Terry xR. 𝘚𝘤𝘩𝘶𝘴𝘴𝘭𝘦𝘳

In this visionary conversation, Google DeepMind cofounder and CEO Demis Hassabis delves into the history and incredible capabilities of Al with head of TED Chris Anderson. Hassabis explains how Al models like AlphaFold - which accurately predicted the shapes of all 200 million proteins known to science in under a year - have already accelerated scientific discovery in ways that will benefit humanity. Next up? Hassabis says Al has the potential to unlock the greatest mysteries surrounding our minds, bodies and the universe. https://lnkd.in/erYwtSE4 #AI #TED #DeepMind

4

1 Comment

Save your spot! Today is the LAST DAY for EARLY BIRD Tickets to our #SeattleAIWeek event "Designing AI Enabled Products and Services" at SURF Incubator. We'll have networking over wine & cheese and I'll be presenting. If you're and AI designer, builder, founder, investor, or otherwise enthusiast or simply curious, join me and the folks at SURF. Here are some questions to ponder: 💡 What will the User Experience of AI applications look like? 💡 How does it look like now and where is it going? 💡 How do we make sure we're carrying proven methods and principles we've learned in deterministic systems forward into probabilistic interfaces and build on top of them rather than taking a step back by ignoring them? AI is here to stay and the companies who succeed past the hype will be the ones that understand their customers and solve real-world problems. https://lu.ma/0weyg4sr #Startups #UXforAI #AI #UX #Seattle

2

Ok who’s heard of Data Poisoning? I was absolutely fascinated to learn about this and find out how artists are fighting back against Generative AI by using tools like Nightshade and Glaze to upload “poisoned” images of their work online, as they couldn’t stop the models from scraping their work. But now when the models are asked to replicate the artists work, they produce corrupted images from the poisoned data. Fight on! #generativeai #ai   From MIT Technology Review – “Poisoned data samples can manipulate models into learning, for example, that images of hats are cakes, and images of handbags are toasters. The poisoned data is very difficult to remove, as it requires tech companies to painstakingly find and delete each corrupted sample.” https://lnkd.in/gQrjCv5j https://lnkd.in/gQNJ-3pA

18

1 Comment

Today is a massive milestone for Coactive AI. We have closed our Series B and another $30M in funding! Our goal is to unlock the untapped potential of visual content for applications ranging from intelligent search to video analytics – no metadata or tags required. Despite images and videos capturing more and more of everyday life, enterprises have struggled to do much more than store it, making it more of a tax (or liability) than an asset. They have been stuck in a world of tag, load, search (TLS), where they need to tag raw assets with either human or machine annotations, load those annotations into their systems as metadata, and search based on those annotations. That process is slow, expensive, and inflexible and does not scale to the image and video content we have today. And with the rise of Generative AI, it won't be able to keep up anymore. Coactive is here to fix that by flipping the process on its head with a load, search, tag (LST) approach. With our platform, enterprises can load and index the raw images and videos directly and make them searchable – no metadata or tags required. We do all the heavy lifting in a scalable and secure way, leveraging the latest in multimodal AI to understand the pixels and audio directly and give enterprises a new set of superpowers. Find clips in less than a second, not hours. Develop a sixth sense to detect content that violates standards & practices or editorial guidelines. Cover breaking news before AI can generate it. And turn raw video files from a tax to an asset that generates new revenue streams. With Series B, we are kicking off the next chapter of the big data movement by unlocking the value of visual data like Splunk did for log data or Snowflake did for semi-structured data. It is an ambitious vision, but over the past year, we have already seen some of the largest tech-enabled enterprises from Fortune 500 retailers, media & entertainment companies, and community platforms use Coactive to create valuable applications from their massive volumes of unstructured visual content. Applications that we never imagined when we first started, like analyzing movies down to the minute to determine which scenes capture audience attention or accelerating marketing campaigns with better search and reuse of existing assets. Now, we are expanding our platform to tackle even more advanced use cases across our core applications in intelligent search and video analytics. Thank you so much, Cherryrock Capital and Emerson Collective, for co-leading the round, Greycroft for your significant participation, and Bessemer Venture Partners, Exceptional Capital, and Andreessen Horowitz for doubling (or tripling) down on us. I couldn't imagine a better dream team to tackle this challenge with! Read more in Forbes: https://lnkd.in/d7fsY48u

621

96 Comments

When you bring effort to every engagement you are telling your culture accountability matters. When you take the time to work on a problem, idea or solution with someone, you are telling them they matter. It’s not a surprise sustained high performing cultures have effort, accountability and teamwork as cornerstones of their existence. #pathtoyes #growth #growthmindset #leadership #culture https://lnkd.in/eqtTvhcK

