Stian Aagedal

Bitmovin's Split-and-Stitch encoding disrupted the status quo for video-on-demand processing, and enabled huge leaps in turnaround time improvements. We've spent the past decade refining and enhancing our split-and-stitch algorithm to increase efficiency and improve quality with innovations like Smart Chunking. Check out our latest blog to learn more about how Bitmovin's split-and-stitch rises above the competition and can speed up your encoding workflows without sacrificing any quality. https://okt.to/bX2Kn6 #videoencoding #Encoding

16

1 kommentar

Does AI (and AI dev) threaten the open web? Mathias Biilmann Christensen, CEO & Co-Founder of Netlify, makes a strong case why that threat is real, and why it’s important to defend it. It was a pleasure hosting Matt on The AI Native Dev podcast, going deep on many topics, such as: - AI agents have become a new type of Netlify user - how do you adapt to that? - How AI use of a non-AI dev tool (Codemod) boosted their productivity the most - The connecting lines between AI music generation and AI dev New tech paradigms are easier to grasp in closed, opinionated environments, controlled by one entity, than in an open and composable ecosystem. AI is clearly following that path, but where is it leading? Are we ok with all AI dev capabilities coming from our one vendor of choice, or insist on being able to assemble it from best of breed? These are just a few bits of a great episode. Matt’s authenticity and clear passion for AI and the open web meld perfectly with his deep insights, so much that the episode is a bit longer than planned - I just couldn’t stop the conversation! #Netlify #AIWeb #AINativeWeb #AINativeDev #MattBiilmann

61

3 kommentarer

DeepSeek v2 is an insane piece of engineering! Finally got around to skim the tech report https://lnkd.in/dYkeYGxz Quite technical, but if you're up for it, highly recommend reading it. It gives a glimpse into what the immediate future of Machine Learning looks like: An entirely new engineering discipline which is focused deeply on optimizations / squeezing equal quality out of less and less compute. This is a very different discipline than research of novel architectures to reach new capabilities. Here are just a few tricks out of their box: * Using highly granular experts for their mixture of experts and deploying them on different machines. They then load-balance the inter- and intra-device expert selection and communication bandwidth with load-balancing loss functions during training. * Minimizing the memory footprint of KV cache by computing keys and values via a joint low-rank latent vector, rather than a full blown K and V matrix, a trick similar to LoRA. * Taking the liberty of dropping computation of random expert outputs on any device, if the device is getting swamped and the expert has lower affinity score than others on this device. This is a particularly interesting one because it basically cements non-determinism into your inference. The output depends on the global load of the system which is not controlled by you. Amazing to see how research has laid the foundation of something that is now being purified into an engineering discipline.

80

2 kommentarer

The AI landscape is changing so fast that it's important not to write code that ties you to a specific AI provider. Here is a real-world example: We're running an exciting AI project at Gazella AB and yesterday we ran into some issues where Anthropic's 3.5 Sonnet didn't perform well enough. With a single line of code changed, we changed to OpenAI's GPT 4o and it was able to solve the task perfectly. All this because we used Vercel's AI SDK which is AMAZING! The developer experience is awesome and the portability between AI models is as easy as changing a single line of code, which means we'll never build ourselves into a corner with any AI model. Highly recommended. Check it out: https://lnkd.in/dzR-ZvfE

35

5 kommentarer

New Drama in OpenAI 🚀 Ilya Sutskever (co-founder & Chief Scientist) and Jan Leike have quit OpenAI The Two most respectful AI scientists in the world lead the OpenAI Superalignment team, which was set up to ensure that AI systems will be safer. It is not the most confident thing, and then two foremost safety scientists leave the company. I remember when the AI field didn’t have drama; it just had interesting papers and excitement. I hope Ilya creates his new company; I speculate that he will join Elon Musk because Elon convinced Ilya to join OpenAI. #ai #openai #samaltman #llm

