Sander Klous

Is it a cat or a dog? Find out how AI is always guessing in my keynote video at https://lnkd.in/geSVtAEi (Notice how the cameraman is frantically trying to follow me on stage 😂 ) This was the opening keynote of June 28 2024 at the OWASP® Foundation global conference in Lisbon. The talk takes you through how great and how terrible AI is through appealing examples: 📌 AI disasters, including forbidden police applications, package hallucination, and hacked missiles 📌 AI triumphs, such as the role of AI in my children's book, opencre.org/chatbot, and AI hallucinating my hair 📌 How OWASP® Foundation is a leader in the AI security conversation with the OWASP AI Exchange at owaspai.org and the OWASP Top 10 For Large Language Model Applications at llmtop10.org. 📌 Our involvement in the AI Act 📌 Software Improvement Group's research on the software quality problems of AI systems 📌 My appeal to my fellow security professionals, that without us, AI will be our doom. Nothing cynical. Just saying. Please use it to your advantage and share it with your peers. if you will #ai #aisecurity

33

1 commentaar

Recently, I was lucky to stumble across https://practical-ai.nl/, an organisation dedicated to public AI education (especially focusing on practical use cases of it). I just finished attending a webinar on AI tools that assist across the various stages of candidate selection and talent acquisition processess. I was positively surprised by not only the clear and comprehensive structure of the event but also the wide variety of presented tools. An aspect that particularly impressed me was that multiple of the selected tools were clearly routed on a rigorous scientific foundation AND are also subjected to continuous testing, focusing on both their effectiveness and verifying their ethical value! Sadly, this is something that cannot yet be said for all AI applications in this field... Thus, I am very happy to see such companies being highlighted at this event. Finally, I truly appreciate the openness and willingness of the speakers to also engage with the critical and technical questions of the audience, partially even in-depth, making it truly feel like an educational event instead of just an elaborate ad-block haha SO, IN SHORT: If you are interested in practical applications of AI, I'd definitely recommend checking out their website, https://practical-ai.nl/, and giving one of their (FREE :D) webinars a try. At least I will most certainly continue to do so!

5

Everyone's talking #AI, but what does it really mean for your IT transformation? Mavim's got the answers. Mavim Webinar Wednesday Series webinar is brewing up something extraordinary. Picture this: Prof. dr. Bas van Gils and Prof. dr.Barry Derksen, from Antwerp Management School, are set to unravel the mysteries of AI's impact on businesses. And guess who's steering this ship of knowledge? None other than Prof. dr. Hans Mulder, our seasoned moderator extraordinaire! 🎤 Our discussion promises to be as entertaining as it is enlightening. Here's what's in store for you: ✨ Strategic insights into AI's business prowess. ✨ Unpacking the essence of trust in AI. ✨ Striking the perfect balance between tech and humanity. ✨ A sneak peek into the future where AI turbocharges decision-making. ✨ Embracing collaboration for innovation. Why should you join? Because in a world where change is the only constant, staying ahead of the curve is the name of the game! Our webinar arms you with the tools and tales to not just survive, but thrive in the digital jungle. 📅 Save the date: November 13, 2024  |  09:30 - 10:00 EST  |  15:30 - 16:00 CET 🚀 Don't miss out on this opportunity to geek out over AI with fellow enthusiasts! Sign up now and let's ride the wave of innovation together: https://lnkd.in/gRjCshYU #AI #Webinar #digitaltransformation #ERP #microsoft #BPM #architecturaltransformation #Mavim

38

2 commentaren

Looking forward to sharing the panel on Exploring AI in Real-world Scenarios with Maarten van den Outenaar and Wouter van der Loop. Thank you Outvie and Mirela Mellone for having me. Are you also there? DM me for catch up and/or meaningful discussions #AI #AIAct #digitaltransformation #futurereadiness #womenindata #womenintech #womeninai

58

3 commentaren

With some help of Egon Willighagen I created my own (straightforward) blog-environment (webpage) on Github. I noticed that the blogposts I wrote during my time at TU Delft | Library and Vrije Universiteit Amsterdam (VU Amsterdam) had been archived and are not visible anymore. So now that I am leaving Npuls and heading to the KB nationale bibliotheek, time to get something “of my own”. And though it is not perfect, and I still have to learn more about Github and Markdown, I am pleased enough to share this. I could find the blogs from TU Delft on the Internet Archive (not from VU Amsterdam), and some of the blogs from Delft were assigned a doi, so are always findable (sidenote, persistent identifiers should rule! Menno de Waal!). The ones from the VU were more difficult to show; I have listed them in one pdf, but had a problem with getting the internal links right. Anyway, here is my modest new blog environment! Up to new stories to share! https://lnkd.in/ez4bfj9y 

