Mike Kaput's Recent LinkedIn Posts

We asked 80 AI professionals if most jobs will be automated in 18 months. 85% said no...but almost nobody said it won't happen at all. Every week, we survey listeners of The Artificial Intelligence Show to get an informal snapshot of how professionals are actually thinking about AI. It's not scientific. It's 80 people who care enough about AI to take 30 seconds and tell us where they stand. This week, we asked about Microsoft AI CEO Mustafa Suleyman's prediction that most white-collar work will be fully automated within 12-18 months. The pushback was overwhelming. But it was pushback on the timeline, not the direction. Half of respondents said it's partially realistic. Some tasks will be automated in that window, but not most. Another 35% said it's too aggressive and meaningful automation is more like 3-5 years out. Only 8.8% said they're already seeing this level of automation in their own work. And just 6.2% dismissed the premise entirely. That's what stands out to me. This is an AI-literate audience. They follow AI closely. They use these tools. And they're being remarkably level-headed about it. They're not buying the hype timeline. But they're also not pretending automation isn't coming. The rest of the data tells some interesting stories too: 📹 When asked about AI-generated video using real people's likenesses, not a single respondent dismissed the concern. 22.5% have already moved past "should this happen" to "how will it be governed." 📚 75% say their organization is doing something with AI training, but only 31.3% have formal programs. The largest group is still in informal experimentation mode. Again, just an informal snapshot from 80 respondents. But it paints a picture I keep seeing in this data week after week. People know what's coming. They're just not convinced it's coming as fast as the headlines say. That feels about right. Full breakdown with methodology:

Entry-level hiring is down 14%. This isn't an entry-level problem. Anthropic just published a study looking at how AI is actually being used versus what it can theoretically do. One of the findings: job posting rates for workers aged 22-25 entering AI-exposed fields have dropped roughly 14% compared to 2022 levels. Companies aren't doing mass layoffs. They're just quietly not hiring. That's the part that should get your attention. Not because of what it means for new grads, but because of what it previews for everyone else. If the acceleration continues on the path we're seeing, every knowledge worker is going to have to ask themselves some hard questions about how exactly they create value. Those questions may have similarly shaped answers regardless of your level. The best advice I'd give an entry-level hire right now is to become the most AI-enabled version of whatever it is they do. That's also the best advice I'd give myself.

The biggest AI problem isn't technology. It's this: The intern doesn't understand it.  The manager fears it.  The executive ignores it.  IT tries to control it. Result? Chaos. The solution is simple (but not easy): AI literacy for all. Not just the tech team.  Not just the innovators.  Everyone. From intern to CEO. Because AI isn't a department... It’s not a specific discipline... It's a universal capability. One that affects everything and everyone.

"I'll focus on AI next quarter." Said every leader, every quarter, since 2023 Here's the thing about "next quarter":  It never comes. You won't find the time.  You won't clear your calendar.  You won't have less pressure. I get it (and also struggle with it)… But we need to be honest with ourselves: It's unlikely we're suddenly going to do: - Three hours of AI work a night - A month-long bootcamp - Some huge AI initiative That's OK. But let's acknowledge it. And respect that fact. Instead focus on more small wins. More consistency. Make ONE small AI move today… During the work you’re already doing. Every single day. Inch by inch. Task by task. Workflow by workflow. Tomorrow you'll still be busy.  But, eventually, you'll be busy with AI.

A hard truth about AI adoption? (And it surprised me...) 100% of leaders know it's critical. But only a fraction do something about it. Often, the barrier is in their heads. More than one leader has said to me: “I’ve been meaning to build a GPT.” It takes 30 seconds to do it. So, why haven’t they? It’s not a technical gap. It’s a psychological one. While you're waiting for the perfect moment...  The perfect tool...  The perfect strategy... Someone else just spent 30 seconds getting started.

Here’s an opportunity in AI… That a LOT of people miss: We all want to be more productive. We want AI to: Eliminate tasks. Make work faster and better. Automate workflows. But there’s a next step that most people miss: Fill this time with MORE human stuff. Use AI to handle the analytics… So you can pick up the phone. Use AI to draft the reports… So you can build relationships. Use AI to generate ideas… So you can have better conversations. I think an increasingly important question is: How can we use AI to capture huge efficiencies… So we can do more of what DOESN’T scale?