146

6 kommentarer

🤖 Creating a fully autonomous AI developer to write and maintain code is still the holy grail of software development. There are many brilliant teams currently working on this, and each of them have their own unique approach. The team at Cosine have made a massive step towards making this a reality. “Cosine achieves state-of-the-art results on the SWE-bench benchmark” They have achieved a score of 43.8% on SWE Bench, which was verified by OpenAI. What's even more incredible is they have managed to do this with a lean team and a relatively limited budget. Though it’s important to note Cosine isn’t looking to replace human developers entirely, but rather augmented them with human-like assistants that they can collaborate with on any kind of coding task. Yang Li + Sam Stenner + Alistair Pullen Read more about their collaboration with OpenAI 👇 https://lnkd.in/gu7xXVPn

21

What a week in AI! And the news just keeps coming, such as the release from EvolutionaryScale, a new AI research lab for biology, has raised $142 million in a seed round led by Nat Friedman, Daniel Gross, and Lux Capital (we'll cover that next week). - Looks like OpenAI's GPT-4 is in for some poetic justice from Anthropic's Claude Sonnet, with Sonnet 3.5 closing in on GPT-4o in LMSYS arena while being significantly cheaper (at $3 per MTok vs GPT-4o’s $5 per million of tokens and GPT-4's $30 per MTok). - xAI partnered with Dell Technologies and NVIDIA to build an AI factory, and Ilya Sutskever launched Safe Superintelligence (SSI) for AI safety. - Roblox is developing a 4D engine for interactive AI experiences that enhance realism and interactivity. - The MCT Self-Refine algorithm significantly improved success rates in solving Olympiad-level math problems (although we see some allegations on plagiarism put forward). - Kentauros AI, a new applied AI R&D lab, released the AgentSea open source agent platform. The toolkit is intended for agent creation and deployment, like a k8s style orchestrator called SurfKit that spins up agents, devices and tools. - A new MoA model outperformed GPT-4 at a fraction of the cost, achieving state-of-the-art results on multiple benchmarks.

71

GenAI alone is insufficient for most users. 🤷🏻‍♀️ But if you combine GenAI capabilities with No Code reliability, that's a different (and much more valuable) story. We wanted to test this hypothesis and whether we could weld AI + No Code together. So, we asked ourselves, can we get an AI engine to generate valuable and fully interactive products from very limited input? (Spoiler alert FE/UX steals the show). We started with just a POC on the frontend & the AI layer, documenting the whole process on medium and reddit, as this meant we could experiment with the channel while gaining significant traction and very valuable raw feedback on the tech stack from the community. As it turns out, we learned that, the infra layer to launch these type of AI powered products, it's much more complex than it may seem. Sometimes even more difficult than together the actual product; meaning pushing to production the AI + FE + BE seamlessly integrated (with CI/CD would be the cherry on top) is tough. This got us thinking, if this is challenging for expert teams to do, what's left for non-technical business owners that want to take advantage of the best technology of our generation? And I mean fully take advantage, not arguing with a chat, but actually using it's brains' power to boost critical business units like CS/User Acquisition or even heavier Dev tasks. Anyway, we launched graph.cubode.com when it wasn't ready, to see users reaction and improve it as we moved along. Content creation began immediately post-launch, and within weeks, we've organically attracted double-digit daily users. It looks like users are eager to leverage AI-generated, reliable, repeatable, and customizable data visualizations. AI generated, reliable, repeatable and customisable data visualisations are something people want to take advantage of. We're developing at an incredible pace, with daily updates improving user experience, frontend, data handling, parsing, AI engine performance, Notion data connections, and various other aspects. I would love for you reading this (specially if you made it this far) to test the platform (be patient please it's work in progress) and share your feedback And for those building in AI, I'd love to hear about your experiences with infrastructure layer challenges. We've now fully automated this aspect and believe it could be a valuable solution to other companies looking to leverage AI technology, specially those limited in technical team capacity. If you've gone through it, I'd love to connect and hear your about your drama. 🍿🗣️

15

2 kommentarer

Here are some thoughts from Anton Osika on AI for programming from an event and presentation he did yesterday!

