Anna Tidstam, PhD, EMBA

Exciting Update in #AI #Legislation! 🇪🇺 🤝 MEPs & Council reach provisional agreement on the AI Act. 🥣 Right of consumers to launch complaints and receive meaningful explanations. 🤖 Rules aim to ensure safe, ethical AI in Europe. 🚫 Bans on social scoring, manipulative AI, and limits on biometric use. 💼 Fines for violations range from 35M euros to 7.5M or 7% to 1.5% of turnover. Check out the official press release of the European Parliament here: https://lnkd.in/eUqMxNBs 👏 A milestone for responsible AI governance, balancing innovation with ethical standards. Let's shape the future of AI in line with our values! My biggest hope is that the transparency requirements will promote further research and open-source development in the field. ☕ Feel free to repost to spread the word, and follow for more on trustworthy AI. #aiact #EuropeanUnion #InnovationWithIntegrity #trustworthyai #ZFambassador ZF Group

28

1 Kommentar

Apple targets Google staff to build artificial intelligence team - SWI swissinfo.ch https://buff.ly/4biSyYK - 🔍 Apple has been hiring AI experts from Google to expand its global AI and Machine Learning team - 🏢 Apple has a secretive European laboratory in Zurich, Switzerland - 📉 Apple's shares have slipped, putting pressure on the company to announce game-changing AI features - 💬 Apple is focusing on deploying generative AI on mobile devices - 🧠 Apple has been developing AI products, such as Siri, based on neural networks - 🤝 Apple's top AI team includes former key figures from Google - 🧑‍💻 Apple has acquired numerous AI start-ups in the past decade - 📱 The next big thing for Apple could be AI smartphones with more powerful chips - 🤐 Apple has been tight-lipped about its AI plans - 💡 Apple's foray into generative AI features may be revealed at its Worldwide Developers Conference in June

2

1 Kommentar

LLMs are part of the AI hype…

2

2 Kommentare

For those who missed it, here a recording of the insightful and fascinating talk between Martin Ebers Brian Williamson Boniface de Champris on #AI #Regulation in #Europe and how the #AIAct can be implemented properly in the #EU and as a model for others. Please do listen in! #CCIAeurope #innovation #digitalEU CCIA Europe

2

A great post from Stephen Gilbert which includes an explainer and link to his most recent paper in Nature Reviews: Cancer. As we are all no doubt aware, current medical device regulation essentialy prevents generalist AI and LLMs from being used in medical devices. Sure, lots of people are developing products for use in the healthcare setting, but they are chosen with great care to ensure they don't fall foul. Meanwhile, countless patients and clinicians are making use of LLMs in an uncontrolled manner. How do we square this circle? The paper, a neat 2 pages, covers this nicely. Perhaps we reign in LLMs to narrow use cases? Maybe we develop a mandatory human in the loop solution and wrestle with automation bias? What IS clear is that blocking GenAI in healthcare isn't a long term strategy.

7

1 Kommentar

#Robotaxis and #autonomoustrucks are the two #autonomousvehicle use cases that receive the most attention. #AI is a key enabling technology for both. To achieve the determinism required in mobility, #AV software platforms incorporated a combination of statistical and symbolic AI. A new crop of AV #startups, led by Wayve and Waabi received large financing rounds for their "end-to-end" #neuralnetwork approaches, similar to those used in #generativeAI. Could these approaches result in vehicles that are cheaper to produce and operate than those utilizing the approaches used to date, adhere to regulations, and be accepted by autonomous vehicle users?

17

1 Kommentar

👉🏼 Artificial intelligence (AI) in diagnostic and therapeutic decision-making-a tool or communication partner? 🤓 J Matthias Wenderlein 👇🏻 https://lnkd.in/gmbMY4PC 🔍 Focus on data insights: - AI as a tool for processing data - Decision intelligence (DI) integrating social sciences and decision theories - Importance of individual patient information in decision-making 💡 Main outcomes and implications: - Improvement in decision-making processes - AI should be viewed as a tool, not a communication partner - ChatGPT's broad training sources enhance its capabilities 📚 Field significance: - Enhancing diagnostic and therapeutic decision-making - Integrating AI with clinical expertise - Advancements in decision intelligence 🗄️: [#ArtificialIntelligence #DecisionIntelligence #DataInsights #ClinicalExpertise]

2

“Four out of five models hallucinate a significant proportion of sources by producing invalid URLs,” even when used with retrieval augmented generation (RAG), “which first performs a web search for relevant sources before producing a summary of its findings.” Even “in the GPT-4 RAG model, we find that up to 30% of statements made are not supported by any sources provided, with nearly half of responses containing at least one unsupported statement. This finding is more exaggerated in the other four models, with as few as 10% of responses fully supported in Gemini Pro, Google's recently released LLM.” “For example, one response by GPT-4 RAG indicated that criteria for gambling addictions are equally applicable across all individuals and groups. But the source it referenced concluded the opposite, finding that "the assumed equal impact of each criterion lacks support in the findings." In another example, the model recommended a starting dose of 360 joules for a monophasic defibrillator (one where the current runs one way to treat a patient with cardiac arrest), but the source only mentioned biphasic defibrillators (where current runs both ways). That failure to distinguish can matter greatly, as there’s been a shift in #technology toward biphasic defibrillators that in fact utilize lower electric currents.” “The ability of LLMs to substantiate answers varies substantially by type of inquiry. Performance is best for MayoClinic and UpToDate and worst for Reddit. Only 30% of the answers to inquiries based on Reddit can be fully substantiated by sources with GPT4 RAG.” “Our findings suggest that LLMs perform worst for exactly the kind of patients that might need this information the most. Where inquiries are mediated by medical professionals, LLMs have an easier time pointing to reliable sources. This has substantial implications for the distributive effects of this #technology on #health knowledge.” GenAI is lightly regulated. “Medical devices that interpret patient information and make predictions in medium-to-high-risk domains are carefully evaluated by the FDA.” On the other hand, “sites like WebMD that strictly report known medical information from credible sources are not regulated by the FDA, and the same holds for GenAI and LLMs where “the critical question is whether such models produce accurate references to substantiate their responses. Such references enable doctors and patients to verify a GenAI assessment and guard against the highly prevalent rate of hallucinations.” As LLMs continue to grow in their capabilities and usage, regulators and doctors should carefully consider how these models are being evaluated, used, and integrated. #innovation #artificialintelligence #startups #hype #AI https://lnkd.in/gRpmC-Ee

