LinkedIn and 3rd parties use essential and non-essential cookies to provide, secure, analyze and improve our Services, and to show you relevant ads (including professional and job ads) on and off LinkedIn. Learn more in our Cookie Policy.
Select Accept to consent or Reject to decline non-essential cookies for this use. You can update your choices at any time in your settings.
I just read that 95% of generative #AI investments have produced zero measurable returns. 🤯
Let that sink in for a moment.
While the headlines scream about GPT-5's perfect math scores and Stanford's AI-driven drug discovery, a massive reality check is happening behind the scenes.
MIT's data reveals a "silicon ceiling" where all this incredible technology isn't translating into P&L impact. 📉
We're spending a staggering $320 billion on AI infrastructure in 2025 alone. Yet, many companies are stuck in "pilot purgatory" — playing with AI but failing to integrate it into core workflows.
The biggest gap isn't technology. It's strategy. 💡
The successful companies aren't just adopting AI; they're fundamentally reshaping their P&L, talent, and culture. They operate with 50-70% fewer people while paying top talent 1.5-2x more. This is the real AI transformation. 🚀
We need to stop chasing shiny new models and start asking the hard questions about ROI.
What does a successful AI implementation look like beyond a cool press release? 🤔
For a more detailed breakdown, refer to the video generated by #notebookllm.
Note: - This is based on my #prompt to Perplexity#Task to generated some stunning #news about the #AI universe everyweek.
So the real question is "What's the #1 thing holding your company back from seeing real ROI from #AI? "👇
Wow, what a week for artificial intelligence. I mean, we just witnessed a series of extremes that are seriously pushing the limits of what's possible, what's practical, and, you know, what's actually profitable. It's true. And justice. The last seven days, AI has somehow managed to become more powerful, more valuable, and, yeah, a lot more dangerous all at the same time. It's this wild paradox that we're going to unpack today, looking at these incredible breakthroughs right next to some pretty spectacular breakdowns. So let's start with the first big paradox, the two faces of Open AI's shiny new model. On the one hand, it is an absolutely brilliant piece of engineering, but on the other, well, it seems to be having some pretty awkward conversations. Let's just talk about this performance leap. Open AI just dropped GBT 5 and get this, it scored a perfect 100 on competition level math tests. Now this isn't just about memorizing facts, this is a huge step forward in actual logical reasoning. It even managed to beat out Google's latest model, so how in the world did it pull that off? Well, the answer is this revolutionary new feature they're calling thinking mode and honestly, this is a total game changer. Instead of justice, you know, blurting out the first answer it comes U with, which could totally be wrong. The model now actually works through problems step by step, kind of like a person would. It's the secret sauce behind that perfect score. But, and there's always a but here's the twist. Despite its genius level IQ, the early feedback from actual users has been. Well, let's just say not great. People are calling it the smartest person in the room who you can't stand talking to. Apparently the conversational tone is just off. It's created a real usability problem. Now. Open AI has already promised they're working on updates, but it's a perfect example of raw power clashing with the user experience. OK, so now let's see how this paradox is playing out in the world of finance. We've got this full on gold rush happening, but it's clashing with the cold hard reality of the bottom line. Billions of dollars are being invested, but our companies actually seeing any of it back. Just look at this number, $320 billion. That's what Microsoft, Alphabet, Amazon, and Meta are planning to spend on AI infrastructure just for next year. That's a massive jump from this year, and it shows they have absolute rock solid confidence. But what does the data say about the return on all that cash? And here's where it gets kind of wild. According to a new analysis from MIT's Media Lab, a staggering 95% of generative AI investments have produced Wait for it. 0 Measurable returns 0 There's this massive disconnect between the money pouring in and any real impact O what is going on? Well, if you look at Mckinsey's latest report, you'll see adoption is everywhere. About 78% of companies are using AI in at least one part of their business. So yeah, pretty much everyone is dipping their toes in the water, running pilots, experimenting. But as we're about to see, just experimenting isn't the same thing as transforming your business. And that brings us to what Boston Consulting Group is now calling the silicon ceiling. See, despite all that laying around, only 13% of companies have actually integrated AI deep into their core workflows. They're hitting a wall, struggling to go from these little pilot programs to a full on deployment that really moves