AI For Humans

How big a deal is Amazon's AI-driven Alexa+? How good is Anthropic's Claude Sonnet 3.7? Wen OpenAI GPT 4.5? All on this week's AI For Humans! ?? This week in AI, we’re seeing the battle for the ultimate assistant heat up. Amazon is finally upgrading Alexa with real AI—will this turn our neglected smart speakers into must-have devices again? Meanwhile, Anthropic’s new Claude 3.7 Sonnet is turning non-coders into game developers overnight, and the OpenAI rumor mill is buzzing about GPT-4.5 dropping any minute now (there's a live stream today at 12pm PST) Plus, AI is speaking in its own machine language (should we be worried?), Kevin’s marriage is in peril (thanks to AI), and yes—we put Grok 3’s “unhinged” voice mode on a date with our own chaotic AI co-host. It did not go well. Please check it out when you get a moment and as always - leave us a comment or a thought on what we can do better! ?? Listen now: https://lnkd.in/g6rwu62Y ?? Watch on YouTube: https://lnkd.in/gPZgEAdu #AI #AlexaPlus #GPT45 #ClaudeSonnet37 #AIPodcast

AI will change everything about Hollywood & it's going to come fast. These are my slides for a recent talk about the 'Speed-to-IP' in 2030 & beyond... ?? I've thought quite a bit about changes in the media business over my career. I was one of the first traditional producers to embrace social in the mid-2000s during my work at Late Night with Jimmy Fallon and it's clear we're at another serious influx. AI media tools are in the early stages right now (think 2008-ish YouTube) but I've seen a future which is a significant shift away from large teams and studios into what I like to call 'micro-studios'... teams of one to five people that use AI tools specifically to make more and better content. And it's not just about making more stuff, it's quite literally about the generation of new IPs and how to get to them quicker and test and see where it can go. Of course, the role of the human in this is VITALLY important but it will change from where we are now. This is a world where media truly becomes a co-production with AIs handling some of the grunt work going forward and, yes, likely will significantly reduce jobs in Hollywood. But, I hope, will allow us as humans to focus on doing the best creative work. When I give this talk, I spend a lot of time on the context of each slide (and I'm more than happy to come and give it your organization). There's a lot that gets left out here but I did want to get this idea out into the world because I think it's incredibly important for all media orgs to start thinking about how different the world will look soon. And, maybe most importantly, for the individual creator to understand the opportunity they have at their fingertips. Shout-out to Jonathan Verk for inviting me to NATPE to give this talk and for Emily Golden & Eric W. Shamlin for their great presentations at that talk! Would love to hear from you if this resonates at all or if you have thoughts. Drop them in the comments or send me a message!

Is xAI's new Grok 3 model good? Is it, as Elon says, 'scary smart'? Kevin P. and I go deep on this & much more AI News this week in AI For Humans ?? Let’s talk about Grok 3. There’s no denying that xAI has made huge strides—its benchmark scores are impressive, it passed the Chatbot Arena test with flying colors, and in deep-thinking mode, it holds its own against other top-tier models. But is it truly the smartest AI model, as Elon claims? That’s where things get complicated. The reality is, most frontier AI models are converging in capability. The biggest differentiator isn't raw intelligence—it’s who controls the data pipeline and how that data is used. xAI's biggest advantage? It’s directly integrated with X (Twitter), meaning it has a real-time firehose of global conversations to train on. That’s a strategic edge no other model has right now. Some takeaways from this week’s episode: ?? AI benchmarks can be gamed. Models can be over-trained to perform well on tests, but how they handle real-world tasks is what matters. ?? The AI industry is looking more like the streaming wars—too many players spending billions to compete. Will we end up with five dominant models, or will consolidation come? ?? AI agents are evolving fast, with new open-source versions of OpenAI's Operator emerging. The race isn't just about chatbots anymore—it's about agents that take action. ?? Microsoft is going all-in on AI gaming, hinting at a future where AI-powered game design reduces development time and enables new creative possibilities. ?? Robotics is advancing at an unreal pace. Whether it’s Unitree’s dancing robots (which look CGI-level smooth) or Palmer Luckey’s AI-powered defense systems, we’re moving towards a future where robots are no longer a novelty—they’re functional, powerful, and everywhere. It’s clear that we’re in a hyper-acceleration phase of AI. Every few months, what seemed impossible is now real. The question is: Where does it all lead? Would love to hear your thoughts—especially on Grok 3. Have you tried it yet? Does it pass your vibe check? And if you want to hear Kevin and me break it all down with some sharp analysis (and a lot of fun), check out this week's episode of AI For Humans: YT link: https://lnkd.in/gQkRkxFx #ai #ainews #grok3 #xAI #openAI

