Info
Artikel von Dries Faems
Aktivitäten
24.892 Follower:innen
-
Dries Faems hat dies geteiltI wanted to test out the video generation capabilities of Opus 5.5 and the result really blew my mind You need to realize that this video was created from scratch with one single prompt. Opus 5.5 developed the scenario, collected and created the necessary images, composed the music and edited everything together Maybe the most surprising thing was that it was able to generate sarcastic humor about my desperate influencer efforts. Best quote: The thumbnails evolved...the blazer did not 😁
-
Dries Faems hat dies geteiltThis academic semester, we see the widespread adoption of agentic AI capabilities among students... Next semester, it will be multi-agentic capabilities. If you want to mentally prepare yourself for this, check out the Ultra thinking model in ChatGPT. In this particular thinking mode, you can already kick off a team of agents to execute complex workflows with one single prompt. In the article, I describe how I tested it on a literature-background task based on a defined set of papers. The run took roughly 24 minutes, created several agents, and even reached the token limit before the work was complete. The interesting part came afterwards. The first draft was useful, but the workflow behind it was arguably more valuable: it could be packaged as a reusable skill for similar literature-background tasks. One expensive run could therefore create a process asset for future work. Open and read the article below if you want to see how this example unfolded, and what it suggests about turning complex AI work into a reusable way of working.
-
Dries Faems hat dies geteiltData collection will never be the same again... The chrome extension of ChatGPT has become very good. Seeing ChatGPT collecting data for you live is quite an amazing experience... I sped up the video, so in reality this took like 3 minutes
-
Dries Faems hat dies geteiltWhat is the perfect AI workshop for you? At WHU Executive Education we have developed a portfolio of offerings to help you in taking the next step in your own personal AI trajectory In my personal 3D world website, you can now take a quizz to find out which offering would best fit your needs. Just go to: https://lnkd.in/e7aGFhwt Go to the AI workshops room and take the quizz
-
Dries Faems hat dies geteiltPersonal websites are so 2025...Now you need to build your personal 3D world As I was experimenting with the newest Claude Opus 5.5 model, I tried to transform my personal website into an interactive 3D world that people can visit. It was surprisingly easy...Link to the 3D world in the comments
-
Dries Faems hat dies gepostetAI is forcing academia to confront a persistent weakness: we often treated the amount of writing as evidence of the amount of thinking. For decades, the number of pages someone produced was a rough indicator of the work behind it. A 40-page paper, a 50-page grant proposal, or a 20-page PhD proposal required time, persistence, and sustained engagement with a difficult question. That connection has become much weaker. Generative AI can now produce dozens of pages in minutes. The pages may look coherent and well referenced, but their existence tells us much less about the originality of the thinking, the quality of the reasoning, or the contribution being made. This creates a serious problem for the academic system. Many academic processes still reward visible output because output once provided a workable signal of effort. Today, page count is increasingly easy to manufacture and increasingly difficult to interpret. The academic system will therefore have to change. However this is also an opportunity as it will force us collectively to focus more on quality than quantity. This transition will be uncomfortable. It will require better evaluation, more substantive feedback, and fewer convenient metrics. The question is whether universities will use this moment to reward better work, or simply invent new ways of counting it.
-
Dries Faems hat dies geteiltThis is one of my yearly educational highlights....This week we again organized a massive speed networking event to give the current WHU – Otto Beisheim School of Management bachelor students the opportunity to get feedback from WHU alumni who are experienced startup founders and/or investors 290 students, representing 55 groups, engaged in multiple speed networking rounds with 42 alumni It is great to see that our alumni show the willingness and commitment to give back to WHU in such large numbers!
-
Dries Faems hat dies geteiltMe enjoying Indian summer in Germany, while orchestrating Codex and Claude code to do the hard work... Multi tasking is getting a new meaning for me While enjoying some some final sunshine, ChatGPT Ultra is spinning off a team of agents to help me writing a literature background section for a paper, whereas Claude Code is coding a customized video editing application These are the moments where I feel we are at a tipping point. How we work on a daily basis is changing quite dramatically...
-
Dries Faems hat dies geteiltWhen papers are getting outdated even before they are printed, we know we have a problem... I was reading an In-Press paper from the Entrepreneurship Theory and Practice journal that is part of a forthcoming special issue on AI. The authors (Kleinert and Urbig) do a good job in testing the capabilities of LLMs in predicting early stage venture success. They used models like GPT4, GPT4o and Gemini 2.5 Pro, which they describe as the most advanced models available when they executed their experiments (May-September 2025) The problem: Today these models are already totally outdated as the agentic capabilities of these models was very limited. I am quite sure that the predictive capabilities of current models will be inherently different. How can we still publish relevant research if submission-to-publication timelines can not keep up with technology?