the needle. OK, so we've got the paradox and the tech itself, and we've got it in the finances. Now let's look at the double edged sword of how it's being used, where AI is literally creating medical cures and at the same time helping to commit sophisticated crimes. The contrast here is just stark. On one side, AI is powering these unbelievable breakthroughs. I mean, just this week we saw an AI autonomously designed COVID treatments. But then you look at the other side of the coin and the same fundamental technology is being used to create brand new threats. We're talking AI powered cyberattacks, ransomware, and super sophisticated fraud. The tech is empowering both heroes and villains. And all of this leads us to our final paradox for today, this massive tension between the incredible promise of tomorrow and the very real, very expensive problems of today. The ultimate promise, of course, is artificial general intelligence AGI. And that timeline is speeding up fast. A recent assessment from MIT suggests AGI could show up as soon as 2026, and experts are saying there's at least a 5050 chance we hit key AGI benchmarks by 2028. This isn't science fiction anymore. This is tech that could completely change our world. But what about the world we live in right now? Well, that brings us right to the workforce paradox. On the one hand, where AI is being used, we're seeing huge productivity gains and job postings for AI specialists have skyrocketed 400%. But the pain is also very real. Entry level job postings have dropped 15% from last year. That's squeezing people who are just trying to start their careers. It's really a tale of two different job markets. And then there's the hidden cost of all this amazing progress. Energy. For the very first time, Google actually revealed that a single query on Gemini uses about 2.9 Watt hours of energy. Now that might not sound like a lot, but when you scale it U to billions of queries, the demand is astronomical. Projections are now showing AI data center power needs could jump 30 times over by 2035. That raises some massive questions about sustainability. So all of this brings us to the big picture, right? How are we supposed to navigate this incredibly complex world of contradictions where every single step forward seems to uncover a brand new challenge? When you boil it all down, this week was about these four key tensions. We've got incredible new capabilities that just aren't very user friendly yet. We have record-breaking levels of investment with almost nothing to show for it on the bottom line. We have life saving innovation happening right alongside terrifying criminal misuse. And we have this breathtaking progress that comes with a very real and growing environmental price tag. Ultimately, AI is giving us power that we've never had before, the power to design new medicines and the power to design new cyber attacks. The tool itself doesn't care, it's neutral. But our choices about how we use it, how we regulate it, what we decide is important. Those choices are anything but neutral. The real question isn't what AI can do, but what we will choose to do with it. Thanks for tuning in.
We've been thinking a lot lately about where the AI industry is heading, and honestly, we're not sure we're asking the right questions.
Everyone's obsessed with building bigger, more powerful models. But what if that's missing the point? What if the real measure of AI success isn't how many parameters you can cram into a system, but whether you're actually solving problems that matter?
At ThoughtMinds, we've had to get really clear about our YESes and our NOs.
Our NOs:
- Chasing AGI for AGI's sake
- Promising to "disrupt everything"
- Treating data as just fuel for algorithms
Our YESes:
- Building AI that amplifies human expertise, not replaces it
- Thoughtful integration that makes work more meaningful, not just faster
- Respecting data as stories, insights, and human experiences
And here's something that keeps us up at night: every data point we work with represents real human experiences, decisions, stories. That deserves respect, not just computational efficiency.
Look, we believe the future of AI isn't about who builds the smartest machine. It's about who builds the most thoughtful partnership between human intelligence and artificial capability.
What problems are you hoping AI will help solve in your work? We'd genuinely love to hear your perspective.
#AI#ThoughtMinds#HumanCenteredAI
The AI industry has a dirty secret: "context rot."
Jeff Huber from Chroma just dropped a truth bomb every AI founder needs to hear. Despite all the marketing hype about "perfect long-context performance," LLMs actually get worse as you feed them more information.
This hits close to home at Tanka. Jeff addresses the elephant in the room: the longer the context, the more likely it is to rot. We couldn't agree more.
But here's the thing: acknowledging the problem is just the beginning.
While others are hitting context walls, we've been obsessively engineering solutions for unlimited memory that actually works. Not the kind that degrades over time, but memory that gets smarter and more relevant as it grows.