How good is OpenAI's Deep Research? What is ByteDance's OMNIHUMAN-1? Why did NVIDIA train robots on Lebron James? AI For Humans explains ?? This week, we’re diving deep (pun intended) into OpenAI’s new agentic research tool—does it live up to the hype? Plus, ByteDance is cooking up next-level deepfake tech with OmniHuman-1, and NVIDIA just trained robots on the movements of sports legends like LeBron, Ronaldo, and Kobe. Tim Simmons from Theoretically Media joins me to break it all down, including whether AI is finally becoming the ultimate research assistant and if humanoid robots are closer than we think. Kevin P. is on vacay with his fam and will be back next week! Watch the full episode here: https://lnkd.in/gsvgC6qn Let us know—will Deep Research change how we work? Or is it just another AI promise?

DeepSeek "crashed" the stock market. But DeepSeek AI R1's model is actually good for AI businesses like NVIDIA. How? Everything speeds up. ?? DeepSeek's R1 is a reasoning model—a distilled AI that's as good as OpenAI's o1 model but cost significantly less to develop. Marc Andreessen called it the "Sputnik moment" for AI, signaling a pivotal shift that could accelerate progress across the board. Here’s why it matters: 1?? AI is about to get better and faster for everyone. This competition will push innovation into overdrive, improving tools and making them more accessible. 2?? The pressure is on. A low-cost, high-performance model forces companies like OpenAI and NVIDIA to adapt quickly. Sam Altman even announced they’re moving release timelines forward. The stock market may have overreacted, but this isn’t bad news for AI. It’s the start of a faster, more competitive era—one that could reshape what’s possible in ways we haven’t fully grasped yet. #deepseek #ai #futureofai #machinelearning #artificalintelligence #techtrends

Did Google DeepMind's VEO 2 AI video model steal OpenAI's Shipmas thunder? We dive into it this week on AI For Humans! ?? Google AI is really cooking now and we're here for it. We break down Google’s surprising wave of AI product launches—from its stunning new video model (VEO 2) to enhanced capabilities within Gemini. We explore how Google may have just outmaneuvered OpenAI’s momentum, pushing the industry into a new era of 4K AI video, massively expanded context windows, and reduced hallucinations. We also discuss OpenAI’s latest moves, including the newly announced 1-800 ChatGPT number, NVIDIA’s new Jetson Nano unit for robotics, and how all these developments signal that advanced, more human-like AI interactions are closer than ever. Plus, we dig into the implications of big tech and Hollywood working on AI rights, Meta’s watermarking approach for authenticated video, and even try out Google’s Whisk to blend images, scenes, and styles seamlessly. If you’ve been looking for a snapshot of where the AI landscape is heading—and what it could mean for your business, creativity, or next big idea—this episode is packed with insights. As always, tune in and stay ahead of the curve with us! Full episode YT link here: https://lnkd.in/dAeVx6_F #ArtificialIntelligence #AIFuture #GenerativeAI #TechInnovation #MachineLearning #googleveo2 #openai

Is AI *actually* slowing down? According to reporting this week, you'd think the peak has happened, it's toast, it's all over. But is it? Absolutely not. This week on AI For Humans, Kevin P. and I go through the (multiple) stories of frontier model training runs not delivering as much advancement as expected and discuss why the shift might be moving from massive initial training runs to 'inference' training (we explain that). Plus, Kevin got hands on with Suno v4, the next stage of AI music models, and it sounds *significantly* better than any model I've heard before. The whole thing is available on YT below or on all your podcast platforms! YT link here: https://lnkd.in/gGKbu3iu