-
Dries Faems gefällt dasDries Faems gefällt das🔥📜 Fresh off the press - new publication in LQ. Very excited to share our new publication in The Leadership Quarterly with Samuel Meyers and Yochanan Bigman titled "Psychological Reactions to Subordinates’ LLM Use at Work." 🧠 This paper finds a highly robust and practically relevant effect: Supervisors show lower trust in subordinates who have used LLMs for a work task. 🔢 The data: It did not matter how much experience the supervisor had with LLMs themselves, how (un)important the task was that the participants were performing, or what the context was (writing academic papers vs. social media posts) - supervisors distrusted employees who used LLMs. The reduced trust, in turn, shaped perceptions of lower performance, OCB, and promotability. And: the more employees used LLMs (a lot vs. a bit), the bigger the penalty was. This finding is important as research shows that most people use LLMs and that these bring substantial performance benefits. Yet we punish employees for doing what everybody does. ⁉️ Is there a solution? We find that the only way to reduce the penalty substantially was explicit organizational policies advocating for LLM use. If rules are non-existent, leaders punished employees similarly as if LLM use was prohibited. 🛠️ The methods: Especially for psychological-behavioral research, the effects are very robust. We studied >2,300 leaders across 6 studies. The effect is large (equivalent to Cohen’s d of 0.86 based on our mini meta-analysis). We found the effects for BEHAVIORAL trust, which means people were willing to put monetary gains/losses on the line (using the very cool rely-or-verify trust game with real money) because they felt LLM users are less worthy of trust. We replicated the results over and over again with various measures (perceptual vs. behavioral trust) and using countless robustness checks. 👁️🗨️ The backstory: Many moons ago, my dear colleague Yochanan Bigman asked me whether I had the capacity to tackle a hard R&R in a top journal. It was a Registered Report project (yeah!), so no data was yet collected and, of course, I said yes. I got to know the unbelievably good PhD student Samuel Meyers who was leading the project. We got another R&R (yeah!) and then... got rejected (oh no). What next? Where should we send an abductive study that is also a registered report? The target was clear; let's aim high and go for The Leadership Quarterly. After multiple challenging reviews before the in-principle acceptance, we got the IPA and now "only" had to collect the data. We did and, after multiple reviews after the IPA, we can finally share this project with the public. I want to thank Andy Loignon and the reviewers for pushing the paper in the right direction and being fair, rigorous, and supportive of our ideas. Also thanks to Nicolas Bastardoz and Jost Sieweke for insights on causality. Link in the first comment WHU – Otto Beisheim School of Management The Hebrew University Business School
-
Dries Faems gefällt dasDries Faems gefällt dasTogether with my colleague Johannes Tafferner, I took the 2nd place at the OpenAI Hackathon at IdeaLab! - WHU Founders' Conference 2026. The challenge, “From Voice to Action – Build with OpenAI’s Live Audio Model,” was a great opportunity to move from thinking and researching about entrepreneurship & innovation to actually building something from scratch. Over the course of the hackathon, we built Walkie, which is our take on how voice AI can change the way we interact with and explore the world around us. I particularly enjoyed the speed of going from an idea to something tangible in such a short amount of time. Thanks to Immanuel Schwall from OpenAI, the IdeaLab! team, and everyone involved in putting the hackathon together. And most importantly, thanks Johannes. Turns out that spending even more time with a colleague from the chair can be quite productive.
-
Dries Faems gefällt dasDries Faems gefällt dasWas lernt man eigentlich so an der WHU – Otto Beisheim School of Management? Ich hab letzte Woche gelernt, wie Müll-Startups an Schulen funktionieren, warum Dänce-Software ins Wohnzimmer gehört und wie man sich gegen KI-Cyberangriffe wehrt. 🤯 Das Ganze nicht im Hörsaal, sondern beim Speed Networking mit Dries Faems, und ehrlich gesagt ging's auch richtig rund. Wilde Ideen, schnelle Runden, und ich durfte meinen Senf als Founder und Investor dazugeben. Ich hab an der #WHU mal promoviert und dachte, ich komm als der mit den Antworten. Rausgegangen bin ich mit mehr Ideen, als ich mitgebracht hab. Es macht einfach ziemlich viel Spaß mit Student:innen zu arbeiten die Bock haben und flott im Köpfchen sind. Und auch das mein Feedback: Uns in Deutschland fehlts nicht an Ideen. Ich glaub uns Älteren fehlt manchmal die Zeit, den Jungen zuzuhören und ihnen die Tür aufzuhalten. So geht's wieder aufwärts. 🚀 Ich bin immer noch geflasht, die WHU-Jugend ist einfach super nice, und ich komm jederzeit gerne wieder. Danke, Dries, für die Einladung 💚 Ehrliche Frage an euch: Welche Gründungsidee aus eurem Umfeld hat euch zuletzt richtig umgehauen? 👇 PS: Dieses Bild zeigt mich wie ich zufälligerweise in Dries Sitz Platz genommen habe, natürlich war ich grad zufällig vor Ort und hat nichts mit KI zu tun. #WHU #Startups #Entrepreneurship #Innovation