1

Writer Releases Palmyra-Med and Palmyra-Fin Models: Outperforming Other Comparable Models, like GPT-4, Med-PaLM-2, and Claude 3.5 Sonnet the Writer Team has developed two new domain-specific models: Palmyra-Med and Palmyra-Fin. Palmyra-Med is designed for medical applications, while Palmyra-Fin targets financial tasks. These models are part of Writer’s suite of language models and are engineered to offer exceptional performance in their respective domains. Palmyra-Med-70B is distinguished by its high accuracy in medical benchmarks, achieving an average score of 85.9%. This surpasses competitors such as Med-PaLM-2 and performs particularly well in clinical knowledge, genetics, and biomedical research. Its cost efficiency is truly praiseworthy, priced at $10 per million output tokens, substantially lower than the $60 charged by models like GPT-4. Read our take on these super cool models: https://lnkd.in/gD9Mwt7r Model-Palmyra-Med: https://lnkd.in/gwQvkZWb Model- Palmyra-Fin: https://lnkd.in/gGsmn4V2 Writer May Habib Waseem Alshikh

54

3 kommentarer

AI summaries of customer feedback are boring and useless The problem with AI summaries is… that they summarize. Averaging everything you’re summarizing into a boring grey goo For example, feeding Spotify app store reviews to GPT gives you something like: Spotify app reviews reveal a mix of praise and criticism. Positive Feedback: • Music Discovery: Users appreciate Spotify's music discovery features and recommendation algorithms • User Interface: The interface is praised for its ease of use, making it simple to find songs and create playlists Negative Feedback: • Ads and Free Tier Limitations: Users on the free tier report frequent ads which can be frustrating  • Subscription Issues: Some users experience difficulties managing subscriptions • Content Recommendations: Some users feel the algorithm suggests unwanted genres This is boring/ useless… The purpose of a summary is to hit highlights, not to homogenize everything. A better version might look like: 𝐔𝐬𝐞𝐫𝐬 𝐚𝐫𝐞 𝐡𝐚𝐯𝐢𝐧𝐠 𝐢𝐬𝐬𝐮𝐞𝐬 𝐰𝐢𝐭𝐡 𝐬𝐡𝐮𝐟𝐟𝐥𝐞 𝐦𝐨𝐝𝐞 𝐚𝐧𝐝 𝐬𝐨𝐧𝐠 𝐪𝐮𝐞𝐮𝐞 (𝐫𝐞𝐩𝐨𝐫𝐭𝐞𝐝 𝐛𝐲 𝟒𝟎𝟎 𝐮𝐬𝐞𝐫𝐬 𝟒% 𝐨𝐟 𝐭𝐨𝐭𝐚𝐥) "𝑟𝑒𝑐𝑒𝑛𝑡𝑙𝑦 𝑖𝑛 𝑠ℎ𝑢𝑓𝑓𝑙𝑒 𝑚𝑦 𝑠𝑜𝑛𝑔𝑠 𝑤𝑖𝑙𝑙 𝑓𝑟𝑒𝑒𝑧𝑒 𝑎𝑓𝑡𝑒𝑟 𝑎 𝑓𝑒𝑤 𝑠𝑘𝑖𝑝𝑠"  "𝐼 𝑚𝑎𝑘𝑒 𝑝𝑙𝑎𝑦𝑙𝑖𝑠𝑡𝑠, ℎ𝑖𝑡 𝑡ℎ𝑒 𝑝𝑙𝑎𝑦 𝑏𝑢𝑡𝑡𝑜𝑛, 𝑎𝑛𝑑 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 4 𝑠𝑜𝑛𝑔𝑠 𝑝𝑙𝑎𝑦 𝑜𝑣𝑒𝑟 𝑎𝑛𝑑 𝑜𝑣𝑒𝑟 𝑖𝑓 𝐼 𝑑𝑜𝑛’𝑡 𝑟𝑒𝑠ℎ𝑢𝑓𝑓𝑙𝑒 𝑚𝑦𝑠𝑒𝑙𝑓" "𝑇ℎ𝑒 𝑞𝑢𝑒𝑢𝑒 𝑖𝑠 𝑏𝑟𝑜𝑘𝑒𝑛 𝑓𝑜𝑟 𝑚𝑒. 𝐴𝑑𝑑𝑒𝑑 𝑠𝑜𝑛𝑔𝑠 𝑎𝑟𝑒 𝑟𝑒𝑚𝑜𝑣𝑒𝑑" 𝐈𝐬𝐬𝐮𝐞𝐬 𝐰𝐢𝐭𝐡 𝐜𝐚𝐫 𝐦𝐨𝐝𝐞 𝐦𝐚𝐲 𝐛𝐞 𝐜𝐚𝐮𝐬𝐢𝐧𝐠 𝐬𝐚𝐟𝐞𝐭𝐲 𝐢𝐬𝐬𝐮𝐞𝐬 (𝐫𝐞𝐩𝐨𝐫𝐭𝐞𝐝 𝐛𝐲 𝟖𝟎 𝐮𝐬𝐞𝐫𝐬 𝟎.𝟖% 𝐨𝐟 𝐭𝐨𝐭𝐚𝐥) “𝑁𝑜𝑤 𝑒𝑣𝑒𝑟𝑦 𝑡𝑖𝑚𝑒 𝐼 𝑛𝑒𝑒𝑑 𝑡𝑜 𝑓𝑖𝑥 𝑚𝑦 𝑒𝑞𝑢𝑎𝑙𝑖𝑧𝑒𝑟 𝑐𝑢𝑧 𝐼’𝑚 𝑜𝑛 𝑡ℎ𝑒 ℎ𝑖𝑔ℎ𝑤𝑎𝑦 𝑎𝑛𝑑 𝑤𝑎𝑛𝑛𝑎 𝑡𝑢𝑟𝑛 𝑢𝑝 𝑡ℎ𝑒 𝑣𝑜𝑙𝑢𝑚𝑒 𝐼 ℎ𝑎𝑣𝑒 𝑡𝑜 𝑑𝑖𝑠𝑎𝑏𝑙𝑒 𝑐𝑎𝑟 𝑚𝑜𝑑𝑒 𝑓𝑖𝑟𝑠𝑡? 𝐻𝑜𝑤’𝑠 𝑡ℎ𝑎𝑡 𝑠𝑎𝑓𝑒𝑟" “𝐼 𝑤𝑖𝑠ℎ 𝑡ℎ𝑒 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠 𝑤𝑜𝑢𝑙𝑑 𝑠𝑡𝑎𝑦 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑜𝑟𝑖𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑟𝑒𝑔𝑢𝑙𝑎𝑟 𝑎𝑛𝑑 𝑐𝑎𝑟 𝑚𝑜𝑑𝑒” 𝐓𝐡𝐞𝐫𝐞 𝐢𝐬 𝐚 𝐬𝐩𝐢𝐤𝐞 𝐢𝐧 𝐢𝐬𝐬𝐮𝐞𝐬 𝐰𝐢𝐭𝐡 𝐨𝐟𝐟𝐥𝐢𝐧𝐞 𝐦𝐨𝐝𝐞 (𝟒𝟎𝟎 𝐮𝐬𝐞𝐫𝐬, 𝟒% 𝐨𝐟 𝐭𝐨𝐭𝐚𝐥 𝐮𝐩 𝟏𝟎𝐱 𝐌𝐨𝐌) "𝐷𝑜𝑤𝑛𝑙𝑜𝑎𝑑𝑒𝑑 𝑠𝑜𝑛𝑔𝑠 𝑎𝑟𝑒 𝑢𝑛𝑎𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 𝑤ℎ𝑒𝑛 𝑜𝑓𝑓𝑙𝑖𝑛𝑒, 𝑒𝑣𝑒𝑛 𝑡ℎ𝑜𝑢𝑔ℎ 𝑡ℎ𝑒𝑦 𝑠ℎ𝑜𝑤 𝑎𝑠 𝑑𝑜𝑤𝑛𝑙𝑜𝑎𝑑𝑒𝑑."  "𝐴𝑝𝑝 𝑘𝑒𝑒𝑝𝑠 𝑑𝑒𝑓𝑎𝑢𝑙𝑡𝑖𝑛𝑔 𝑡𝑜 𝑜𝑓𝑓𝑙𝑖𝑛𝑒 𝑚𝑜𝑑𝑒 𝑒𝑣𝑒𝑛 𝑤ℎ𝑒𝑛 𝐼 ℎ𝑎𝑣𝑒 𝑖𝑛𝑡𝑒𝑟𝑛𝑒𝑡 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑖𝑜𝑛." This is a sampling, not a summary of the most interesting specific comments made by users, including specific facts and statistics. Moreover it is actionable and gives a starting point as to where we might dig further. This is not something AI is good at out of the box. It requires the ability to quantify what is being reported by users, decide on the most important points and present an output that is suitable to the use case Others see the same?

