Nate Cooper's Linkedin Analytics

No, it is not. And here's why. 👇 🔴 We’re far from a world where AI can fully handle creative work on its own. Human input is still essential to craft ideas, review outputs, and guide AI decisions. But some things will change. 🟢 Repetitive tasks will be automated. 🟢 Creative work will become more efficient. 🟢 The focus will shift from "how" to do a task, to "what" task should be done, and from finding “the answer” to asking the right questions. Professionals across all industries will feel the pressure to master AI tools—and eventually, they will. As Fiverr’s famous campaign said: AI won’t take our jobs. It’ll take them to the next level.

Yikes. 🐜 "To execute the scheme, some of the defendants recruited “ants”, who pretended to be legitimate survey takers but instead were paid a nominal fee for completing surveys that produced fraudulent market research data. Some of the defendants even served as “ants” and fraudulently took large quantities of surveys themselves and received significant payment for their “ant” work." https://lnkd.in/gW4DZdXz

I really wish people in my industry would learn how large language models work so they stop putting out misinformation about the supposed search function. You do not rank in a large language model there's no such thing as a ranking signal in a large language model. Every large language model has a search engine that it uses to bring back documents TO SUMMARIZE (THEY ARE SUMMARIZATION ENGINES) when you put in a supposed query, to help ground the large language model so it doesn't write things that are fictitious. It will still write things that are fictitious. It is still a predictive text model Yes, that's a simplification there's a lot more that goes into it but it doesn't have a bunch of documents it brings back, the search engine bring back to the documents. Also generative AI search, again not really search it's a summarization engine, is less than 1% of all traffic to websites so it's not something you really have to stress about. Do your SEO well so that you're one of the documents that it summarizes from and hopefully they pull your link as a citation. does not mean it will always pull your link as a citation because they can pull a hundred documents or more to summarize from, but that is your best way to appear in the summarization engine. Oh and yes you need structured data because that is how OpenAI will pull in products into its new product engine and Bing uses it for its generative AI and Bing also feeds co-pilot and OpenAI.

“You should listen to opinions that don’t necessarily align with yours,” said my friend and shared a link to the latest Joe Rogan. Fair point - I listened (1:08-1:26 in the episode). And as someone who works in this field daily, I want to clarify a few things—because it’s too important to let misinformation slide. Here’s what I have to say:   1. “I’m not a climate scientist, but…” That’s exactly why we listen to climate scientists. Dismissing climate action as a “social contagion", led by "narcissistic psychopaths” is an insult to the science, and to the people already facing the consequences of climate change. Climate science is not about 'tyrants using fear and compulsion'. It’s about informed, measured risk management built on rigorous, peer-reviewed research. 2. “The Earth is in a cooling period” Misquoting a Washington Post article doesn’t change the science. The referenced article actually warns about unprecedented rates of warming, comparing today’s human-driven climate shift to the asteroid that wiped out the dinosaurs. This isn’t “natural fluctuation”—it’s a crisis moving faster than ecosystems and societies can adapt (read it yourself, link in comment). 3. “Carbon levels are only high if you pick your starting point” Carbon levels rose 35 parts per million over 1,000 years coming out of the last ice age 20,000 years ago. We've done the same in just 14 years. And we’re 8 billion people now, not hunter-gatherers. We're pushing our planetary systems at a breakneck pace, far beyond historic norms. 4. “But the Earth is 20% greener than it was 30 years ago” Yes—but not for the reasons claimed by Peterson. NASA’s most recent research shows this greening is largely due to human intervention, particularly massive tree-planting programs in China and agricultural efforts in India—not simply CO₂ fertilization. It’s a positive trend, but it doesn’t offset the loss of natural vegetation in biodiversity-rich regions like the Amazon and Indonesia. The gains are real, but so are the losses—and the costs of deforestation and habitat collapse are immense. 5. “Developing countries need fossil fuels to grow” Actually, many countries are leapfrogging straight to cleaner, decentralised energy systems. This isn’t charity—it’s smart economics. Clean energy brings jobs, lower energy costs, and improved health outcomes. Energy access and climate action go hand in hand.   Rogan and Peterson challenge the foundations of climate science with a blend of scepticism and conspiracy, arguing that climate action is about control, guilt, and fear. In my view, it’s the opposite. It’s about taking responsibility, not burying our heads in the sand, and using the best science, tools, and innovation we have to protect communities, ecosystems, and our future. Curious to hear what others think…

Some notes on "innovation" and the "Disrupted States of America"