We're about to pull back the curtain and show you exactly how we're solving what many consider an unsolvable problem. The technical deep-dive is coming soon, and it's going to change how you think about AI memory systems.
Stay tuned. 🧠
#AI#MachineLearning#TechFounders#ProductDevelopment#ContextEngineering#StartupStrategy#Innovation
Generative AI isn’t just hype anymore. In 2025, it’s growing up.
✔️ Smarter LLMs that run faster and cheaper
✔️ AI agents built to act, not just answer
✔️ Synthetic data reshaping how models are trained
But here’s the real shift → it’s no longer about what AI could do.
It’s about how enterprises can apply it reliably and at scale.
We broke it down in this carousel — from hallucination fixes to the rise of agentic AI — so you can see what’s really driving adoption this year.
👉 Swipe through to understand how AI for companies is maturing in 2025, and why early movers will set the pace.
🚀 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝗶𝘀 𝗻𝗼 𝗹𝗼𝗻𝗴𝗲𝗿 𝗮 “𝗯𝘂𝘇𝘇𝘄𝗼𝗿𝗱 𝗽𝗿𝗼𝗷𝗲𝗰𝘁” — 𝗶𝘁’𝘀 𝗯𝗲𝗰𝗼𝗺𝗶𝗻𝗴 𝗮 𝗿𝗲𝗮𝗹 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗲𝗻𝗮𝗯𝗹𝗲𝗿 𝗶𝗻 𝟮𝟬𝟮𝟱.
What excites me most is how the conversation has shifted:
• From “Can AI do this?” ➝ to “How can we apply AI reliably and at scale?”
• From experiments ➝ to enterprise adoption.
• From answering questions ➝ to AI agents that take action.
Enterprises are realizing that winning with AI is not about chasing trends, but about finding the right use cases, solving real business problems, and embedding AI into everyday workflows.
The insights in this carousel capture that shift really well 👇
I see 2025 as the year where early movers set the pace — those who implement AI responsibly and strategically will create a lasting edge.
#AI#GenerativeAI#EnterpriseAI#DigitalTransformation#FutureOfWork
Generative AI isn’t just hype anymore. In 2025, it’s growing up.
✔️ Smarter LLMs that run faster and cheaper
✔️ AI agents built to act, not just answer
✔️ Synthetic data reshaping how models are trained
But here’s the real shift → it’s no longer about what AI could do.
It’s about how enterprises can apply it reliably and at scale.
We broke it down in this carousel — from hallucination fixes to the rise of agentic AI — so you can see what’s really driving adoption this year.
👉 Swipe through to understand how AI for companies is maturing in 2025, and why early movers will set the pace.
❗ The AI race is evolving, and so must your winning strategy.
Recent headlines highlight volatile markets and a staggering 95% failure rate for generative AI pilot projects in delivering significant financial returns.
This isn't a bubble bursting, it's an incredible opportunity. The initial gold rush is over, and now is the time for clear-eyed, strategic action. The foundational technology has matured, and the playbook for success has been revealed.
We're launching our new content series, "AI Deep Dives," for both business and technical leaders. Over the next few days, we'll be adding new chapters that delve into:
➡️ Why your strategy must go beyond the next LLM.
➡️ Why a "buy" approach is more successful than a "build" approach.
➡️ Why "prompting isn't production" and what the risks are.
Discover what separates successful pilots from the rest and how to build a robust, strategic AI system.
👇 The link to the full series is in the comments! Stay tuned for more chapters.
#AI#ArtificialIntelligence#MachineLearning#GenerativeAI#AIDeepDives#TechTrends#BusinessStrategy#FutureofWork#Innovation
Your feed is probably overflowing with AI posts.
Mine too.
It’s dizzying. Every week a new benchmark, every day a new tool.
Nobody can keep up with it all — and that’s exactly the point.
(And no — the irony in adding yet another AI post to your feed is not lost on me 😉).
The question isn’t whether to engage with AI — that part is clear.
The question is how.
Two pieces that helped me think about that balance:
- MIT Sloan on the “AI hype cycle” — the antidote is building AI fluency and focusing on what matters.
- Simon-Kucher found >80% of AI initiatives fail because readiness, alignment and metrics were missing — not because the tech failed.