34

3 kommentarer

The new ChatGPTo vision features are pretty scary. In our last live stream, Morten Rand-Hendriksen and I discovered some disturbing things about the new Chat GPTo algorithm. ChatGPT effortlessly deciphers binary code, revealing hidden messages and personal identifiers with astonishing accuracy. Could secrets or passwords be exposed with just a glance at an image? It silently stores and collects data, creating an echo of our interactions. What happens when the AI remembers more than we intended? It can inadvertently dox someone's information by remembering what it shouldn't have. ChatGPTo recognizes locations from images and it's able to pinpoint exact places with just a glance, raising unsettling questions about privacy and security. Join us on a short journey through the shadows of AI capabilities, where the line between convenience and creepiness blurs. Here's the scoop 1. The ChatGPT algorithm's ability to read binary code and interpret it to an extent where it can deduce names and information from images. 2. The unsettling feature of the algorithm storing memories by default and recalling personal information without user consent. 3. The unexpected and potentially dangerous implications of the algorithm's predictive capabilities, such as identifying specific geographic locations by just looking at a photo.

69

15 kommentarer

Meta's LLama 3 is the first real open source LLM competing with OpenAI and GPT-4 or outperforming it in certain benchmarks. Does this mean that OpenAI is losing the game? In some areas that might be true, but for any person speaking Finnish or some other Finno-Ugric language, or any other rare language, GPT-4 still outperforms all other closed or open models. The current benchmarks do not measure these real-world use cases: they measure how well the models perform some expert tasks or answer tricky questions. And that's why most users will still stick to OpenAI or other models that are the most useful for everyday pragmatic use.

16

3 kommentarer

Some of our slightly contoversial principles at Sagittal AI: Adapt the AI to humans. We can write code to bring AI to working human practices. It's easier and more scalable - and better - to meet the humans where they are, and bring the AI to them. Team-oriented AI is much better than "single player AI". Context and success in business lives in teams. Empowering teams rather than individuals with AI is the right way to drive value. Getting sh*t done is better than chatting. Chat is a crappy interface. Hook AI into the tools (but not in a "general agent" way - there lies madness). But integration is king. Don't be all things to all people. Be opinionated. The LLM is general-purpose but if you think your value will be brought by wrapping an LLM with "You are a Chief Marketing Officer" you're sorely mistaken. Do the tough work of undertanding the domain. At Sagittal.ai, we put our money where our brains are - that LLMs are powerful but only useful if tied into specific, real business value, and we encode knowledge into usage of the LLM. If you're interested in accelerating your software development teams, send me a message!!

9

2 kommentarer

I discovered this library while trying to create a bridge between Pinescript and MQL5 in EricTraderSDK. Simply put, suppose you have a piece of Pinescript that generates profit on TradingView and you want to convert it into an Expert Advisor (EA) in MT5 to place orders automatically. This process will involve three challenging issues: - The system needs to understand the Pinescript language (the scripting language of TradingView) and parse it into an Abstract Syntax Tree (AST) data structure. - Implement a technical analysis (TA) library similar to how TradingView does in MQL5. - A module to convert AST data into MQL5. You can find a library that solves the first issue by the author Yunseong Hwang here: https://lnkd.in/gNANtxVJ

3

Recently, tech social media blew up with a discussion of whether we need AI control systems (or whatever you want to call them*). Again. The sensationalist takes could be approximately summarized as: “We’d better do something urgently so AGI doesn’t outsmart us!” versus “AI is still pretty dumb, chill out.” I’d like to offer a perpendicular perspective in the linked blog post I wrote for you. (In which we both chill out and build something.) In short: "Nevermind AGI, GenAI, or anything AI. We're not smart enough to design, build, and steer complex technology responsibly and technology is getting more complex quickly. Let's build tools that make us more capable." Or, more bluntly: "Have you noticed that the people in charge are never the human-in-the-loop? There's a loop they need to be in... and they're not there." *For the semantics fan club, I put a glossary of 57 AI-control-related terms for you at the bottom of the blog post. If you're tempted to argue about word meanings, please enjoy those first. 57 of them should surely provide all the nuance you can guzzle in one go. And color-coded too! <3