Here’s a quick AI win… That most people are missing: AI has quietly revolutionized research. Just in the last few months. Deep research tools… Google’s NotebookLM… Models equipped with better web search… These tools can do days of research in minutes. Most people aren’t yet aware of that. Or haven’t fully explored what’s now possible. So, if you haven’t already: Getting familiar with AI-powered research is one of the highest ROI things you can do right now.

Can't WAIT to go back to school... (AI Marketing School that is.) On Sept. 2, I'm joining some of the best in the biz to talk AI at AI Marketing School from 12pm - 3pm ET. I'll be talking AI for content workflows, but... There are a ton of other tracks you should check out. Join us? Register below 👇 https://lnkd.in/ggX3dECc

Companies spend millions on… AI software and consultants, but: Won't invest in AI training. That's like buying a Ferrari...  And never learning to drive. AI doesn't matter if your people can't use it. So your biggest AI investment isn't technology. It's education. Train your team.  Empower your people.  Build AI literacy. Because the best AI strategy in the world fails... Without people who can execute it.

75% of marketing teams don't have an AI roadmap. That's not a typo: According to our latest data… Three out of four teams are just...winging it. They have no roadmap of AI use cases. No plan.  No priorities. Here's the uncomfortable truth: You can't afford to figure it out "later." Later is now.  Tomorrow is too late. Start simple: → List out your daily tasks  → Pick one to do better with AI → Test for 30 days  → Repeat Stop waiting... Stop deliberating... Just get started.

With OpenAI’s launch of its agentic AI web browser Atlas this last week, I’m thinking a lot about what it means for businesses. It seems likely that, at some point in the next couple years, what we all do online will likely be mediated at some point by an AI agent. At the very least, our experience will be filtered by a highly personalized instance of a large language model. If that happens, businesses lose a lot (perhaps all?) control or influence over the research and buying journeys. The entire process will be intermediated by an AI tailored to a user's specific, personal tastes. So you won't really control what people see related to your business. Or how they see it. That'll be AI's job. And I suspect you'll have far less control over how AI does that job. I don’t really know yet what all of this means or what to do about it. But it strikes me that the immediate goal might be to simply make sure everything about a business (pricing, features, value, etc.) is transparently available online, in a format that makes it easy for agents to find and understand. That way, you know an agent can at least add you to the conversation when a user asks it for information, pricing, or product options.

There's a disturbing AI trend gaining momentum: The spread of "nudify" or "AI undressing" apps. These tools can generate realistic nude images from a single photo, and they're spreading rapidly on platforms like Telegram and Discord, as AI ethics researcher Rebecca Bultsma highlighted this in the post below. Look, it’s an awful topic to discuss. But I think we need to at least make ourselves aware it’s a thing. And it’s a thing that’s becoming more widespread. (These apps have millions of monthly views.) I don’t know what the answer is here… I'm not sure there's any real way to stop the technology itself, especially with open-source models. It seems the platforms should come down hard on these apps, but even that's a partial solution. As Rebecca's post notes, so much of this is found and disseminated through private channels like Discord and Telegram, which are difficult to police. (Not to mention, I’m deeply skeptical platforms will do an adequate job policing these.) It feels like we have to accept an ugly truth: Nobody is really coming to save us on this. If the technology can't be stopped and the distribution can't be fully controlled, it puts the burden right back on us at the individual and family level. (Whether that’s fair or not.) It means we're the ones who have to have the hard conversations and educate ourselves, our families, and our communities about what's now possible…and about what is and isn’t acceptable in our own use of this technology. And, as uncomfortable as it may be, the first step in that process is awareness.

OpenAI has launched ChatGPT Atlas, its new AI-powered web browser...but I’m struggling with it. It’s designed to be a companion that lives alongside your web activity, capable of summarizing pages, comparing products, and analyzing data from any site. The standout feature is the new "agent mode," a preview tool that can autonomously take actions on websites under user supervision, like navigating a retail site or even purchasing groceries. Despite how impressive the technology is, after trying it, I’m not currently finding a lot of relevant, valuable agentic use cases in my own work in a browser at the moment. The core issue for me is the "AI verification gap." For an agent to be truly useful, it doesn't just need to work… I need to be able to verify that it worked correctly. Right now, it feels like verifying an agent's autonomous actions (and checking for all the potential security issues or simple mistakes) would take me significantly more time and energy than just doing most things myself. I have no doubt this is where the technology is going. But it seems that until this verification gap closes, these powerful agentic tools may struggle to move from technically possible to practically useful for a lot of what I do. I’m sure that will change, but I’m personally not there yet. Does anyone have any killer use cases that they love for agentic AI within your browser?