42

4 commentaren

It feels like everything changes with acceleration in AI. This is probably not just an emotion but a fact. Therefore, super important to frequently align with the wider AI expert community. This week the Amsterdam ML Engineering community got together at the ABN AMRO headquarters (thanks for hosting the event!). Here's what I learned: - A robust ML platform and MLOps practices are instrumental in facilitating the organization to apply ML at scale in an efficient, collaborative and safe way. - A vector database is super beneficial to get a LLM to give the right answers based on your data. Using Retrieval Augmented Generation (RAG). - How to optimize the LLM-RAG-Vector_Database-Your_Data combination is both experience and art . - Results can be amazing and really make your data (e.g. a municipality website) much better accessible for users 🚀 . - GenAI is a powerful tool and its impact on society can only be predicted now. We need to spend time thinking through its potential negative impacts and ensure we can respond fast enough as a society to mitigate those 👩‍🚀 . - There are so many interesting people developing this domain that any drinks & bites session is too short 😉 #MLengineering #learning #Amsterdam #community #digitaltransformation

31

2 commentaren

Theory of Constraints (TOC) in ICT Service Management by Michel Stijlen The subtitle of this presentation is "Deliver more, faster and more reliable ICT services and projects". Michel shows the buy-in process to implement TOC to ICT Service Management, starting with the objective of an ICT department and then showing the dilema that ICT services face. In this presentation Michel shares the elements (injection, what for and how) of the solution (50% reduction in lead times, properly assign task priorities, aggregation of work) to significantly increase the performance of an ICT department. Free excerpt at https://lnkd.in/gWUvwJ3E #Theoryofconstraints #management

3

Wondering how #DigitalTwins differ from #causality and #SyntheticData - and why it matters? Watch my latest talk to explore these distinctions and the game-changing potential of Digital Twins for bespoke medicine, improve decision-making, and other areas: https://lnkd.in/e8R6AHbh

30

PHILOSOPHY IN THE LIGHT OF AI Hegel or Leibniz Sjoerd Van Tuinen "Reality itself is nothing but the feedback effect of the choices between possible worlds that each of its habitants is constantly making. Irreducible to the past, this means that the common is no longer a question of what is but of what can be. The totality of the system only exists as implied by a concept or field of knowledge. Every specialized discipline, laboratory, artistic practice, political or religious institution, and every new technology must be understood as a partial mirror of this implicate order, as a “transversal optic” (Serres, Le système de Leibniz 3) that, in its difference mode of producing resemblance, allows us to “see deeper” (Leibniz, Philosophical Papers 153), but in such a way that what is anamorphically reflected is not the more universal movement of the Idea but a superdiverse world in endless variation...Such is the specificity of Leibniz’s logic of versions: a model – for example, in mathematics or urban planning or software design – is never an original, let alone an aim in itself, but a further problematization of the common world...Instead, history is itself turned into a database for endless “applications.” Things and concepts never cease to emerge out of chaos in a grinding, nonlinear fashion. It is not the orderly world of a classical metaphysician but the anarchic world of a mannerist diplomat who has no fear of going beyond common sense but is always confident enough to take care of it. " "Philosophy already has a long history of coming to terms with artificial intelligence (AI). But if the future of the concept is indeed inseparable from artificial languages and ubiquitous computing, then philosophy must also be able to understand and rewrite its own history in this unnatural light. To this end, I distinguish two manners in which modern philosophy has pursued the artificial cultivation of intelligence... As technology has become our destiny, this leads them to the problem of the continued relevance of humans to the history of a general self-authorizing intelligence ... Philosophically speaking, this identification reaches back to the proto-structuralist system of Leibniz, which aims for universal communication. Leibniz’s many inventions of formal languages, from the binary system and the universal characteristic to magic and mechanical calculating devices, constitute a proto-AI that functions as the operative code of an inclusive civility....Leibnizian cosmotechnics instead bet on a proto-cybernetic reason that contributes to the distributive composition of an unnatural common sense, all the while protecting multiplicity against its collectivization by a self-naturalizing good sense." https://lnkd.in/gDQXSKby