207

46 Kommentare

The EU's AI Act is making waves, and for good reason. So, let's take some time to understand how it works. In a deep dive developed by Maarten Stolk, founder of Deeploy on EUVC Here's how the AI Act works. 1/ RIsk-based approach. The AI Act classifies AI systems based on their potential risk. Think of it as a tiered system with stricter regulations for higher-risk applications. 2/ High requirements for high risk. AI that threatens fundamental rights (like social scoring systems) will have to comply with strict requirements. Will include risk assessments, transparency, record keeping, safety & more. 3/ Focus on mitigating harm. The core principle is clear: the higher the risk, the stricter the rules. This ensures the responsible development and deployment of AI. We know the EU AI Act will reshape AI innovation in Europe and beyond. How do you see it impacting the future of AI development? What are your thoughts? Read our full deep dive on EUVC. 🔽 https://buff.ly/3UFeDLF #artificialintelligence #venturecapital #europe

3

👉🏼 Comparative accuracy of ChatGPT-4, Microsoft Copilot and Google Gemini in the Italian entrance test for healthcare sciences degrees: a cross-sectional study 🤓 Giacomo Rossettini 👇🏻 https://lnkd.in/eg_9XUiC 🔍 Focus on data insights: - The study evaluated the accuracy of three AI chatbots (ChatGPT-4, Microsoft Copilot, Google Gemini) in predicting correct answers in the Italian entrance standardized examination test for healthcare science degrees. - Statistical analysis revealed significant differences in performance among the chatbots, with ChatGPT-4 and Microsoft Copilot outperforming Google Gemini. - Narrative coherence analysis indicated that logical reasoning was the predominant correct answer provided by the AI chatbots. 💡 Main outcomes and implications: - AI chatbots demonstrated promising accuracy in predicting correct answers in the standardized examination test. - ChatGPT-4 and Microsoft Copilot showed superior performance compared to Google Gemini. - The narrative coherence of AI chatbots primarily exhibited logical reasoning in their responses. 📚 Field significance: - Adoption of AI chatbots as educational tools in healthcare science programs may enhance learning experiences. - Candidates should cautiously integrate AI chatbots into their study routines to supplement traditional learning methods rather than replace them. 🗄️: #AI #chatbots #healthcare #education #datainsights

2

With #genAI, how will #humans and #computers #collaborate in the #future? And what does this mean for #digital #assessments & #certification? This question was asked by one of our wonderful audience during our recent #webinar on #smart #assessments hosted by Constructor together with bizExaminer, and moderated by the incredible Maria Mashkeeva 🙌 . Let me just say how incredible it is to do those #webinars, as you #learn so much from the #audience, and I keep learning from one of the industry's veterans, Bernhard Waglechner. Deeply grateful for the #opportunity and how we are interacting within the ecosystem. 🙏 With the rise of #genAI, the future will see a harmonious collaboration between #humans and #computers 🤖🧠. This synergy will revolutionize how we approach assessments and certifications. At #bizExaminer, we're exploring how AI can enhance scalability and personalization in certification programs. The key lies in leveraging AI to create smarter, adaptive evaluations that cater to diverse needs, ensuring quality and integrity. 🤓 This shift means assessments will be more dynamic, data-driven, and capable of providing real-time feedback. It’s about moving beyond one-size-fits-all models to create tailored pathways that truly reflect individual capabilities. 💡 Together, we can navigate this exciting journey and transform the landscape of digital certifications! 🚀 #EdTech #AI #Innovation #FutureOfWork #DigitalTransformation #SmartExams #BizExaminer #ConstructorTech

12

#EuropeanAI LSTMs (developed by Sepp Hochreiter) in the form of the #xLSTM architecture are back! And outperform transformers (the architecture behind OpenAI) for certain computer vision tasks! Shout-Out to ELLIS - European Laboratory for Learning and Intelligent Systems (founded by Bernhard Schölkopf et al)! A new model developed by Maximilian Beck and Benedikt Alkin, named "Vision-LSTM" (#ViL) (link in the comments) uses #xLSTM for image processing - instead of transformers. Early tests show it works very well - eg outperforming other models on ImageNet-1K (a benchmarking data set used by AI researchers). And because #xLSTM scales linearly with sequence length instead of quadratically, it could overcome computing limitations (example in the comments)! So #ViL is a strong alternative for eg. medical imaging tasks. This is very interesting, because the #xLSTM architecture was originally designed for understanding language - and not even for computer vision. If you are an AI startup, looking for funding, please reach out to us at DTCF | DeepTech & Climate Fonds

27

1 Kommentar