It’s easy to get headline whiplash trying to track the real-world impact of AI. Not long ago, a widely-cited MIT report suggested 95% of generative AI pilots were failing. This week, a new large-scale Wharton AI adoption study dropped, and it paints a dramatically different picture. The Wharton study, now in its third year, found that 75% of business leaders are already seeing a positive ROI from their AI investments. Fewer than 5% reported negative returns. And, nearly 90% of companies plan to increase their AI budgets over the next year. So, which is it? Are we failing or succeeding? It turns out, these two reputable sources are showing vastly different realities because they're measuring different things. The MIT study, for instance, had a specific definition of "success" (deployment beyond the pilot phase with measurable KPIs). The Wharton study surveyed senior decision-makers at large enterprises (1,000+ employees) about their perception of ROI. I have my own opinions on which study makes more sense to me. But the real point, for me, isn't about which study is "right" or "wrong." It’s about how easy it is to get distracted. It feels like everyone (doomers and optimists alike) is tempted to grab the headline that justifies their emotional perspective. I get that. I fall into the trap all the time. It's valuable to track these studies, and it's even more valuable to dig into their methodologies to try and extract the signal from the noise. But I’m finding that getting too attached to any one study, or getting caught up in the debate, is its own kind of distraction. It pulls focus from the only thing that really matters at the end of the day: The personal, practical work of figuring out how to apply AI to your job and your life to make your job and your life better.

I'm super excited for our AI for Agencies Summit coming up in February... = It's designed specifically for leaders who want to use AI to: - Increase efficiency and productivity - Enhance creative output - Deliver more value to clients - Build scalable, future-ready agencies 🔗 Register for FREE here: https://lnkd.in/eniNqXv4

"I tried AI once. It didn't work." Of course it didn't. Neither did: Your first bike ride… Your first presentation… Your first sales call… Your first anything. AI is no different. You'll fail.  A lot.  And that's the point. Every failure teaches you: → What doesn't work → What to try next → What to avoid → What to improve The people winning with AI? They failed just as much as you... They just didn’t stop experimenting.

Everyone's trying to keep up with AI. Here's what no one will tell you: It's impossible. And that's okay. New tools launch daily.  Models release weekly.  Features update constantly. You'd need 3 full-time jobs just to track it all. But here's the plot twist: You don't need to know everything.  You just need to know what matters. What AI means for: Your job.  Your goals.  Your growth. Focus beats FOMO every time.

One thing we teach in our workshops… That may be helpful: Your job is a bundle of tasks. Here's what I mean... That big fancy job title? It's not 1 big responsibility. When you break it up... It’s really 100s of smaller tasks. And AI can help with many of them. So if you’re stuck trying to find AI use cases… Try this: Stop zooming out. Start zooming in. Look at: → Your to-do list → Your calendar → You next meeting agenda Each one is a treasure trove of use cases. Stack enough of those use cases… And suddenly you’ve got a whole AI-powered job.

"I don't have enough data for AI." Wrong. You have data.  You just don't know it yet. Every conversation: Data.  Every meeting: Data.  Every email: Data.  Every process: Data. Documents, language, content… It’s all data you can use. But here's the catch: If it's not documented, it doesn't exist. Start today:  → Write down that process  → Record that explanation  → Document that decision  → Save that conversation Next week when you need AI's help? You'll have everything ready. The best time to start was yesterday.  The second best time is now.

Smart professionals have mentors. Smarter ones also have AI advisors. For instance, you can create: → A GPT trained on your job → A GPT trained on your domain (i.e. copywriting) → One trained on your department → One for each key project → One for each core skill you want to improve → And so on and so forth. So, anytime you need advice? Your team of advisors is always available.

AI has a -20 favorability rating. And it's probably going to get worse. That's from a new NBC News poll of 1,000 registered voters, asked to rate their feelings on public figures, institutions, and topics. Here's where AI landed: 🔹 The Pope: +34 🔹 Stephen Colbert: +10 🔹 Trump: -12 🔹 ICE: -18 🔹 AI: -20 🔹 Democratic Party: -22 🔹 Iran: -53 Only the Democratic Party and Iran scored lower. I think this is just the beginning. The benefits of AI are real. But they're abstract. Scientific breakthroughs, productivity gains, new capabilities. They take time to show up in people's lives. The costs are personal. It's your neighbor who lost a role. It's the entry-level job that quietly disappeared. It's the team that got restructured. As the technology touches more areas of daily life, as more people experience disruption firsthand rather than just read about it, I'd expect this number to get worse. Possibly a lot worse.