4

When I started with the AI Annotatielab I always thought wouldn't it be valuable to research what we do? 3,5 years later Mathieu Koot wrote his master thesis: Breaking Down Barriers: Exploring The Motivations and Challenges in Adopting Repsonsible Data Annotation (RDA) Practices for AI Developers. I would love to share some of his findings: The main feature of RDA is the responsibility towards workers in the form of good working conditions, this is seen as a gain creator as it has also been connected to the expectation of higher quality annotated data. Providing a service for annotating data creates the benefit of streamlining the annotation process, helping AI developers save time to focus on core activities. One of the presumed benefits of RDA, the collaboration relationship between client and annotator has been cited as a benefit to both parties. An open communication channel would allow for continuous feedback and understanding of the data for the annotator giving more meaning to the work while also providing the client with more control over the iterative process. This has been noted by the interviewees as adding control over the quality of annotated data allowing iterative improvements. Some conclusions: The first motivator for AI developers to transition to RDA is the emphasis on the ethical treatment of workers. Secondly, the RDA model is believed to deliver high data quality, translating into more reliable AI training. Yet as the value proposition of RDA, as seen in the results, is believed to address both the pains they face and gains they desire, there are several barriers that may deter AI developers from adopting RDA. The premium associated with RDA services can be a significant deterrent, especially for developers with limited budgets or those handling smaller volumes of data. Lastly, while RDA emphasizes data privacy, the perception that outsourcing inherently compromises security can still deter developers. Ensuring data privacy in a way that convinces sceptical developers remains a significant challenge for RDA providers. For the full paper, please go to: https://lnkd.in/geFTiSmb

43

3 commentaren

Interessante lezing van Iris van Rooij over hoe AI kan worden gezien in historisch perspectief sinds ongeveer 1960. Quotes uit de aankondiging: “Yet, as we formally prove herein, creating systems with human(-like or -level) cognition is intrinsically computationally intractable. This means that any factual AI systems created in the short-run are at best decoys.” en “In other words, AI in current practice is deteriorating our theoretical understanding of cognition rather than advancing and enhancing it.” en “The situation could be remediated by releasing the grip of the currently dominant view on AI and by returning to the idea of AI as a theoretical tool for cognitive science. In reclaiming this older idea of AI, however, it is important not to repeat conceptual mistakes of the past (and present) that brought us to where we are today.” Dit laatste is uiteraard het belangrijkste. Het is altijd een zoektocht naar nuttige toepassingen van techniek zonder schade aan mens en planeet. Bij 4vitae wordt alweer een tijdje gekeken waar AI (of beter machine learning) nuttig toepasbaar is om het ontdekken van vervandering/impact in narratieve en warme data te ondersteunen en/of versnellen. Lezing is al geweest. Had er graag bij geweest.

8

1 commentaar

Freek van Ede, a cognitive neuroscientist at Vrije Universiteit Amsterdam, delves into the intricate realm of attention. Focusing on internal attention processes, he explores how the brain filters through vast information to shape future behavior. Read more about his fascinating research here: https://lnkd.in/g289E5gc #neuroscience #performance #brain #visualization #focus

1

How can Digitalization help to implement #CSRD? How can digital technologies helpt to reduce organization’s ecological footprint? What are the key technologies? What is responsible AI? Some of the topics I will cover in my keynote speech at the Vereniging van Compliance Ethics & Integrity professionals on June the 19th. Looking forward to it, #ESG is a shared responsibility and requires intensive cooperation across the different silo’s. Susan du Becker Albert Plugge Michelle Fisser

17

2 commentaren

Jan Leike (previous co-lead of Superalignment at OpenAI) opens up about why he decided to resign. Related… https://lnkd.in/epGDeTZF And… https://lnkd.in/e8fut_CC Why is Superalignment important? It ensures that superintelligent AI systems act in ways that align with human values and intentions. It requires accurately specifying human preferences, designing AI systems that can understand them, and creating mechanisms to ensure the AI systems pursue these objectives. See… https://lnkd.in/en7TqR9b What are the worse case scenarios if Superalignment doesn’t work? See… https://lnkd.in/e6F5P75r

