Nils Berger

The Rise of AGI by Next Year? Sam Altman, CEO of OpenAI ,is making waves once again with bold predictions and ambitious plans in the realm of Artificial General Intelligence (AGI). In a tech landscape that often plays it safe, Altman and his team dared to speak of AGI in times when such notions were deemed "impossibly crazy" by many. During a recent dialogue with Garry Tan, founder of YCombinator, Altman outlined why now might just be the best time to start a tech company. He illuminates the idea that major technological breakthroughs, such as the one we are on the cusp of with AI, are the perfect storm for upstarts to shine. As tech giants cruise comfortably during slow-moving times, these dynamic shifts bring the power back to innovators and startups. The writing on Altman’s wall reads that AGI might not be as far-fetched or distant as previously thought. With continuous, compounding progress in AI technology over the past few years, the dream of achieving superintelligence is potentially "thousands of days away". This optimistic path indicates that, if momentum holds, monumental advancements in intelligence, energy, and even space exploration could be within reach. What's more, Altman emphasizes the transformational power of abundant energy and intelligence. These twin pillars could not only elevate human innovation but also redefine the physical and cognitive work landscapes, making a world of abundance a tangible goal. As Altman paints this picture of boundless potential, the crux remains in conviction—having the tenacity and focus to pursue a single bold idea, no matter the odds. For those looking to dive into the tech startup world, the time is ripe with possibilities, backed by relentless belief in groundbreaking technology like AI. In a world that might not fully grasp the magnitude of these innovations, Altman’s message is clear: Startups have the chance to disrupt and lead where larger companies lag behind. Conviction and the courage to explore uncharted territories will define the leaders of this new era of technological revolution. You will find the whole interview here https://lnkd.in/egSF-iN7 #ai #innovation #agi Future Sales Society ___ ♻️ REPOST and inspire your network to leverage exponential technology. I'm Tim Cortinovis. My posts are about the future of sales and the use of AI in business.  You can expect 4-5 posts a week. Click my name + follow + 🔔 to never miss out on thriving news. Are you looking for an AI deep dive for your organization? I am available to speak about the latest in AI globally - just send me a DM. 

3

2 Kommentare

'Smart' Contracts must be the dumbest name in #web3. Why? Well, the coding architecture that bears this name isn't particularly 'smart'. Smart Contracts are predominantly found in the 'EVM Compatible' Ecosystem. What is it? It is a (centralized) 'community of cryptographic architectures' that revolve around Ethereum. 'EVM' stands for 'Ethereum Virtual Machine'. Some of these architectures are blockchains, some aren't. From those that are not, some are transparent they are not, while the open information some provide about themselves are less clear. Many of the 'EVM Compatibles' have a link to Ethereum itself, but some don't. They use a common coding system, and this assists with interoperability. For example, it is very easy for a wallet provider, like MetaMask or Coinbase, to adjust to accommodate new 'EVM Compatibles'. A wallet designed to accommodate HNS Coins, Handshake Names/Domains, Bitcoin Ordinals, Bitcoin Stamps, Dogenals, and many other types of Web 3 related assets can't. Different types of addresses, identities and codes related to 'EVM Compatibles' have a common indicator in that they begin with the two characters '0x'. Most promo spiel on 'EVM Compatibles' will champion interoperability as if it being 'a good thing' is a given. It is not. Interoperability is a gateway, and if that gateway is open for joint operation between two or more architectures, it is also open to intrusion by bad actors. 9ja Cosmos uses this phenomenon to great effect in the 'Sinosignia' project because the single asian alike character (example: 𨰏 ) integral to the Handshake #blockchain and the Sino Amazon (tokenized artwork) cannot both inhabit the same wallet. We also have the '.det0x' names tokenized to #optimism while the 'Root Zone' is on Handshake. (Aside: if something claims to be a Web 3 Domain, check that it's 'Root Zone' is integrated through a blockchain core rather than secured through a Smart Contract. Some are not secured at all! If a provider is unclear about how it's Root Zone is secured, or doesn't mention it at all, that's not a good sign). 'EVM Compatibles' are not useless. but they are not decentralized. They are dangerous to use on their own when attempting to secure or 'mint' something collectible, and/or of tradeable value. Securing these kinds of assets with Smart Contracts is very risky. It can lose you those assets, and/or contribute to your identity to be stolen. $12.3 BILLION IN LOSSES — THAT’S THE AMOUNT RECORDED DUE TO SMART CONTRACT VULNERABILITIES GRACIOUS JOHN