My National Science Foundation grant on AI education was cancelled on Friday. Why? Not that I was actually told, but if I had to guess, it's because of this sentence in the grant's abstract: "Informal learning for young people increasingly takes place on social media, which is also full of myths, misconceptions, and misinformation around AI." This research is--was, I suppose--focused on interdisciplinary groups of students as learners and educators to create high-quality, accurate content about AI for the general public with a focus on the topics most important for young people. The government announcement on Friday cited blanket reasoning that "NSF will not support research with the goal of combating ‘misinformation,’ ‘disinformation,’ and ‘malinformation’ that could be used to infringe on the constitutionally protected speech rights." Which, though an incredibly important area of research that absolutely did not deserve to be cut regardless, is not what my grant is actually about. What we appear to have here is a simple word search for the word "misinformation" and without further consideration for context, the clawing back of money for research in an area that in theory is important to the current administration (AI workforce development). The loss of this funding means that this work will not happen, a PhD student will not have research funding next year, and the undergraduates who would have been involved will not get this research funding and experience. (Also, selfishly, I will not have summer salary that I was counting on.) We are also having to scramble a bit to make sure that we can pay the stipends to students that were set to be paid out at the end of the semester in two weeks, for the work that they already did on the project this semester. (They WILL be paid even if it has to come out of my own discretionary funds, though I'm hoping it doesn't come to that.) Feel free to share this as an example of the material impact and utter carelessness that we're seeing in these cuts to federal research funding in the U.S.

I've been working on the Trillion-Dollar LLM Hallucination Problem for about a year and a half. I've noticed that the Big Tech companies (OpenAI, Google, Anthropic, Meta, Apple) have not solved the problem; in fact, they are recently reporting that the hallucination problem is getting worse as they try to advance into reasoning and planning, and both OpenAI and Anthropic are saying that they need to do further research in this area. There are many vocal critics of LLMs (Gary Marcus, Yann LeCun, Dr. Jeffrey Funk, Denis O., Richard Self.) The common refrain is that present-day LLMs are a dead end on the path to human-like intelligence, and that some kind of new architecture is needed. I agree with that. I have a dozen friends on LinkedIn who have alternative architectures that could be candidates. (Kyrtin Atreides, Vincent Granville, John Ball, Srini Pagidyala, Peter Voss, Walter C., Luciano Zorzin, Hristo Georgiev, etc.), and I am following them and cheering for them to succeed. And of course there is Yann LeCun's JEPA architecture and Fei-Fei Li's World Model, and whatever mysterious ideas are being developed by Ilya Sutskever and Mira Murati at their new Billion-Dollar startups. For me, the Trillion-Dollar Question is: Can we find a new architecture, that can build on the remarkable successes of LLMs, while overcoming their serious problems? I have found it productive to focus on the Hallucination Problem. LLMs hallucinate because they are sampled auto-regressive next-word predictors, trained on internet-scale text data. Once they are trained to predict next words very well, they do not use their training data at inference time. They are able to predict next words very well, to make syntactically correct language, but they do not explicitly remember any statements from their training data. They discard any memory of statements which could be considered true. They predict likely next words. They are not constrained to make true statements, and they have no way of reliably doing so. To remedy the problem, I have had to develop Hallucination Detector and Deep Attribution Networks, to allow an LLM to natively remember and refer to its training data. This is a separate, integrated subsystem that functions as a compressed memory of factual statements. And finally, an Executive Selector can override the LLM when it is veering into a hallucination, and substitute a key next word that is directly derived from the relevant true statements in the training data. The language part of the human brain (Broca's Area) is NOT the memory controller (Hippocampus), and it is not the Executive Selector (Frontal Networks). This new architecture uses an LLM for what it is good for: producing syntactically correct language. It augments the LLM with memory to get the facts right, and a Hallucination Detector and Executive Selector to satisfy the joint objectives of producing valid language and factual statements. #ai #llm