31

13 commentaren

Two great articles in the Dutch newspaper Het Financieele Dagblad on how to innovate with AI. Article 1, https://lnkd.in/e-MQkFYR, discusses how de Volksbank hired a philosopher to work alongside its data and AI experts to spark dialogue about the inherent trade-offs when using AI. “Having a discussion does not mean that you can no longer do anything,” says Krijger. 'Ethics is not like driving with the handbrake on, it is ensuring that your safety system is properly adjusted. This ensures that you are a safer, more reliable and, in the long term, more effective participant in society.'" Article 2, https://lnkd.in/eqq-2bpd, emphasises the commitment required to gather the necessary data for effective AI implementation. "It was a long-term investment, says Daniel Gebler (46). Because before Picnic Technologies could abstract serious, useful information from data, the company was two years further along. AI is not a quick solution, but requires stamina, [...]." Both Daniel and Joris underscore the importance of thoughtful planning and consideration when innovating with AI. They recognise AI's tremendous potential but stress that a quick and dirty fix will not yield the desired results. Experimentation and careful planning are essential. Great insights, Daniel Gebler and Joris Krijger👏! #AIinnocation #longtermplan #dialogue

39

8 commentaren

In a time when polarisation seems to be growing, I recently came across two articles that made me curious about AI’s potential role in shaping our perspectives. 📰 The first (https://lnkd.in/e9Vh9W5X), by Joris van Os in Het Financieele Dagblad, takes a somewhat satirical approach, comparing AI to a modern-day "Oracle of Delphi." With a touch of humour, it recounts how AI was asked about the widening left-right divide. The response, mentioning “political fragmentation, populism, and media influence,” actually made a lot of sense—even if it came from a machine. It started to push the boundaries when it went on to suggest that we "look beyond the shadow of differences and see the light of unity," but the sceptical author had to admit that he wasn't at all dissatisfied with the answers he received. 📰 The second (https://lnkd.in/eeFjv7Gf), from The Economic Times, was more serious and looked at how brief conversations with ChatGPT were found to soften the beliefs of conspiracy theorists. Through fact-based dialogue, AI was able to challenge some of their strongest views—raising the question of whether AI can offer a new way to approach difficult conversations. I was pleasantly surprised by both articles. In a world where it’s easier than ever to gather information and ask questions that might feel awkward face-to-face, I’ve found AI can help us explore tricky topics more openly. Of course, AI doesn’t always get it right, but when the facts are out there, it can help you prepare for deeper conversations much faster. Could AI help us bring a little more balance into our discussions? What are your thoughts?

51

14 commentaren

The next big thing: AI Agents   AI continues to evolve at an incredible pace, the latest trend is Agentic AI. They represent a new approach in how humans and AI interact to achieve complex goals and a new way of building intelligent systems.  At Avans Hogeschool, we want to explore the potential of AI Agents in our education, research, and partnerships. What potential have these agents? How can they reshape the way we work, teach, and innovate? And what do they mean for our current ways of doing things?   I’m looking for an expert speaker who can inspire our teachers, students and partners. We need someone who can explain: What is possible with AI Agents today and tomorrow? Why these developments are transformative? How organizations and individuals can prepare for this shift? If you know someone—or if you are someone—who could guide a diverse audience through this exciting new chapter of AI, please reach out. #AIAgent

21

7 commentaren

On May 22 I will join Nele Roekens Kave N. to discuss algorithmic impact on the rights of persons with disabilities. We will explore: ➡ The potential challenges, risks and opportunities associated with implementing AI in the context of disability rights; ➡ The potential role of NHRIs in addressing the associated risks and challenges. From my side, I will cover the recent Commission and non-commission programs I evaluate, including LLM and SLM for public services and accessibility, ontologies of assistive, health and human capacity emerging technologies, online safety and guardian models (including preventing the spreading disinformation targeting patients, countering hate speech and online attacks), unique patterns related to cognitive, sensory and physical impairments, opportunities and limitations governors face, how NHRI may cooperate to strengthen protection mechanism and feedback loop presented by AI, Digital and Accessibility Acts. 📝 Staff of NHRIs and equality bodies can register here until 21 May: https://lnkd.in/eCCHvWpq 🔍 For further information: https://lnkd.in/e4F-Daam with European Network of National Human Rights Institutions Jerneja Turin #ai #ethics #policy #accessibility #disability

36

2 commentaren