15

How NVIDIA And Insilico Medicine Are Accelerating Drug Discovery With Generative AI #GenerativeAI is the technology behind #ChatGPT and other conversational, content-generating tools. And while the technology can be used in all sorts of fun and interesting ways, it also has a very serious role to play in the world of #drug discovery. https://lnkd.in/ejjXZ-kp

48

5 Comments

The Stacks. TLDR. (<2 min read) Sandboxes and bad questions We opened this month's Stacks by sharing our new AI integration, why Sandboxing is the right choice for our database and why this is the exact feature that you didn't ask for, but needed. Our second AI editorial shares the importance of asking good questions when doing research and how better search strategies won't completely save us. Fibro: Make a Decision Already Our fibromyalgia content gives you the confidence push you need to stop referring out these patients and making the freaking diagnosis yourself. Standards of care guidelines call for primary practitioners to just call it already. If you use the current guidelines, you're not likely to flag the wrong person. We promise. ED in men (and not the one you're thinking) Eating disorders are maybe more common in your patients with ovaries, but the consequences of missing it in the other 50% of your patients is just as serious. We share how to flag your patients with eating disorders regardless of gender, and why you need to be more careful looking at every patient that sits across from you. Thumbs down for B3 on cholesterol In our sneak peek content (typically for members' eyes only), we share the importance of clinically relevant outcomes for cholesterol care and how niacin doesn't fit the bill for cholesterol lowering. Hashimoto's and Ovarian Reserve A few important papers were published in the last 18 months creating even more of a concrete connection between HT and POI. If you're working with people TCC who have a positive TPO, you may want to read this ASAP. Want to read the whole thing? Click here: https://joom.ag/hStd

1

Silvio Sangineton shared this research study as a reply on another post. I am reposting it as it's super interesting. This is a punch line for me. It can help people be more creative. It improves weaker writers' ability to produce better content. It helps with ideas, but near the end, this paragraph is telling. While these results point to an increase in individual creativity, there is risk of losing collective novelty. In general equilibrium, an interesting question is whether the stories enhanced and inspired by AI will be able to create sufficient variation in the outputs they lead to. Specifically, if the publishing (and self-publishing) industry were to embrace more generative AI-inspired stories, our findings suggest that the produced stories would become less unique in aggregate and more similar to each other. This downward spiral shows parallels to an emerging social dilemma (42): If individual writers find out that their generative AI-inspired writing is evaluated as more creative, they have an incentive to use generative AI more in the future, but by doing so, the collective novelty of stories may be reduced further. In short, our results suggest that despite the enhancement effect that generative AI had on individual creativity, there may be a cautionary note if generative AI were adopted more widely for creative tasks. https://lnkd.in/ghMWAWfN

1

A very interesting and innovative decision by the US Copyright Office (USCO) relating to a novel that was written with extensive help from Generative AI. The USCO determined that the human author is considered the author of the “selection, coordination, and arrangement of text generated by artificial intelligence.” This means no one can copy the book without permission, but the actual sentences and paragraphs themselves are not copyrighted and could theoretically be rearranged and republished as a different book. This is a very interesting middle ground relating to human/ai generated content. Looking forward to further developments on this front... https://lnkd.in/dP3Vd-bT #generativeai #genai #intellectualproperty #copyrights #aicopyrights #ip

31

1 Comment

Brian Eno uses Generative Art to trump Generative AI. This is fitting. Brian Eno is one of the foremost and earliest musical pioneers of electronica. Brian Eno has been doing procedural 'generative' creations since the late 1970s. And one of his most famous albums, On Land, was entirely procedurally constructed. The difference is that there is human intention. This is Generative Art. ---- NYTimes: "‘Eno’ Review: Creativity, 52 Billion Billion Ways A new documentary about the groundbreaking artist Brian Eno breaks its own ground, too." NYTimes: https://lnkd.in/gAJkjiPC ----- The Verge: "The making of Eno, the first generative feature film Director Gary Hustwit tells us about his living documentary and the technology that changes it for every screening." (Kevin Nguyen) "Brian Eno has taken many musical forms: producer, technologist, glam-rock star. Eno, the new documentary about the musician, also takes many forms, though more literally. Each showing of the movie, which opens today in New York at Film Forum, will be a different version. It is, according to its makers, “the first generative feature film,” meaning pieces of it will change shape and structure per viewing, thanks to some clever software ingenuity designed by director Gary Hustwit and his partner Brendan Dawes. "While Eno may be more famous as a band member of Roxy Music or producing David Bowie’s Berlin trio, the form of this documentary fits its subject: Eno himself has been making generative art for decades now, since avant-grade minimalism of 1975’s Discreet Music to the mutating soundtrack of Spore. This is different from generative AI, though, which uses massive training models to infer what it should spit out. Eno is crafted through 30 hours of interviews and 500 hours of film — a curated and ethically sourced data set — with certain pieces weighted to be more likely to appear. Basically, it follows a set of rules and logic written by Hustwit and Dawes. According to The New York Times, there are 52 quintillion possible versions of Eno. Eno is an unexpected documentary in other ways, too. You might expect a movie with so many possibilities to be broad, yet the scope remains narrow. Rather than taking a sweeping look at the musician’s long career, it expounds on his philosophies about creativity. The film deploys some great archival footage, but there are no other talking heads. It’s an approach you might expect from filmmaker Hustwit, best known for Helvetica. But the ambitions of Eno are greater than the film itself. Hustwit wants to see a future exploring generative filmmaking using Brain One, the name of the software behind Eno . The director spoke to The Verge about his ever-changing documentary, the bespoke patent-pending technology that fueled its creation, and how the movie inadvertently met this AI moment and did something more ambitious by thinking smaller about generative art." (keep reading) The Verge: https://lnkd.in/g9zkfCvE #brianeno #generativeart

