Salim A.

🇨🇭 Apertus, Switzerland's Open-Source AI Revolution, Now Available Globally on AWS SageMaker JumpStart Congratulations to ETH Zurich, EPFL, and CSCS on the release of Apertus - Switzerland's first fully open, multilingual large language model family. Today, we are announcing that Apertus is accessible worldwide through Amazon SageMaker JumpStart, making it easier than ever to harness powerful AI capabilities for your business needs. This cooperation brings Swiss AI innovation directly to your cloud infrastructure: Organizations around the world can deploy Apertus with single-click simplicity directly in their private and secure AWS environments, empowering them to experiment, learn, and scale their AI solutions with confidence. Ready to get started? Explore Apertus on AWS today. 👉 Find all details on our blog post: https://lnkd.in/dqc45nDM #AWSSwitzerland #AI #MachineLearning #SwissAI #Apertus #SageMaker #OpenSource #Innovation Martin Jaggi Mauricio Moncada Imanol Schlag Nicolas Jourdan Malte R. Arlind Nocaj Christian Keller Stefano Mallè

152

3 Kommentare

How can GPT-5 can be the most capable AI model so far yet still feel “modest”? Over the past week, I've been tracking the tensions in how people are reacting. They crystallise into five paradoxes: 1️⃣ The moving-goalposts paradox: As AI clears old benchmarks (like the Turing test), we decide they were never good measures of intelligence. The smarter the system, the less its wins feel like proof. 2️⃣ The reliability paradox: Fewer mistakes make the rare ones harder to predict AND more jarring. Accuracy rises, but trust doesn’t rise in lockstep. 3️⃣ The benevolent-control paradox: The more an AI anticipates our needs, the more it shapes our choices by default. Empowerment blurs into subtle steering. 4️⃣ The floor-ceiling paradox: Benchmarks show the biggest gains at the bleeding edge, but users notice them most in everyday tasks. High ceilings don’t matter if the floor still limits autonomy. 5️⃣ The negative-space paradox: Progress sharpens the outline of what’s missing. The closer we get to AGI, the more the gaps dominate perception. Progress can be real AND still feel strangely underwhelming. Full essay: https://lnkd.in/euzERbyr cc Gary Marcus Yann LeCun François Chollet Luciano Floridi Steve Hsu Fei-Fei Li

67

25 Kommentare

𝗦𝘄𝗶𝘀𝘀 𝗗𝗲𝗲𝗽 𝗧𝗲𝗰𝗵 𝘀𝘁𝗮𝗿𝘁𝘂𝗽𝘀 𝘀𝘂𝗿𝗽𝗮𝘀𝘀 𝗨𝗦𝗗 $𝟭𝟬𝟬 𝗯𝗶𝗹𝗹𝗶𝗼𝗻 𝗶𝗻 𝘃𝗮𝗹𝘂𝗲 → 60% of Swiss VC went to Deep Tech (2019–2025) → #1 globally for Deep Tech VC share → #1 in Europe for VC per capita in Deep Tech → 96% of late-stage rounds led by global investors → ETH Zurich & EPFL among top 4 Deep Tech spinout hubs in Europe   These metrics are a mark of the emergence of Switzerland as the world’s most capital-efficient Deep Tech hub, where frontier science is not only discovered but increasingly commercialized at scale.   The new Swiss Deep Tech Report, produced in collaboration with Dealroom, Startupticker, and VC firms Founderful and Kickfund, maps this transformation in detail.   Switzerland is no longer just a research leader. It’s a startup ecosystem with a repeatable model for turning scientific rigor into investable companies.   ✅ Deep Tech funding has grown 6x in the last decade ✅ Reached $1.9B in 2024 and projected to cross $2.3B in 2025 ✅ 96% of growth-stage deals are now led by global funds—validation from the world’s most selective LPs   Meanwhile, Switzerland continues to top the Global Innovation Index (WIPO) for the 14th year running, and ranks #1 in the European Innovation Scoreboard.   Some of the standout VC rounds in 2024–2025:   𝗔𝗜 / 𝗠𝗟 Neural Concept — $27M Series B (Jun 2024) EthonAI — $16.6M Series A (May 2024) 𝗔𝗲𝗿𝗶𝗮𝗹 & 𝗦𝗽𝗮𝗰𝗲 H55 — CHF20M Series C Extension (Jul 2024) Spaceti — CHF2.1M Seed (Feb 2025) 𝗦𝗲𝗺𝗶𝗰𝗼𝗻𝗱𝘂𝗰𝘁𝗼𝗿 & 𝗤𝘂𝗮𝗻𝘁𝘂𝗺 SynSense — $27.7M Early VC (Aug 2024) Corintis — Undisclosed double-digit $M round 𝗖𝗹𝗶𝗺𝗮𝘁𝗲 & 𝗘𝗻𝗲𝗿𝗴𝘆 neustark — $69M Late VC (Jun 2024) Bcomp Ltd — $40M Series C (Apr 2024) UniSieve AG — Decarbonising heavy-emitting industries via sieving membranes   Switzerland has long excelled at fundamental science. But this decade belongs to the engineers, founders, and investors turning that science into globally defensible companies.   👉🏻 Grab the complete report here https://lnkd.in/dfPkhrz6   #Venturecapital #AI #Deeptech #Startups   Follow us at APEX Ventures and subscribe to our newsletter for exclusive content on groundbreaking Deep Tech startups: 🔗 https://t2m.io/EV2qHQuo