The more I vibe-code, the more I realize: this isn’t a tool for newbies or non-coders to replace no-code or low-code tools. It’s something much different — a powerful but highly unstable technology, best suited for expert coders. From the outside, vibe-coding can look like magic. It feels as if Ralph Wiggum from The Simpsons could suddenly build the next Google all by himself. But that’s not going to happen — not now, and maybe not ever. A few observations from my recent vibecoding journeys: • You need a strong mental model of what you’re building. Vibe-coding can generate thousands of lines of code in minutes. But is it the code you actually need? Where should the tool evolve further? To give a mental analogy: imagine that building software is like drawing a disk — and over time, your disk should steadily grow larger. With vibe-coding, you can draw very long lines very fast. But if you want a disk, you must maintain a clear mental picture, carefully guiding the tool to fill in adjacent lines until the circle emerges. • You need deep software instincts to avoid bloat and overengineering. Models are naturally eager to produce code. What should have been a simple script can easily explode into an “enterprise application” — bloated with unnecessary layers and abstractions. A trained eye quickly notices that much of the generated code is not only redundant but actively harmful to maintainability. • You need to become serious about software testing. As someone in ML, I was rarely exposed to real-world testing. I knew about testing in theory — the techniques, the best practices — but not how to implement or prioritize them. Vibecoding changed that. If you look at the QuackVerse, our open-source project at AIProduct.Engineer the number of lines dedicated to tests is now almost equal to the lines for business logic. You either learn to test well or drown in your own code. Is all this transitory? Some optimists say: It will never get worse than this — it can only get better. I’m not so sure. Progress in machine learning isn’t linear — it’s stepwise. You can see it in the latest reports on Llama 4: it excels at some tasks but regresses in others where earlier Llama models were strong. Improvements aren’t guaranteed across the board. Sometimes, progress stalls — for years, even decades. Think about chatbots: they were laughably bad for a long time, until suddenly they became astonishingly good. It’s entirely possible that, within our lifetime, we get stuck in a world where AI remains good enough to help, but not good enough to replace coders — and where Ralph Wiggum still can’t build the next Google on his own. In this age of vibe-coding, do you feel like Ralph Wiggum, suddenly enlightened by the ability to summon thousands of lines of code in minutes? Or has it made you realize that software development remains a deeply challenging craft — just like everyone can draw a horse, but not everyone gets to be Picasso?

I asked ChatGPT for 10 ways that autocrats around the world undermine democracy. Here's what it said (unedited)... "That’s a thoughtful and important question. Understanding how democracies can erode is crucial for preventing it. 1. Undermine the Free Press: Discredit independent media as “fake news,” take over public broadcasters, and intimidate journalists to limit dissent and control narratives. 2. Attack the Judiciary: Undermine judicial independence by appointing loyal judges, removing or intimidating uncooperative ones, and limiting judicial review powers. 3. Weaken Checks and Balances: Expand executive power, bypass legislatures, or co-opt them through patronage or intimidation to reduce oversight. 4. Undermine Electoral Integrity: Change election laws, gerrymander districts, limit access to voting, and use disinformation to manipulate outcomes. 5. Exploit Emergencies: Use real or manufactured crises (like terrorism or pandemics) to justify emergency powers and bypass democratic procedures. 6. Silence Dissent: Criminalize protest, restrict civil society organizations, monitor citizens, and use police or intelligence services to harass opponents. 7. Co-opt or Crush the Opposition: Buy off political opponents, ban parties, or jail opposition leaders on dubious charges. 8. Use Nationalism and Scapegoating: Stir up ethnic, religious, or cultural divisions to unite supporters and justify repressive measures. 9. Undermine Truth and Facts: Flood public discourse with disinformation and conspiracy theories to erode trust in institutions and shared reality. 10. Extend Rule Indefinitely: Eliminate term limits, rig referendums, or rewrite constitutions to stay in power indefinitely. These steps don’t always happen in the same order, and sometimes they unfold gradually. Would you like examples of countries where some of these steps have been observed?"

🚨 I've been hearing from university professors that ChatGPT is completely destroying learning and that we'll soon have an AI-powered semi-illiterate workforce. Here's a glimpse into the educational apocalypse: "Sarah, a freshman at Wilfrid Laurier University in Ontario, said she first used ChatGPT to cheat during the spring semester of her final year of high school. (...) After getting acquainted with the chatbot, Sarah used it for all her classes: Indigenous studies, law, English, and a “hippie farming class” called Green Industries. “My grades were amazing,” she said. “It changed my life.” Sarah continued to use AI when she started college this past fall. Why wouldn’t she? Rarely did she sit in class and not see other students’ laptops open to ChatGPT. Toward the end of the semester, she began to think she might be dependent on the website. She already considered herself addicted to TikTok, Instagram, Snapchat, and Reddit, where she writes under the username maybeimnotsmart. “I spend so much time on TikTok,” she said. “Hours and hours, until my eyes start hurting, which makes it hard to plan and do my schoolwork. With ChatGPT, I can write an essay in two hours that normally takes 12.” - "By November, Williams estimated that at least half of his students were using AI to write their papers. Attempts at accountability were pointless. Williams had no faith in AI detectors, and the professor teaching the class instructed him not to fail individual papers, even the clearly AI-smoothed ones. “Every time I brought it up with the professor, I got the sense he was underestimating the power of ChatGPT, and the departmental stance was, ‘Well, it’s a slippery slope, and we can’t really prove they’re using AI,’” Williams said. “I was told to grade based on what the essay would’ve gotten if it were a ‘true attempt at a paper.’ So I was grading people on their ability to use ChatGPT.” - It would be great to hear more from teachers and educators to understand what might be potential solutions. I'm also adding below the essay I published today about the hidden dangers of AI chatbots.