The Pentagon blacklisted Anthropic for two AI safety red lines. Hours later, OpenAI signed a Pentagon deal with those exact same terms. Episode 200 was something else. Here's what Paul Roetzer and I just covered on Episode 200 of The Artificial Intelligence Show: 🔹 The Trump administration designated Anthropic a "supply chain risk to national security" after CEO Dario Amodei refused to let the Pentagon use Claude for mass domestic surveillance and fully autonomous weapons. 🔹 Hours after Anthropic was blacklisted, Sam Altman announced OpenAI had signed a Pentagon deal that included the exact same prohibitions on surveillance and autonomous weapons. 🔹 OpenAI closed a $110 billion funding round, the largest private financing in history, with $50 billion from Amazon, $30 billion from NVIDIA, and $30 billion from SoftBank. 🔹 The creator of Claude Code said he hasn't manually edited a single line of code since November 2025, and that 4% of all public GitHub commits are now authored by Claude Code, a number he predicts will hit 20% by year-end. 🔹 Jack Dorsey cut nearly half of Block's workforce, roughly 4,000 people, and explicitly named AI as the reason. The stock surged 24%. 🔹 A Substack essay called "The 2028 Global Intelligence Crisis" modeled an AI-driven economic collapse, made it to Wall Street trading desks, and triggered a real selloff and a formal Citadel Securities rebuttal. 🔹 NVIDIA reported $68.1 billion in quarterly revenue, beating every analyst estimate. The stock dropped 5.5% the next day. 🔹 Anthropic accused DeepSeek, Moonshot AI, and MiniMax of running industrial-scale distillation attacks, using 24,000 fake accounts to extract over 16 million exchanges from Claude. 🔹 Public support for AI data centers collapsed to negative 24% net approval, now polling worse than nuclear power plants. 🔹 Anthropic published its AI Fluency Index, finding that the better AI output looks, the less people bother to verify it. Oh, and we recorded this one live for the first time, in front of our AI Academy Mastery members! 200 episodes. Still feels like we're just getting started. Link to Episode 200 in the first comment.

Your new hires don't see AI as optional. They see it as oxygen. While we debate AI policies...  They're already using it. While we form committees… They're forming habits. While we assess risks...  They're assessing efficiency. By high school, AI is their default.  By college, it's their DNA.  By hiring, it's their expectation. The risk isn't in using AI.  It's in pretending they're not.

It’s probably time to rethink meetings… Thanks to AI. Typically, during a meeting you: - Take some notes - Assign some action items - Move on But I see AI leaders instead: → Have AI listen in real time → Actively prompt it during meetings → Record every meeting → Use that context for follow ups And I’ve seen it for myself: We’ve solved really tough problems this way… By treating AI not just like a note-taker, but like an actual meeting participant.

I keep catching myself doing things one at a time. That's the old way. The more I work with AI, the more I realize the real skill isn't prompting. It's orchestration. Multiple agents, multiple instances, working on different parts of the same problem in parallel. It's a fundamentally different way of working. Instead of doing step one, then step two, then step three, I'm running steps one through five simultaneously and figuring out the right order of operations to stitch them together. I have to keep reminding myself to think this way. The instinct to do things sequentially is deep. It's how I've worked my entire career. But every time I catch myself and ask, "What else could be running right now?" the output gets dramatically better and faster.

Time and time again… I’ve found it helpful to remind myself: Most people… Think AI mastery takes years  Assume it has to all be complicated Believe you need a tech background to succeed Here's the reality check: →  Your next great prompt can take 5 minutes  →  Basic use cases transform work constantly → This doesn’t have to be rocket science Start with the next important task.  Master it with AI. Move to the next. That’s it. Getting further with AI is really just putting one foot in front of the other, every single day.

Something helpful I’ve found… To get more out of reasoning models: It’s less a conversation… And more a consultation. Don’t: → Fire off quick questions → Expect instant replies → Go back and forth rapidly Do: → Take time to gather context → State a complex goal → Take time process the full output There’s a lot less chat… And a lot more thinking involved.