23

9 Kommentare

I’ve been experimenting with the latest Sora competitor from OpenAI, developed by Luma AI , called Dream Machine. Generative AI for video is so fascinating and has the potential to revolutionize video content production, not just for entertainment but also for creating digital training content. Enterprise training could greatly benefit from using generative AI for video, but a few things are needed: ✅ Higher Resolution and Frame Rate: We need high-quality, realistic video outputs for effective training. ✅ Consistency and Interactivity: For interactive videos, it’s crucial to generate consistent characters that follow a logical storyline across multiple videos. ✅ 360-Degree Video: With VR being integrated into enterprise training, we need generative AI for 360-degree videos. (Blockade Labs, how is your progress on this front?) And if we are dreaming, how about real-time generative AI for interactive videos? Imagine training scenarios where videos are generated on-the-spot based on the learner’s responses. Where do you think generative AI for videos in corporate training is headed? #GenerativeAI #AITraining #DigitalLearning #CorporateTraining #VideoProduction #EdTech #VRTraining #InteractiveVideo #FutureOfWork #LumaLabs #AIInnovation #TrainingTechnology #LearningAndDevelopment #TechInEducation #AIForBusiness #VirtualReality

46

6 Kommentare

Swiss startup FinalSpark aims to build world’s first living processors for AI computing. Their website currently says they offer “instant access to human neurons” where you can “run your remote biocomputing research from your own desk”. This 24/7 remote access to brain organoids is yours from as little as $500 per month. You may be wondering, why not use machines for AI computing? The reason given on FinalSpark’s website is that the human brain is far more efficient than silicon-based chips. They describe using human brain cells as a platform for artificial intelligence as “the next evolutionary leap for AI”. This suggests to me that the human brain is already quite an optimised machine. The question explored by The Queer Book Of Revelation is this: what would happen if engineers discovered that a brain cell in a live host is operating in a more conducive operating environment than a specimen in a laboratory? *** Excerpt from my latest Substack For more, read #TheQueerBookOfRevelation 📸: @son_o_g #SciFi #Technology #QueerLit https://lnkd.in/dRtcEp4q

2

1 Kommentar

🤖 Custom #GPTs: Hype or Hope? When #OpenAI launched Custom GPTs and the GPT Store, there was a wave of excitement. The promise was thrilling - personalized #AI models available on demand, tailored to specific needs and skill sets. For many of us, it felt like the start of something #transformative. 🚀 But now, months later, I can’t help but wonder: where did that excitement go? 🤔 As someone who’s developed multiple Custom #GPTs, I’ve seen both sides. On one hand, these tools can offer real value for specific, targeted tasks. I’ve used them sporadically, deploying them for tailored needs. Yet, despite the initial hype, I see the limitations that have dampened the excitement. 👉 Many of these #models don’t deliver the kind of seamless, "plug-and-play" functionality we were promised. Instead, they require thoughtful design, patience, maintenance, and realistic expectations - things that weren’t highlighted in the initial rush of enthusiasm. So, here’s the question: Were we too quick to assume these models would be the “killer app” for most #business need? In my experience, Custom GPTs often serve as #niche tools - valuable, yes, but far from a silver bullet. The excitement has faded because they don’t always deliver the seamless, intuitive experience many hoped for. 👋 If you’ve experimented with Custom GPTs, what’s your take? Are you using them regularly, are they changing your #workflow - or was the initial excitement just a mirage? Let’s dig into where the real potential lies. - - - - - - - - - - - - 🙋‍♂️ I'm Braňo, and together with transformio and our partners, we're empowering companies, leaders, and teams for the future - starting today. I’m passionate about this work because it’s meaningful and drives real change.

