterday I shared with Sciences Po students how we use #data #analytics and #AI at N26 to create the best #banking experience for our customers. Thanks Ahmed Abbessi for this great opportunity!

#AI is now a key asset in #customerservice by predicting customer needs, handling requests with #bots or assisting agents with #insights. This allows #CS functions to develop from #support capabilities to real #retention & #growth engines Via MIT Technology Review technologyreview.com/2020/09/17/100…

L’#IA dans ses formes actuelles amplifie-t-elle les inégalités, notamment de #genre ? Peut-elle au contraire contribuer à les réduire ? Quelques pistes dans mon article pour #PAM Merci à Ponts Alumni pour cette tribune 🤖👥🌉 —> English version coming soon!

Can #AI in its current forms either amplify or reduce #gender #inequalities? Some lines of thought in my latest #Medium post - also available in the Feb. edition of Ponts Alumni Magazine medium.com/@jp.sleiman/ai…

Salim Laïbi - LLP Quand l'eau devient haram ou makruh ? L’eau devient non halal si elle est polluée par : Des impuretés physiques (urine, excréments, sang, cadavres, etc.) → elle devient najis (impure). Des substances intoxicantes (alcool, drogues) → même en petite quantité, elle devient haram.