Microsoft's AI CEO just predicted full job automation in 18 months. ...It was that kind of week. Here's what Paul Roetzer and I just covered on Episode 198 of The Artificial Intelligence Show: 🔹 Microsoft's AI CEO Mustafa Suleyman predicted that accounting, legal, marketing, and project management will be "fully automated" by AI within 12-18 months. 🔹 Stanford economist Erik Brynjolfsson says AI is now driving a measurable productivity surge, with US growth hitting 2.7% in 2025, nearly double the prior decade, while payroll was revised downward by 403,000 jobs. 🔹 Anthropic CEO Dario Amodei says we are one to three years away from AI systems that match or exceed human expertise across virtually every cognitive domain, and the world isn't taking it seriously enough. 🔹 AI filmmakers claimed they matched a $200 million Hollywood production in 24 hours using ByteDance's Seedance 2.0, and someone generated a deepfake of Brad Pitt fighting Tom Cruise from a two-line prompt. Hollywood responded with cease-and-desists. 🔹 Anthropic, Google, and xAI all released major new AI models in the same week: Claude Sonnet 4.6, Gemini 3 Deep Think, and Grok 4.2. 🔹 The creator of OpenClaw, the fastest-growing open source project in GitHub history, was hired by OpenAI to build its next generation of personal AI agents. 🔹 Meta was granted a patent for AI that keeps posting from dead people's accounts. 🔹 Anthropic raised $30 billion at a $380 billion valuation, making it the second largest venture deal of all time. Oh, and 70% of our audience this week said AI is "accelerating faster than I can keep up." After covering all of this in one sitting, it's hard to disagree. Link to Episode 198 in the first comment.

Reasoning models are starving. Not for data. For context. When using reasoning models… Most people don't give nearly enough context. Say you want help on a marketing plan... Winners give reasoning models this:  → Company history  → Market position  → Competitor analysis  → Past campaigns  → Budget constraints  → Team capabilities  → Success metrics → Failed attempts → And more The more you feed it... The better it performs. Your context = your competitive advantage.

How does your AI adoption stack up? Most companies have no idea. We're running our annual State of AI for Business survey to get a real picture of how organizations are actually adopting AI across every function, industry, and company size. It takes about 5 minutes and your answers are anonymous. The results publish this spring as the State of AI for Business Report. It's one of the only places you can see grounded data on what's actually happening with AI inside companies, not just what the vendors and headlines are saying. Here's what we're looking at: - Where AI is gaining real traction and where it's stalling - What leaders are prioritizing and what's getting in the way - How companies are handling workforce planning around AI - How your organization compares to 1000s of others Your perspective genuinely matters here. The more people who respond, the more useful the data is for everyone trying to figure this out. (Not to mention, if you complete the survey, you'll be entered for a chance to win or extend a 12-month membership to our AI Academy learning platform.) Survey link is in the first comment.

Every $1 spent on software supports $6 in human services. That's from a Sequoia essay arguing the next trillion-dollar company won't sell software. It'll sell the work that software used to require humans to do. A company spends $10,000 on accounting software. Then $120,000 on the accountant who uses it. If the software can do the accountant's job, the market isn't the $10,000. It's the $120,000. Multiply that across every knowledge worker in every industry and the total addressable market for AI isn't the software budget. It's the payroll. Everyone's payroll. The math here is so simple it's almost boring. And that's what makes it serious. When the economics are this obvious and this compelling, people are going to act on it. Some already are.

There were a couple of wild data points this week regarding AI and relationships. A new national survey from the Center for Democracy and Technology reports that nearly one in five US high schoolers say they or a friend have had a romantic relationship with AI. 42% of teens surveyed use AI for mental health support or as a friend. And over a third say it’s easier to talk to AI than to their parents. I keep coming back to the question: What do I do about this as a parent when the time comes to have conversations about this with my son? I don’t have a lot of answers yet. But it feels like simply being aware that this is a real, urgent trend is the first, most critical step. Just taking it seriously might be all you CAN do at this point... (Aside from having open, honest conversations with your kids, if you do have teens.) It’s easy to dismiss this trend as weird or isolated or too scary to engage with. But I try to force myself to engage with it nonetheless. Because I suspect it’ll be one of the biggest issues in AI moving forward.

I think the future of work with AI… Might look something like this: You conducting an AI orchestra… Of many different AI agents or tools. All of whom march to the beat you set. One agent or tool writes.  One researches.  One analyzes.  One creates. You conduct. I like the analogy because: Conductors don't play instruments…  They direct the performance. I suspect all our job descriptions will soon include less doing, and more directing.