12

7 Kommentare

#Agentic AI graphs are the only way to bring #LLMs into production. For developing prototypes, a basic LLM implementation is sufficient. In 5 out of 10 tests you get amazing results and people are impressed. In my opinion, 99% of marketing posts shine on the surface but lack accuracy for real impact. The implementation of Agentic AI graphs is more complex. Deep domain expertise and business logic need to be translated into an AI-augmented process chart (i.e. AI graph). Here are some of the core differences btw. basic LLM implementations and Agentic AI graphs: 1️⃣ Token Limitation: Basic LLM: Often constrained by token limits, struggling with large or complex datasets without explicit segmentation and guidance Agentic AI: Manages token size more efficiently through the use of a graph state, allowing for better handling of complex data structures and interactions 2️⃣ Workflow Complexity: Basic LLM: Limited capability to conduct accurately multiple tasks (depending on the alignment of the LLM to the system prompt). Not feasible for production Agentic AI: Capable of handling complex, multi-step workflows. Basically replicating the entire business process with validation loops for critical steps 3️⃣ Looping in humans: Basic LLM: Generally operates as a standalone system, with collaboration limited to responding to user inputs. Agentic AI: Collaborates with other AI agents or human experts in a workflow, supporting multi-agent systems for complex problem-solving 4️⃣ Customization: Basic LLM: Customization is primarily through prompt engineering, less adaptable to specific enterprise needs Agentic AI: Can be customized with enterprise-specific logic, resources, and other models. It is adaptable to various business-specific operations and needs 5️⃣ Function Calling Accuracy: Standard LLM: Accuracy can be inconsistent, often relying on predefined instructions without the capability to adapt autonomously Agentic AI: High accuracy in function calling can be reached through validation tools & dynamic prompting P.s., our GoTo framework at nexamind for implementing Agentic AI graphs is #Langgraph by LangChain Any thoughts?

72

🍓 Over the last week, OpenAI released its new #AI model called #o1, previously known by the codename "Strawberry". A model described as able to "reflect" on the problem before trying to solve it, increasing the waiting time before the input prompt and the expected outcome. But also clearly highlighting the decisional processes occurring in between. 🔍 But why release it as a #preview? According to Noam Brown, one of the researchers who worked on developing o1 and its reasoning capabilities, o1 models are not always better than #GPT4o. Many tasks do not require reasoning, and sometimes it is not worth waiting for an answer from o1 compared to a fast answer from GPT-4o. One motivation for releasing the o1-preview is to see which use cases become popular and where the models need work. 🔹o1 learns to refine its chain of thoughts and refine the strategies it uses. It learns to recognize and correct its mistakes, to break complex steps into simpler ones. It learns to try a different approach when the current one doesn’t work. o1 represents the effort to automate #analyticalthinking. 📌 The new model is open to exploration, with the core question once again focused back on us: what do we want to explore? #OpenAI #artificialintelligence #ChatGPT