69

4 Kommentare

🚨 Breaking: GPT-5 is finally here! This is OpenAI's most advanced model to date, now powering ChatGPT with frontier-level performance in reasoning, coding, and usability. 🧠 What GPT-5 Brings to the Table • Unified AI Architecture: Combines high reasoning ability with fast, responsive outputs. • Agentic Task Execution: Can handle software generation, calendar navigation, and research report writing autonomously. • Smart Routing: A real-time router determines whether to answer quickly or “think longer” for better output quality. • Creative Strengths: Performs better than previous models in writing, design, and subjective tasks with more natural tone and “better taste.” • Health & Safety Improvements: Hallucinates far less (1.6% in HealthBench), flags health concerns, and better distinguishes between unsafe and safe prompts. 🔍 Benchmark Performance Highlights • SWE-bench Verified (Coding): 74.9% - outperforming Claude Opus 4.1 (74.5%) and Gemini 2.5 Pro (59.6%). • GPQA Diamond (Science QA): 89.4% - higher than Claude Opus 4.1 (80.9%) and Grok 4 Heavy (88.9%). • Humanity’s Last Exam: 42% with tools, just below Grok 4 Heavy (44.4%). • Tau-bench (Agentic Tasks): Mixed results - 63.5% (airline tasks), 81.1% (retail tasks). 🔐 Safer and More Transparent • Significantly fewer hallucinations (4.8% vs. 20.6% for GPT-4o). • Deceptive behavior is lower compared to earlier OpenAI models. • More accurate refusal handling; avoids unsafe tasks while reducing false rejections. 🧰 For Developers and Enterprises Model Variants: • gpt-5: Full-power base model for complex reasoning. • gpt-5-mini: For cost-efficient use cases. • gpt-5-nano: Optimized for ultra-low-latency applications. • API Pricing: $1.25 per million input tokens, $10 per million output tokens. • Control Features: Developers can now set verbosity levels and reasoning depth. 🗣️ Upgrades for ChatGPT Users • Default for Free Users: GPT-5 is now the standard model for everyone on ChatGPT. • New Personalities: Users can choose Cynic, Robot, Listener, or Nerd for different AI styles. • Pro Features: GPT-5 Pro offers higher-quality output via extra computation. Available to Pro users ($200/month) and business plans (Team, Edu, Enterprise). Are you as excited as we are? Let us know in the comments...👇 If you found this helpful, follow us There's An AI For That 🦾

1.120

33 Kommentare

Can you cure cancer with #quantumComputers ? In this edition of the Quantum Pirates newsletter we look into all that. But the answer is in the comments! Also, a lot of summer news from many companies: Qunova’s closed a $10 million Series A round. Haiqu and Xanadu collaborated on the Open Quantum Design—supported by $1.6 million from Canada’s Digital Technology Supercluster—working to deliver the first full-stack, open-source quantum compilation infrastructure to reduce vendor lock-in and broaden technology access. Strangeworks acquired Quantagonia (that was an interesting one! A quantum platform buys al algorithm shop! - no details of the deal disclosed... you know what that means) to integrate quantum infrastructure with optimization and AI tools, accelerating its European reach. Meanwhile ORCA Computing deployed two PT Series photonic quantum systems at Montana State University’s QCORE, marking rapid, practical deployment for applied research under a $26.7 million Air Force-backed program. Key technology advances were also reported: Terra Quantum AG introduced its QMM-Enhanced Error Correction, a quantum gravity-inspired approach aimed at scalable, hardware-efficient error correction; Quantinuum made progress in compact ion trap scaling via microfabrication and semiconductor partnerships with Infineon; and Rigetti Computing launched a 36-qubit system for public cloud access, reflecting incremental momentum in hardware availability. Matrix Product States were shown as superior for privacy-preserving synthetic data, while quantum-AI synergy sped up drug discovery efforts targeting KRas... did they found #quantumAdvantage ? Link in the comments!