Most people approach AI backwards. They ask: "What can AI do?" Wrong question. The right question:  "What do I need done?" Your job isn't one big thing.  It's thousands of tiny tasks. → Writing emails → Scheduling meetings → Analyzing data  → Creating content Stop trying to use AI for your “job.” And start trying to use it for… Each of the tasks that actually make up your job. Don't try to boil the ocean. Start small… Move fast… Stack up AI wins. It doesn’t have to be more complicated than that.

Same AI tool…same task… Completely different results. What changed? The prompt. Bad prompt:  "Write a blog post on X..." Good prompt: Provides context, examples, format The difference?  Night and day. Most people blame the AI when things fail. But here's the secret: A LOT of AI failures are prompt failures. Not tool failures.  Not AI limitations.  Just bad instructions. Want better results? Write better prompts. → Include examples  → Add context → Specify the format you want → Show AI what success looks like The tool is only as smart as your instructions.

Federal Reserve Chair Jerome Powell just said something pretty striking. He said at a recent press conference that "job creation is pretty close to zero.” He then connected that slowdown, at least in part, to what CEOs are now openly telling investors: AI allows them to do more with fewer people. Now, the classic counter-argument I always hear (and honestly, have leaned on myself at times) is that mass displacement is unlikely because our economy is driven by consumption. If people don't have jobs, who will buy the products? But I've been wrestling with a piece of data that complicates that idea. According to The Washington Post, the top 10% of American earners (making $250k+ a year) now account for a record 49.2% of all consumer spending. That's up from 35% in the 1990s. What this suggests to me is that the consumer economy can keep humming, even if a large portion of the population isn't participating at the same level. It seems entirely possible that we could see a future where the stock market is ripping, corporations are posting record profits from AI-driven productivity, and consumer spending looks healthy... All while a significant number of people are struggling with AI-driven unemployment or underemployment. In sum, I’m not sure “the economy” would simply break. It might just become breathtakingly unequal. I’m not quite sure yet what to do with this thought. But it strikes me that it’s worth thinking about, because it seems to imply that a few different economic realities could be playing out simultaneously and less visibly. We don’t necessarily have to see unemployment lines snaking down our streets to be living through severe economic disruption. It may be a lot more quiet or mundane than we expect.

Three years ago, Google declared "Code Red" in response to ChatGPT. Last week, OpenAI declared its own Code Red in response to Google. My, how the tables have turned. We unpack what's actually happening in the AI race, and why 2026 might be rougher than people expect, in this week's episode of The Artificial Intelligence Show. 👇

One trend I’m noticing: AI is really changing strategy. Yesterday’s strategy meeting took: - A bunch of people - Several hours - Whiteboard full of ideas …And maybe 5 solid ideas come out of it. Today’s strategy session has: - Fewer people - An AI reasoning model listening in - Takes 30 minutes …And produces dozens of great ideas. The math has changed... And the math is simple: Less time generating. More time executing. Less people talking. More ideas shipping. Less meetings scheduled. More strategies launched. Time to act accordingly.

69% of AI-engaged professionals say they can't keep up with AI's pace. I co-host a podcast to help people keep up...and I'd put myself in that 69% 😅 Every week, we run an informal survey of listeners of The Artificial Intelligence Show to get a snapshot of how professionals are actually experiencing AI. It's not a scientific poll. It's an informal survey of people who care enough about AI to take 30 seconds and tell us where they stand. This week, 69.4% of respondents said AI is accelerating faster than they can keep up with. Only 1.7% said they're not noticing major changes. I actually find that encouraging. Not because people are struggling, but because they're being honest. Nobody can fully keep up right now, and that's not a failure. It's just the reality of the pace. I do this for a living. Our team at SmarterX spends all day, every day tracking what's happening in AI. And I would absolutely put myself in that 69%. We try, and I think we do a decent job, but "fully keeping up" isn't something I'd claim with a straight face. The rest of the data we collected this week also tells some interesting stories: 📊 94% say AI has changed the total amount of work they do ⚡ 54.5% are getting more done in less time 📈 39.7% say they're doing more work overall, not less ⏰ 84% spend 5+ hours/week using AI tools 🔥 38% log more than 15 hours/week using AI tools The honest answer about AI's pace right now isn't "I've got this." It's "I'm doing my best." And that's fine. Again, just an informal sample. But food for thought nonetheless. Full post with all the data in the comments 👇

Everyone's obsessed with AI tools. But they're missing the point: It's not about having more:  → More platforms.  → More features. → More integrations.  → More subscriptions. It's about doing less:  → Less overthinking.  → Less comparing.  → Less waiting.  → Less perfecting. The most AI-productive people? They master one thing.  Not everything. They document one workflow.  Not rebuild their entire system. They improve one task.  Not transform overnight. They take one daily action.  Not one yearly initiative. 365 small steps > 1 giant leap that never happens. Subtract the complexity.  Add consistency.