5

“History never repeats itself, but it does often rhyme.” - Mark Twain From the opening pages of “Competing for the Future” by Gary Hamel and CK Prahalad. This book was published 30 years ago, but the questions remain just as (if not more) prevalent today. From the forthcoming series on The Innovation Show on the work of Gary Hamel. Look around your company: at the high-profile initiatives that have been launched recently. at the issues that are preoccupying senior management. at the criteria and benchmarks by which progress is being measured. at the track record of new business creation. Does senior management have a clear and broadly shared understanding of how the industry may be different ten years in the future?  Are its headlights shining farther out than those of competitors? Is its point of view about the future reflected in the company's short-term priorities?  Is its point of view about the future competitively unique?  How influential is my company in setting the new rules of competition within its industry?  Is it regularly defining new ways of doing business, building new capabilities, and setting new standards of customer satisfaction?  Is it more a rule-maker than a rule-taker within its industry?  Is it more intent on challenging the industry status quo than protecting it? Is senior management fully alert to the dangers posed by new, unconventional rivals? Are potential threats to the current business model widely understood?  Do senior executives possess a keen sense of urgency about the need to reinvent the current business model?  Is the task of regenerating core strategies receiving as much top management attention as the task Is my company pursuing growth and new business development with as much passion as it is pursuing operational efficiency and downsizing?  Do we have as clear a point of view about where the next $10m, $100m, or $1bn of revenue will come from as we do about where the next $10m, $100m, or $1bn of cost savings will come from? What % of improvement efforts focus on creating advantages new to the industry, and what % focuses on merely catching up to our competitors?  Are competitors as eager to benchmark us as we are to benchmark them? Am I more of a maintenance engineer keeping today's business humming along, or an architect imagining tomorrow?  Do I devote more energy to prolonging the past than I do to creating the future?  How often do I lift my gaze out of the rut and consider what's out there on the horizon? What is the balance between hope and anxiety in my company; between confidence in our ability to find and exploit opportunities for growth and new business development and concern about our ability to maintain competitiveness in our traditional businesses; between a sense of opportunity and a sense of vulnerability, both corporate and personal? Image by Christophe Vorlet: https://vorlet.com #Innovation #saturdaythoughts

35

2 Kommentare

Future of advertising? Conference calls? Point of sale? Education? The differences between AI generated lifelike faces and real person are already almost imperceptible. Within the next year or two, we will be incapable of differentiating. 🤔 #AI #Diffusion #Transformers

11

1 Kommentar

#technology #unitedstates My Substack "Aufklärung. The New Enlightenment" is now also a podcast, AI voice cloned from my natural voice using Speech Synthesis by ElevenLabs https://elevenlabs.io/ in Eleven Turbo v2.5 which trained on 40 minutes of my original recordings to create a very high fidelity, professional voice clone cleaning up the clone for external noise or crystal-clear, isolated speech using ElevenLabs Voice Isolator giving me various voice model options to choose from to do an automatic text to speech synthesis using my own voice clone from 5 voice model options of a Substack article I had published earlier in April of this year. I picked the voice model closest to my natural pronunciation, but with some minor variations by the voice AI where it could have done better. It would be nice if I can edit in those variations in intonation and pronunciation without the AI technical bias based on the voice model I used.

1

What is the biggest challenge for AI startups right now? 🤔 Computing? Model efficiency? Scalability? In my opinion, it's gaining the trust of customers and investors in the built solutions and products. Trust is about believing in the AI startup's vision and its technological capabilities. Building this trust means demonstrating transparency, reliability, and a clear path to addressing real-world problems. #ai #artificialintelligence #startups #vc #venturecapital

5

A novel idea… However, is it necessary? Instead of using augmented reality and other technologies, to interrupt people, in order to capture their attention, why not simply put offers out in the world where consumers can choose whether or not they want to engage with your brand🤔 A simple, QR code, thoughtfully placed with thoughtful and concise call to action, would likely accomplish better results in a fraction of the cost. Scan the QR code attached in the comments to see an example from Jib Technologies, Inc. #product #sampling #augmented #reality #ar #zeroparty #data #mobile #shopper #marketing #optinadvertising

9

1 Kommentar

https://lnkd.in/ebVX5xE3 Claude 3? GPT-4?, Gemini?, Llama3? Your own Fine-tuned model? - Struggling to choose the right #LLM for your text summarization needs? Which model is best for you and your use case? Our latest video reveals the secrets to evaluating and comparing models like a pro! You'll discover: - The key metrics you need to assess model performance  - How to objectively compare fine-tuned models to their base versions  - Tips for choosing the best fine-tuning approach for your use case Our latest video dives deep into the key metrics you need to know: ✅ #BLEU: Precision-focused, n-gram based evaluation ✅ #ROUGE: Recall-oriented, measures content coverage ✅ #SemanticSimilarity: Assesses meaning beyond exact word matches Discover how these powerful metrics can help you: 🎯 Objectively compare fine-tuned models to their base versions 🎯 Identify the most effective fine-tuning approaches for your use case 🎯 Make data-driven decisions when investing in AI language models Plus, we'll share insider insights on making smart investments in AI language models for your business. Don't waste time and resources on subpar models. Watch our video now and learn how to select the most powerful tool for your text summarization tasks! https://lnkd.in/ebVX5xE3 #AI #LanguageModels #TextSummarization #MachineLearning #NLP