43

3 Kommentare

We need to talk about how ChatGPT is shaping the way our brains work. MIT’s Media Lab just ran one of the first neuroscience studies on generative AI — and the results are concerning. The Study • 54 participants, ages 18–39 • Wrote SAT-style essays in three conditions: solo, with Google, or with ChatGPT • Brain activity tracked via EEG across 32 regions • Essays reviewed by educators for quality The Findings 1. Lowest brain engagement. ChatGPT users showed the weakest neural activity — consistently underperforming at neural, linguistic, and behavioral levels. 2. Memory loss in action. When asked to re-write essays without AI, most ChatGPT users couldn’t remember their own work. Brain waves tied to memory and creativity (alpha/theta) stayed low. 3. Declining effort. Over months, AI users got progressively lazier — many defaulted to “copy-paste” prompts and editing instead of writing. 4. Writing quality suffered. Teachers called their essays “soulless,” “formulaic,” and “lacking depth.” By contrast, the Google and solo groups wrote with more originality and reported higher satisfaction. 5. Critical thinking risk. Researchers warn heavy reliance on LLMs could weaken neural pathways that build resilience, memory, and creativity — especially for developing brains. Why It Matters Lead researcher Nataliya Kosmyna released the findings early, before peer review, because she feared policymakers might rush to integrate AI into education (“GPT kindergarten”) without understanding the risks. She argues: developing brains are most at risk. Other studies from the Media Lab have also linked heavy AI use to increased loneliness and reduced motivation. Early signals suggest the same patterns may hold in fields like software engineering — where critical thinking and problem-solving are essential. The Takeaway AI can boost speed and convenience. But the trade-off may be memory, depth, and engagement. The best strategy isn’t avoiding AI — it’s using it intentionally: • Start with your own thinking • Bring AI in later to polish or expand • Use AI as a tool, not a substitute for thought This is less about banning ChatGPT — and more about deciding what kind of thinkers we want to become. Read here: https://lnkd.in/erug8TSc #AI #Leadership #CriticalThinking #Neuroscience

99

6 Kommentare

🌍 BREAKING: The UN just made history on AI governance. For the first time, the United Nations General Assembly has adopted a resolution creating two global mechanisms to guide how humanity navigates artificial intelligence: 🔹 Independent International Scientific Panel on AI – 40 experts worldwide, tasked with producing evidence-based assessments on AI’s risks, opportunities, and impacts. 🔹 Global Dialogue on AI Governance – an annual forum for governments, industry, civil society, and academia to build shared approaches and bridge divides. 💡 Why it matters: AI is advancing faster than policy. Without coordination, we risk fragmented rules, growing inequalities, and weakened trust. These mechanisms aim to: • Promote safe, secure, and trustworthy AI • Support capacity-building in developing countries • Ensure transparency, accountability, and human oversight • Advance open data, open models, and interoperability ⚖️ For now, this resolution is non-binding — but it lays the foundation for international cooperation. Much like other global compacts, its true power lies in shaping norms, aligning priorities, and building trust across borders. This is a historic step: AI governance is moving from national debates to the global stage. 👉 What do you think — will global cooperation help ensure AI serves humanity, not the other way around? Link to the full resolution in comments 👇 #AIforGood #AIGovernance #GlobalGoals #UN

162

58 Kommentare

Did you know that AI models don't just hallucinate? They also produce what researchers call “Machine Bullshit” — convincing but untrue output. This includes: - Empty Rhetoric - Weasel Words - Paltering - Unverified Claims - Sycophancy In public health, this can mean fabricated research, wrong diagnoses, and biased algorithms. Once embedded in decision-making, these errors are far harder to detect than hallucinatory falsehoods. As public health leaders and practitioners, we need to implement technical safeguards, but the bigger fix is human vigilance and review. Accuracy in public health is non-negotiable. Read the full article here: https://lnkd.in/ekWUatjG

60

1 Kommentar