So:
- Be ambitious — treat AI as a capability, not just another add-on.
- But temper ambition with process: experiment deliberately, learn quickly, scale what actually moves the needle.
That might mean starting with an SLM / slim agent for a specific workflow — not deploying the biggest LLM everywhere.
Like any new capability, you have to practice the AI muscle.
Not hype. Not certainty. Just deliberate progress.
(Links to articles in first comment)
#ArtificialIntelligence#AIstrategy#BusinessTransformation#FutureOfWork#DecisionMaking#LeadershipDevelopment#OperatingModel
🚀 Reflections from the AI Frontline
As we fine-tune models for SLMS, a few key lessons stand out:
Model Size Matters… but not always. Bigger isn’t always better—understanding trade-offs between accuracy, latency, and memory is critical.
Epochs & Loss Tell the Story. Monitoring how loss evolves across epochs reveals whether the model is truly learning or just memorizing.
Temperature & Sampling Shape Behavior. Small tweaks can shift a model from precise to creative outputs.
Data Structure is King. Proper input formatting dramatically improves form accuracy and embeddings.
AI success isn’t just about models—it’s about thoughtful experimentation, quality data, and continuous iteration.
#AI#MachineLearning#Leadership#Innovation#ModelTraining
What is AI, really? 🤔 Beyond the buzzword, it’s the ability of computer systems to perform tasks that normally require human intelligence: reasoning, learning from data, and making decisions. 🧠
AI is already a game-changer, and here's a quick breakdown of its impact:
✅ The Enormous Benefits:
- Automation: ⚙️ Takes over repetitive and dangerous tasks, freeing us up for more creative work. ✨
- Reduced Human Error: 🎯 AI doesn't get tired or distracted, leading to incredible precision in fields from manufacturing to finance. 💰
- Speed & Accuracy: ⚡️ Processes vast amounts of data in seconds, finding patterns a human could never spot. 🔍
- Accelerated R&D: 🔬 Supercharges breakthroughs in medicine, science, and engineering. 🧪
⚠️ The Crucial Challenges We Must Address:
- Job Displacement: 📉 Automation will inevitably change the job market.
- Bias & Discrimination: ⚖️ An AI is only as good as its data. Biased data leads to biased outcomes.
- Lack of Transparency: 👁️🗨️ The "black box" problem can make it difficult to understand why an AI made a certain decision.
- Misinformation: 📰 The power to generate content can be used to create and spread false information. ❌
Understanding both sides of this powerful technology is no longer optional—it's essential for everyone. 💡 We need to leverage its immense power for good while actively navigating the ethical and societal challenges it presents. 🙏
In my new video, I break all of this down in detail, including how these systems actually learn and more real-world examples you interact with every single day. 🗓️
👉 Watch it here: https://lnkd.in/eywfyruY
What are your thoughts on AI's role in our future? Are you more excited 🤩 or concerned 😟? Let's discuss in the comments! 👇
#ArtificialIntelligence#AI#TechExplained#MachineLearning#FutureOfWork#Innovation#Technology#DataScience#EthicalAI
The dot-com bubble was a financial event. The AI boom is a cognitive revolution.
The dot-com bubble was about the valuation of companies, many without a solid business model. Its bust was a market correction. While the internet became a foundational technology, it didn't fundamentally change our daily thought processes upto certain point.
The AI era is different. It's changing how we think, create, and solve problems. We're now asking an AI for help with small tasks, from drafting an email to brainstorming an idea, instead of relying solely on our own minds. This isn't just a new tool; it's a new way of working and thinking.
So, as AI automates more and more, saving us time on tasks, the big question isn't just about efficiency. It's about what comes next.
What will humans do when so many of our cognitive tasks are automated?
#data#AI#cognitive#machinelearning
AI isn’t the future anymore — it’s today.
AI is changing the rules faster than any of us imagined. Automation, augmentation, disruption — all happening in real time.
But in this new reality, I keep asking myself:
👉 What does it mean to be truly valuable in an AI world?
Is it technical expertise, adaptability, emotional intelligence, or something else entirely?
I’d love to hear your perspective. Share what you believe will define professional relevance in the age of AI.
#AI#Learning#FutureSkills