11

1 Kommentar

As we continue to be impressed by what #NVIDIA has accomplished within this competitive landscape for marketing themselves and their products. They have really raised the bar in terms of technology, strategic alliances, and #customer relationships. Thanks to a strong #RnD focus, #community development and place-based marketing, they have been top players in the market up to the recent past. People, though, if WeAddo were to enter into collaboration with the giant NVIDIA, then this is how we could bring additional delight to their already existing strategies: #Hyper-Personalization through AI: We consider #data #analytics a strong point at NVIDIA, however using our #GenAI #decision #intelligence platform, #SpinOtale, we could help improve the customer experience through hyper-personalized approaches. Democratization of content and product communications and recommendations based on explicit customer-expressed orders and preferences will increase repeat purchasing and grow profits. Innovative Co-Creation Workshops: Instead, we would conduct co-creation sessions in which NVIDIA’s consumers will join in the ideation and development of new products. This kind of customer interaction can provide valuable feedback on new products and their features and help ensure that they meet people’s needs. Expanding Sustainability Initiatives: NVIDIA’s campaigns on internal sustainability would bring further benefits to its stakeholders and consumers by creating campaigns that not only promote environmentally friendly practices but also meet customers’ expectations of social responsibility. Engaging Interactive Experiences: There is also room for the creation of immersive marketing experiences deploying real-time demonstrations of NVIDIA’s technologies. This helps to create more impactful experiences and a clearer message of the product. Localized Marketing Strategies: Localised content is done successfully by NVIDIA, however, we believe it could help push the campaigns even more towards regional markets addressing different geographies and their tastes and customers​. By focusing on these strategic opportunities NVIDIA can strengthen & further enhance its marketing impact.

10

1 Kommentar

Apple Intelligence - AI From Apple Apple Launched there own Gen AI Version called Apple Intelligence. While not the first mover, Apple has taken giant strides with iOS 18, integrating powerful AI capabilities across the OS. Some Highlights : Summarize messages and emails with smart replies suggested based on conversation context - a productivity boost! Text to Siri and expanded voice capabilities, such as understanding stutters/pauses more naturally. Mind-blowing photo editing powered by AI. Siri's new "app awareness" unlocks AI assistance within apps. Image Playground - Generate creative AI images, illustrations, and even animations from textual prompts. Rewrite texts/notes with AI writing assistance to modify tone, brevity and more. Security gets smarter - app locking, face ID protection, and an option to hide apps entirely. Extensive UI customization - from customizable app icons to configurable Control Center pages. While playing catch-up initially, Apple seems poised to deliver one of the most advanced AI experiences on mobile. Which one seems most intriguing? We will go through the features in detail and post about the most interesting ones. Follow me for more updates on tech news and reviews. Let's demystify AI together!

5

1 Kommentar

I have been making web and digital experiences since 1998. I worked in a leading global digital agency before the dot com crash in 1999. I have seen so much happen in digital business in the last 25+ years. I like open innovation. It is an amazing idea. But open innovation is stuck. There are no large companies, there are no real armies of dedicated consultants, there is no community. It is because open innovation is in its core a digital change but did not embrace the rules of the digital world. There are only few local platforms which seem to be designed by and for academics and consultants, not digital natives. Open innovation also did not penetrate small and medium businesses. If it remains in the realm of multinationals then it will not grow. Multinationals can quickly adopt any new trend if it makes them look good, and as quickly throw it away. They do not commit. For open innovation to succeed, for it to become a global standard of doing business, of innovating, it must follow these simple digital rules. Open innovation must copy what already worked. We solved this in our company: 🖐 www.openos.me ✍ www.openinnovation.me #openinnovation #innovation