28

20 Comments

Meta has created a way to watermark AI-generated speech The news: Meta has created a system that can embed hidden signals, known as watermarks, in AI-generated audio clips, which could help in detecting AI-generated content online. Why it matters: The tool, called AudioSeal, is the first that can pinpoint which bits of audio in, for example, a full hour-long podcast might have been generated by AI. It could help to tackle the growing problem of misinformation and scams using voice cloning tools https://lnkd.in/gzFF3Dcv https://lnkd.in/gw9c7FXc

3

The predictable enshitification of maps with ads: https://lnkd.in/g8ptc4_s But hey, what if it was all more contextually relevant? Like what if you really needed gas and it showed you gas stations nearby? Isn't that useful? That's still a very likely "no" -- esp. if the ad-tech company is using your personal information against your personal best interests. E.g., even if you do need gas, the system can hurt you if the ad auction winner is more expensive, got worse reviews, and was farther out of your way than the other options you weren't told about. FWIW, at Keyhole (which became Google Earth after acquisition) we also played with how to monetize maps via ads. Serving earth data is very expensive, esp. for a small company without cloud resources. The best idea we had was to promote map labels to show up earlier or more prominently when zooming or panning in, based on a prior auction. Turns out, you never want to show all possible map labels at once -- TMI. Highlighting a key city name often leverages negative space (no labels) around it. So the ad auction would be one way to pick and prioritize which comes first to make everything more readable. Contextual picking is even better. We also played with less conspicuous visual ads. But none of these were very satisfying. All felt kind of icky, so we never did it. Google won our attention in the 2000s not simply because search worked better than the other options, but because the text ads were small and easy to ignore. Remember that? Ad-tech business models always drive to the worst outcome over time.

8

And voice synthesis just got amazingly better I wrote about voice cloning just one short week ago - about how OpenAI’s VoiceEngine could synthesize voice from a 15 second audio clip. Pretty amazing stuff. Welp, here we go again - Product Hunt yesterday featured Replica Studio’s new Voice Director which let’s you create unique voices in a Voice Lab, and export text to speech using the voices that you created. The demo is here: https://lnkd.in/gXna_g5v and it is fun to watch. I tried Voice Director and it’s pretty cool. Web-based (though they also have a downloadable app) Voice Lab is their lab for mixing and matching voices to synthesize new voices. The source voices come from professional voice actors. An intuitive web-based interface let’s you mix the voices together to synthesize new voices. You can save voices and then use them with text strings to generate voice. So simple; so cool. Also interesting is that Replica did a deal with SAG-AFTRA so that voice performers can work with Replica to create replications of their voices and get paid. From Replica’s release: “We license our training data from a growing network of voice actors, ensuring best practices around consent and compensation. We are honored to be the first AI company to partner with SAG-AFTRA, and pioneer AI licensing models that create new revenue streams for actors and talent.” The applications are many, but the gaming industry is absolutely going to go through massive change from voice engines like Replica and image/scene generators like Scenario.com, both of which have APIs. Now you can dynamically generate scenes, characters and voices and directly integrate into a game. Oh the times, they are a changing. This one is pretty cool.

8

4 Comments

🚀 Rise of AI: Confronting the fear within 🚀 As AI like GPT-4o transforms our world, fear and excitement collide. While change can be daunting, it’s also a chance to redefine our future. Let’s harness courage, embrace innovation, and unlock our true potential. The future is bright! 🌟✨ #AIRevolution #CourageOverFear #Innovate #FutureReady #AITransformation #NewBeginnings https://wix.to/yaQWTlM