I tested AI coaching for 30 days. The results shocked me. As the founder of a peak performance coaching community, I'm about to share something that might put me out of business. It’s a topic we debate plenty in the Transform Perform Coach team chat. But the truth is, some simple prompts, with an LLM, can help you level up in life. Start with this prompt, courtesy of Zain Kahn's Superhuman newsletter (link in comments👇). Prompt: "Act as a purpose-driven business coach. Help me discover my Ikigai — the intersection of what I love, what I’m good at, what the world needs, and what I can be paid for — and guide me in turning it into a business or career path. Start by asking me reflective questions in each of the four Ikigai dimensions to uncover my passions, strengths, marketable skills, and potential value to others. Once we’ve mapped those, suggest business ideas or career directions that align with my Ikigai. For each suggestion, explain how it connects to the four elements and provide examples of real-world businesses or roles that reflect similar paths. Include prompts for journaling, self-assessment tools, and small experiments I can run to test alignment. End with a one-page Ikigai roadmap summarizing my core themes, possible paths, and first action steps toward building a purpose-driven life or business." This simple prompt can help you to... Discover your purpose through these core elements: 👉 Align your passions with marketable skills 👉 Identify where your talents meet global needs 👉 Test small experiments in your chosen direction 👉 Track insights through guided self-reflection Transform your career path: 👉 Start with 15 minutes of daily journaling 👉 Map your strengths against market opportunities 👉 Create micro-projects to validate your direction 👉 Build evidence through real-world testing So why would I encourage you to start experimenting with AI coaching in this way? The ultimate paradox of coaching: ✨ My role is to help you not need me ✨ Your success means becoming your own guide ✨ True growth happens when you lead yourself So, are you ready to begin? ♻️ Share this with someone who needs an empowering high five 👉 Follow Liz Bradford for insights to boost your wellbeing, career and augment your business.

Service design thinking aligns people, processes, and systems. We agree with Aishwarya Dwivedi that great products don’t just come from designing good features. She argues they come from understanding and designing the whole system behind them. Many teams focus too much on the interface and overlook the people, tools, and processes that make everything work. Service design thinking helps teams spot problems early, avoid launch chaos, and build smoother user experiences. Check out her article: https://lnkd.in/gzWPDUfD To address these challenges, she encourages teams to:​ → Map the entire user journey, including behind-the-scenes processes. → Identify all stakeholders and systems involved at each step. → Recognize data dependencies and potential failure points. This approach helps product teams spot issues early and build smoother user experiences. 💬 We asked Aishwarya why she wrote the article: “I still remember the moment it happened. We had just lunched a feature and a critical bug had surfaced, and the PM on my team couldn’t solve it because he only understood his piece of the puzzle. That’s when I saw it clearly: in the name of efficiency, most teams work in silos. I’ve seen it in companies I’ve been part of, and I’ve heard the same from friends and people I’ve coached. We celebrate shipping features, not stitching systems. That’s when I realized the real need to understand how the entire system works. And for me, service blueprints became more than just a framework—they became a way to see the whole picture. “ Love it. Helio helps you test ideas with users early so that you can spot problems before launch. It connects user feedback and UX metrics to the experience, not just the interface, making it easier to build great products. #uxresearch #productdiscovery #marketresearch #productdesign