9

3 Kommentare

AI Chatbots vs. Search Engines: The Ultimate Guide You Need 💡 As the founder of ExpertEase AI and an AI solution specialist, I frequently encounter this pivotal question: "Are AI chatbots just fancy search engines?" Let me clear this up once and for all! 👇 In this comprehensive guide, I break down: ✔️ The fundamental differences between search engines and AI chatbots ✔️ When to use each tool for maximum efficiency ✔️ Real-world applications and limitations ✔️ Future implications for business and technology Key Insight: While Google helps you find information, AI chatbots help you create and transform it. The distinction might seem subtle, but it's revolutionary for business applications. 🎯Pro Tip: Understanding these differences is crucial for: • Strategic decision-making • Resource allocation • Technology implementation • Future-proofing your business 🔍 Curious about more AI insights? Follow me, Deniz Subasi - MBA, for weekly deep dives into AI strategy, technology implementation, and business innovation. Let's navigate the future of technology together! #AIStrategy #BusinessInnovation #TechLeadership #FutureOfTech

5

How long does a company last beyond its #visionaryentrepreneur / #creator & what parameters go into making that post-creator runway lasting? ... is a question that I have often flipped for discussion in many fora. Answers could be useful to #entrepreneurs dedicated to creating a lasting legacy. And also for #investors to analyze their #portfolio. For me the question again took center-stage on the back of the news of #WarrenBuffett cutting his #Apple stake by half. While reasons were. attributed to portfolio balancing, preparing cash pile, rate cut etc. there is a strong buzz about the legendary investors waning confidence in the company. Its been fifteen years since #SteveJobs & there's been no major innovation. #SmartPhone market share in China where it sells over 50 million #iPhones has dropped & the company that marketed cutting edge innovative creations is mostly seen marketing Steve Jobs legacy. So how long & what makes a company last beyond its creator? #Inertia & state of motion is surely one. The size & orbital velocity that the entrepreneur manages. Even the mathematical formula is interesting. Its about the momentum an #entrepreneur creates & the competence of #professionalmanagers to keep up the velocity after him. #Tatas is a case in point closer to home where the group has existed & grown for over a 100. #GenNext & their capability & interest is another factor perhaps. #MukeshAmbani & #Reliance is an example & even #Adani but the group will need to be tracked beyond the #successionplanning that #GautamAdani has announced. #FormulationSecrecy perhaps? Like in the case of #Coke where #product formulation is guarded is guarded since 1886. Or occupying a clear #niche as we see in many #European #manufacturing companies like #AlfaLaval #Schneider #Buhler etc. most created over a century ago. #geographical leadership? like say #Amul that rules Indian dairy since 1946 And cpuld be even #Processleadership like say #Citibank or #Government support like in case of #SBI or #IndiaPost or #CorporateValues like #Tatas or #Geopolitical & #Currency muscle like in the case of many #American enterprises. Post-creator legacy can be a combination of many such parameters. #Promoters & #Entrepreneurs can pick from a long list. More importantly #investors can evaluate their #portfolio companies to assess where they are on the existential curve. Connect if you'd like to chat on the topic Perspective welcome #CorporateStrategy #BusinessStrategy #IndianEconomy #CareerTainment

7

📊 AI's Future: Insights from Leopold Aschenbrenner's "Situational Awareness" Former OpenAI researcher Leopold Aschenbrenner's thought-provoking essay offers a glimpse into AI's potential trajectory. Key takeaways: 📈 Trust the trendlines: Deep learning's consistent progress defies skepticism. 🚀 Rapid advancements: AI continues to surpass expectations across various benchmarks. 🧠 AI researchers by 2027?: Plausible scenario of AI performing complex cognitive tasks. 🤝 From chatbots to coworkers: Anticipate more agent-like AI interactions. 📉 The data dilemma: Potential bottleneck as we exhaust available internet data for training. While speculative, Aschenbrenner's analysis provides valuable insights into high-level AI discussions in Silicon Valley. As we navigate this rapidly evolving landscape, it's crucial to stay informed and adaptable. #ArtificialIntelligence #AIFuture #DeepLearning #TechTrends

3

2 Kommentare