2

Nvidia’s Rapid Ascent to a $2.7T Market Cap: What’s Their Moat? For a company founded in '93, Nvidia's rise to a $2.7T market cap has been extraordinary. So, what really is Nvidia's moat? Let’s break it down. PART 1 — SOFTWARE The story begins in the early 2000s. Jensen Huang, Nvidia's CEO, and his team were meeting researchers who were using their products in innovative ways. Most researchers were hacking graphics packages to run complex parallel compute tasks. It was far from ideal. Enter Ian Buck, a visionary who saw the potential of running general-purpose programming languages on GPUs. Nvidia funded his Ph.D., and after graduation, Ian joined Nvidia to commercialize this tech. In 2006, Nvidia released CUDA: - Compute - Unified - Device - Architecture CUDA made all those parallelization hacks accessible to everyone, becoming the default choice for researchers. It allowed for easy customization of the low-level hardware, which developers loved. Today, startups like MosaicML evaluate available technologies and inevitably choose CUDA. The ecosystem around CUDA has grown so robust that its lead is virtually unbeatable. This software layer is at the core of Nvidia's moat. PART 2 — HARDWARE The other side of Nvidia's moat is hardware, specifically AI supercomputers—not just graphics cards for crypto and gaming. The story of these supercomputers began in the late 2000s. As Nvidia developed CUDA, Jensen asked his team to build a supercomputer to help them create better chips. The result was a massive supercomputer, weighing 100 pounds, that strung together many GPUs with world-class networking for ultra-fast computing. In the early 2010s, Jensen gave a talk about this AI supercomputer. Elon Musk got wind of it and said, "I want one." In 2016, Jensen donated one to Elon Musk's then little-known nonprofit, OpenAI. He even hand-delivered it—there's photographic proof! OpenAI quickly realized the supercomputer was exceptional for training large neural networks. The 2016 Pascal architecture delivered an impressive 19 TFLOPS of FP16 operations—19 trillion floating point operations per second. But that was just the beginning. Since then, Jensen and the Nvidia team have been leading the industry, exponentially increasing TFLOPS. The latest Blackwell architecture delivers a massive 5000 TFLOPS—over 260 times the AI computing power in just 8 years. And it sells for more than $75K. Yet, buyers like Meta, OpenAI, Google, and Amazon can’t get enough, as their internal ASICs are nowhere near Nvidia's level. As a result, Nvidia's profits and market cap continue to soar, cementing its position as a leader in AI hardware and software.

1

Concerns around image licensing, AI training datasets, and data providence are a household issue for me. We have an incredibly hard-working artist in our family, and they are understandably concerned as to how their work is used, without attribution or compensation, as generative AI continues to move forward unregulated. While my wife and I use AI daily for all kinds of tasks, we still debate the merits (and dubious intellectual property protections) of generative AI around the dinner table. This is especially true of artistic outputs like images, videos, and the like. How can we share our work? How do we get feedback? Most platforms' licensing has been tilted in their favor for years – is all of my work being used to train AI? How can I find out, or even stop it? Is there a place I can share my original work, with like-minded humans, without it being commoditized without my explicit consent? To answer that last question, we're looking at Glaze and Cara https://lnkd.in/dzEgR4cG . I think this is the beginning of a cat-and-mouse game that will take years to play out, but I applaud the work being done to protect artists until we have a reasonable, ubiquitous, proven legal framework. If you publish your work online, I encourage you to look at Cara and Glaze. Cara is a great, volunteer-led project. They have roles available if you have some bandwidth for social Good.

3

2 Comments

Here’s a little article i wrote about superhumans and the future of AI / Human symbiosis. Also, how XR could enable this to a scientific degree. What else should we do before global warming threatens our extinition. #AI, #XR, #Spatial, #Humanity, #thought, #sociology, #psychology, #superhumans https://lnkd.in/g3bVYCgp

2

RAG is all the rage This TechCrunch article does a nice job describing what retrieval augmented generation (RAG) is, its benefits and shortcomings https://lnkd.in/et4WFjvq #ai #artificialIntelligence #hallucination #RAG

11

As the #AI cold war heats up, all sides are under pressure to perform. Artificial intelligence startups are struggling to generate revenue, but platforms leveraging their models are also weighing the risks and rewards through a tighter lens. That’s what makes Webflow’s purchase of Intellimize an intelligent move. While adding #generativeAI for #content to a workflow has reached peak parity, Webflow is looking for ways to expand its value to a growing number of larger teams by offering AI-driven optimization. A leading #website builder with more than 3.5 million users, Webflow will integrate Intellimize’s #personalization and #optimization capabilities into its ecosystem – which already features robust tools and an onboard CMS – to deliver what it calls the world’s first “Website Experience Platform.” As founder and CEO Vlad Magdalin observed in his post about the news, Webflow is endeavoring to solve the problem of deploying cumbersome optimization tools that require significant engineering support. “Our goal is to give marketing teams and web agencies the ability not only to build incredible web experiences, but also seamlessly personalize those sites for different audiences, easily create experiments to test which versions perform better, and to optimize their buyer journey throughout their website to drive conversions and revenue.” We unpack what this means to users – and the market – on CMS Critic.

45

2 Comments