Berufserfahrung und Ausbildung
Erhaltene Empfehlungen
2 Personen haben Dries Faems empfohlen
Jetzt anmelden und ansehenWeitere ähnliche Profile
Weitere Beiträge entdecken
-
Tan Yigitcanlar
How far can large language models (#LLMs) go in reimagining the way we plan, design, and govern our #cities of the future? Our new paper, “#LargeLanguageModels in #UrbanPlanning: A Systematic Review and Conceptual Framework,”published in the #JournalOfUrbanTechnology, explores this very question—examining how LLMs are reshaping urban planning theory and practice. Through a comprehensive PRISMA-guided systematic review, the study identifies where, how, and why LLMs are being applied in the field and proposes a #ConceptualFramework to guide their responsible and effective use in planning contexts. Here are the key findings: 1️⃣ LLMs are rapidly emerging across five domains of urban planning—from #policy and #management to urban #design and citizen #engagement. 2️⃣ These models enhance planning capacity through advanced #DataSynthesis, #ScenarioAnalysis, and #ParticipatoryCommunication. 3️⃣ Current applications remain concentrated in data-intensive areas such as #transportation and #management, with limited integration in design and community engagement. 4️⃣ For effective planning use, LLMs must embody four essential qualities: #professionalism, #inclusivity, #trustworthiness, and #convenience. 5️⃣ A new conceptual framework is introduced to guide future research and #ResponsibleIntegration of LLMs in urban planning. Kudos to the lead author Ke Liu for driving this pioneering work, and sincere thanks to my co-authors Rashid Mehmood, Juan Manuel Corchado Rodríguez, and Xinyu Fu for their invaluable contributions and international collaboration across Australia, Saudi Arabia, Spain, and the USA. The paper is open access, and a copy is enclosed with this post for convenience. Read it online here: https://lnkd.in/gbJF9rAm QUT (Queensland University of Technology) QUT Engineering, Architecture and Built Environment QUT Urban AI Hub Texas A&M University University of Salamanca (Universidad de Salamanca) Islamic University of Madinah
4 Kommentare -
Alina Hafner
🎓 New Working Group launched: Doctoral Student Collaboration (Doktorand:innen-Kollaboration) @ Die Wirtschaftsinformatik. Doctoral research in Information Systems is often highly domain-specific and methodologically diverse, which can make it harder than it should be to find the right peers to exchange ideas with, sense-check directions, or simply discuss research challenges on equal footing. That’s why Thimo Schulz, Chiara Krisam, Leonardo Banh, and I are launching the Working Group “Doctoral Student Collaboration” within Die Wirtschaftsinformatik: a platform to connect PhD candidates, strengthen visibility across the community, and build bridges between topics, domains, and institutions. 🔎 Learn more about the working group: https://lnkd.in/d_Nd2dgF 📩 Join our mailing list: https://lnkd.in/dqfgBsv8 What we aim to enable: • Networking across institutions and research domains • Low-threshold opportunities for exchange – so you can quickly find people to discuss your (highly specific) research topics with • A community hub that supports sustained connections and future touchpoints within the doctoral community 📌 Save the date: our first online get-together 🗓️ May 4, 2026 | 17:00–18:30 (CEST) This first session is about building the network: getting to know each other, exchanging perspectives, making initial connections, and laying the foundation for future community meet-ups. If you’re a PhD candidate in the IS community (or work closely with doctoral students): join us, subscribe to the mailing list, and please share this post with others who might be interested. We’re looking forward to connecting with the community and the exchange ahead. #Wirtschaftsinformatik #InformationSystems #PhD #DoctoralResearch #AcademicNetworking
1 Kommentar -
German Consulate General Vancouver
🇩🇪 𝗚𝗲𝗿𝗺𝗮𝗻𝘆 𝗰𝗼𝗻𝘁𝗶𝗻𝘂𝗲𝘀 𝘁𝗼 𝗼𝗽𝗲𝗻 𝗱𝗼𝗼𝗿𝘀 𝗳𝗼𝗿 𝗶𝗻𝘁𝗲𝗿𝗻𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗔𝗜 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵𝗲𝗿𝘀! 🤖 The 𝘗𝘰𝘴𝘵𝘥𝘰𝘤𝘵𝘰𝘳𝘢𝘭 𝘕𝘦𝘵𝘸𝘰𝘳𝘬𝘪𝘯𝘨 𝘛𝘰𝘶𝘳 𝘪𝘯 𝘈𝘳𝘵𝘪𝘧𝘪𝘤𝘪𝘢𝘭 𝘐𝘯𝘵𝘦𝘭𝘭𝘪𝘨𝘦𝘯𝘤𝘦 (Postdoc-NeT-AI) offers outstanding early-career researchers a personalised gateway into Germany’s vibrant AI research landscape. 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗮𝗿𝗲 𝗻𝗼𝘄 𝗼𝗽𝗲𝗻 – a great opportunity for #PhD graduates looking to connect with Germany’s AI community and explore the next steps in their research journey. 👩🎓 #AIResearch #PostdocOpportunities #GermanyInScience #InternationalCollaboration #ResearchNetworking #Innovation Torsten Geissler