My default used to be: hire the expert. Now it's: can I get there myself? I used to assume that anything outside my core skill set required a professional. Lawyers, accountants, consultants, specialists. That was the responsible thing to do. Increasingly, my first move is different. Before I call anyone, I check whether I can get to a good answer with AI. Not for everything. I'm not replacing human experts across the board. That would be foolish. But there are more and more domains where I can get pretty good results on my own. Good enough to move forward. Good enough to make an informed decision. Good enough that the expensive expert isn't the obvious next step anymore. I don't know exactly how this plays out over time. But if more people start doing what I'm doing, every service provider has to answer the same question: am I worth it when my client can get 80% of the way there without me?

The US blacklisted Anthropic. Yet Claude is still in combat. Here's what's happening in AI, from Ep. 201 of The AI Show: 🔹 The Pentagon labeled Anthropic a "supply chain risk," a designation normally reserved for foreign adversaries, while the military continues using Claude in active operations in Iran. 🔹 Amodei's leaked memo called OpenAI's Pentagon deal "80% safety theater," while Altman admitted to staff the move looked "opportunistic and sloppy." 🔹 Anthropic's new study found AI can theoretically handle 94% of computer and math tasks, but real-world usage sits at just 33%. 🔹 A Sequoia partner argued the next trillion-dollar company won't sell software, it will sell the work. 🔹 OpenAI released GPT-5.4, the first model to surpass human performance on computer use tasks. 🔹 Mollick says the biggest barrier to enterprise AI is whether one executive is willing to assume risk. 🔹 A mathematician watched GPT-5.4 solve a problem based on 20 years of his research and declared his personal singularity had arrived. 🔹 The AP's AI manager told reporters "resistance is futile," while Ars Technica fired a reporter for publishing AI-generated quotes. 🔹 Jensen Huang called OpenClaw "the most important software release ever." 🔹 The Supreme Court let stand the ruling that AI-generated art can't be copyrighted. 🔹 Meta is being sued over AI smartglasses privacy after workers reportedly reviewed sensitive footage. 🔹 Microsoft launched Copilot Cowork, embedding AI agents directly into Microsoft 365. Link to Ep. 201 in the first comment 👇

Everyone's obsessed with AI tools. But they've got it backwards: The secret isn't adding more... More models.  More apps.  More features.  More subscriptions. It's mastering less. One good prompt framework.  One core AI tool.  One use case at a time.  One small win daily.

Most people still use AI as a copilot. The shift to autopilot is already here. There's a meaningful difference between using AI to help you do work and having AI just do the work. Copilot means I'm still in the driver's seat. I prompt, I review, I edit, I approve. The AI makes me faster, but I'm still doing the thing. Autopilot means I describe the outcome I want and the AI handles it. I check the result, not every step along the way. Coding and software engineering are already making this transition. The rest of us are next. I catch myself defaulting to copilot mode all the time. It feels safer. It feels like I'm still in control. But increasingly, the question I need to ask myself is: does this actually need me in the loop, or am I just there out of habit? More and more, the answer is habit. The mental shift from "AI helps me" to "AI does it" is harder than it sounds.

Everyone's scared of AI hallucinations. But here's what they're missing: It's not about avoiding AI.  It's about knowing when to trust it. Sure, hallucinations prevent some use cases. But the bigger danger usually isn’t AI making mistakes… It's humans not knowing how to spot them. Typically, we don't need perfect AI… We need more AI-literate humans.

If you’re an AI-forward marketer… You won’t want to miss this: TODAY is MAICON Day. That means: For one day only… $200 off tickets to MAICON 2025. PLUS, when you buy today… You get entered to win a: → Free ticket upgrade → 1:1 AI expert consult → AI for Agencies Summit pass → VIP/Speaker Party Pass Check it out before it’s too late ↓