PSA to Tech Founders + VCs: The Creative Director Is the Most Underrated Hire in Tech. In a world where AI has commoditized execution, taste is the new growth lever. And the Creative Director? They’re your moat. Not the prompt engineer. Not the growth hacker. Not the product manager. The creative director. In an age of AI, they’re no longer a “nice-to-have.” They’re as critical to your business as your CTO or Head of Product. Because AI didn’t just change how we work, It commoditized it. Design. Code. Strategy. Content. Automated. Outsourced. Replicated. In a world where software is infinite, brand is your final moat. And the creative director is the architect of that moat. ⸻ Welcome to the era of the Creative Polymath. Not the designer with the best Figma shortcuts. Not the marketer chasing shallow engagement and fleeting trends. But the one who: • Thinks like a director • Writes like a screenwriter • Imagines like a set designer • Consumes film, architecture, fashion, subcultures… • And knows what feels iconic, long before the data proves it. They don’t just make decks. They design new realities. ⸻ Because in a world where anyone can generate content… The value isn’t in creation anymore. It’s in the curation. And that requires impeccable taste. Taste you can’t teach with mood boards, dashboards or spreadsheets. This is why the right creative director is no longer a cost center, They’re your cultural strategist, your growth multiplier, your maker of meaning, Your 10x creative engineer. ⸻ The most overlooked strategy in tech? Hiring someone who innately knows what not to ship. Because that’s the creative director’s superpower. Discernment. They will be the soul of your story. The architect of your world. The person your CEO will turn to and ask: “Do you believe in this?” ⸻ And here’s the shift no one is prepared for: As AI levels the playing field, brand becomes the business strategy. Not simply a function of it. And brand won’t be led by committees or CMOs obsessed with data and alignment. It will be led by creatives with vision. With taste. With range. With teeth. ⸻ So if you’re building the next great company… Don’t just hire someone who can make it look good. Hire someone who can make it feel inevitable. Because in the coming decade, your brand will be the only thing they can’t copy. And the creative director? They’ll be the one designing what can’t be commoditized. ⸻ Investors hunt for visionary founders. Startups chase engineering unicorns. But if you’re not also hunting for a creative savant, and world-builder, you’re building a castle with no walls. Read that last part again. ⸻ My advice: take a note from David Carradine’s 1970’s hit show, Kung Fu, Wonder the earth. Find yours. Create your moat. Because in the decade ahead, taste will build empires.

Some of the smartest CEOs we know are saying it: Master AI… or get replaced. So here’s how: BTW, it’s not just CEOs. I’ve been talking to top product leaders, and they have the same expectations of their PMs. And here’s the thing: the gulf between those who just use AI vs those who are masterfully using it widens everyday. To keep you on the right side of the curve, I put together a full roadmap. PART 1 - USING AI AS A PM 1. The AI PM's Playbook: How Top Product Managers Are 10x-ing Their Impact in 2025 - https://lnkd.in/ef8TkEdQ 2. The Ultimate Guide to AI Prototyping - https://lnkd.in/eJujDhBV 3. Tutorial of Top 5 AI Prototyping Tools With Colin Matthews - https://lnkd.in/eit9MnaR 7. How to Use ChatGPT as a PM - https://lnkd.in/eyfUi8ZN 5. We Built an AI Product Manager in 58 mins (Claude, ChatGPT, Loom + Notion AI) - https://lnkd.in/eAZZ8MSb 6. We Built an AI Agent to Automate PM in 73 mins (ZERO CODING) with Anthony Maggio - https://lnkd.in/ebF8hhvx 7. How to PM with AI at Early Stage Startups with Sergio Pereira - https://lnkd.in/eWt7UciA PART 2 - AI PM SKILLS 1. Complete Course: AI Product Management With Paweł Huryn - https://lnkd.in/eN2_-6xi 2. Use AI To Draft Your Email and Slack Messages with MCP - https://lnkd.in/ejMMjAAV 3. 21 Harsh Truths about Product Management in AI with Scott Jones - https://lnkd.in/esYZcN7N PART 3 - BUILDING AI PRODUCTS 1. How to Build an AI Startup - https://lnkd.in/e4MaHeKM 2. Most People are Building AI Products Wrong - Here's How to do it Right - https://lnkd.in/eDGmsvZ5 3. Product Management in the Era of AI with Sam Kawsarani - https://lnkd.in/eGjTrvQe PART 4 - BECOMING AN AI PM 1. How to Become an AI Product Manager with No Experience - https://lnkd.in/e85tyhxy 2. How to become an AI Product Manager - https://lnkd.in/emUdezbS 3. How to Write a Killer AI Product Manager Resume (With Examples) - https://lnkd.in/ehAbcRGy 4. How to Become an AI PM With Marily Nika, Ph.D - https://lnkd.in/eid66mjS 5. 2025 AI PM Job Search Masterclass with Dr. Nancy Li - https://lnkd.in/eqqMD_Ne The next 12 months will separate the AI-masters from the AI-users. Which will you be?