521

80 kommentarer

how do you get jensen NVIDIA to buy your ai start up. you make a Run:ai for it ;) PayPal. we talk about peter theil, elon musk, the rest of the gang as being exceptional. frankly i don't believe that except for one massive differentiator. perhaps it was theil who was aware that solving payment transactions for sellers, buyers on the wildly success eBay - the auction house at your finger tips. the data was there. 1000s of sellers. millies of products. cash flow. interrupted. so solve the interruption. and do so to the point where eBay must buy you. what's so sophisticated about this is that eBay already had all the data on PayPal to the penny. it was a matter of hanging on. and they did. a bit of luck. a bit of timing. an extraordinary exit. and suddenly the gang had some money in their back pocket jeans. this 700 or 800 milly deal between NVIDIA and Run:ai smells the same. Omri Geller came up with a math solution. to matrix multiplication. NVIDIA gpu resources reconfigured. via software. to optimize the optimization of ai models. in terms of timing this deal also has OpenAI's sama to thank. as he jetset 'round the world speaking on the great worry of safety with ai to governments the politicians began to freak out. coincidently many elections were up for grabs. what? with this ai my competition may take over. oooops. better get a handle on this. this fomo was also evident with NVIDIA's stock rising faster than any other in recorded history of market capitalism. trillys in market cap over 2 years. unique. there were others in the mag7 who also benefited from a rising investment tide. Run:ai just kept on running. and their orchestration became that much more valuable. in fact doubled just because of the stock prices. demand. elon creating xAI. setting himself up for a 1 milly NVIDIA gpu cluster. the industry is aware of existing limitations. Run:ai rather than whining, they're win. ning. sadly i was reminded that places like Vector Institute or Mila - Quebec Artificial Intelligence Institute or Alberta Machine Intelligence Institute (Amii) or other global research centres like ETH Zürich University of Oxford prob had the same level of talent yet definitely didn't have the vision, the investment capital, or attitude to do something about it. this is what we mean by ai being a winner takes all game. this orchestration being solved and now to be open sourced means there is no 2nd place. 3 millys seed TLV Partners 10 millys a SCAPITALVC TLV Partners Sequoia Capital 30 millys TLV Partners Insight Partners 75 millys TLV Partners SCAPITALVC Sequoia Capital ---- 118 millys 700 - 800 millys from jensen NVIDIA. bought to optimize all NVIDIA gpus from the get go ++. source: https://lnkd.in/eHJQ8nWK source: https://lnkd.in/etuCyEfJ source; < . >

15

4 kommentarer

Anthropic just released its new multimodal AI model: Claude 3.5 Sonnet. The benchmarks sure need a refresh as AI models keep improving at breakneck speed. But more importantly, each benchmark represents a real-world application, so it's crucial to steer clear of thin wrappers over any benchmarked capability. #claude #anthropic #chatgpt #gpt #openai

5

1 kommentar

🚨 BREAKING: Jan Leike led OpenAI's superalignment team with Ilya Sutskever, and both resigned last week. Here's what you need to know: ➡ According to OpenAI's post from July 2023: "We need scientific and technical breakthroughs to steer and control AI systems much smarter than us. To solve this problem within four years, we’re starting a new team, co-led by Ilya Sutskever and Jan Leike, and dedicating 20% of the compute we’ve secured to date to this effort. We’re looking for excellent ML researchers and engineers to join us" and also "Superintelligence will be the most impactful technology humanity has ever invented, and could help us solve many of the world’s most important problems. But the vast power of superintelligence could also be very dangerous, and could lead to the disempowerment of humanity or even human extinction." ➡ As Ilya and Jen left, the future of OpenAI's superalignment team is now uncertain. Debates around AGI (when? what? why?) and safety issues persist. ➡ Below are some of Jan Leike's perspectives on OpenAI's approach to safety, alignment, and more (link to the full thread below). #AI #AGI #AIsafety #alignment #OpenAI #superalignment

49

5 kommentarer