Oh boy, where do I even start with this new article. Never, ever have I had writer’s block, but in this case, I’m staring at a blank screen like it’s a freaking Rorschach test. I mean, the topic is a doozy. But for me, it feels like I’m trying to sum up a soap-opera season finale in just one friggin paragraph.
So, let’s start by talking about ze early days of AI.
I remember the day ChatGPT3 launched, as if it was yesterday. I had been active with Machine Learning for about a decade, and had even played around with it’s predecessors, GPT2, and GPT1, but the 3.5 version took me by surprise. That day, our world had changed for good (or for worse, depending on your point of view, but do not worry peeps, we’ll get to that later on).
Do you remember when AI was the shiny new toy that you wanted to play with every day?
ChatGPT had us all geeking out. They led us think that we were on the brink of some sci-fi future where robots would be serving us martinis and it would write our emails, create our powerpoints, and save us a lotta work.
And now, if we fast forward to 2025, it is kind of like waking up from a wild night out, if you know what I mean. The excitement is waning, and I am left with a hangover and a few regrets, and I’ll tell you why.
If you like this article and you want to support me:
- Comment, or share the article; that will really help spread the word 🙌
- Connect with me on Linkedin 🙏
- Subscribe to TechTonic Shifts to get your daily dose of tech 📰
The hype train has officially derailed.
I went from “AI is going to change the world” to “AI can’t even do some proper research without hallucinating like a Pinocchio on LSD”. It’s feels kinda like watching Hancock with Will Smith, where the hero turns out to be a drunk zero.
The promises we got were sky-high..
But the delivery….Meh, at best.
AI was hailed as a savior of all things tech, but it has hit its angsty teen phase in 2024. It has loads of potential, but oh-so-much drama and attitude.
In 2022 ChatGPT had me all starry-eyed, and dreaming of revolutions, and industry shake-ups, and the thrilling – creepy – idea that machines would actually outthink us.
But fast forward a hot sec, and what do we have? A mountain of busted dreams, some meh updates, and a sneaking suspicion that AI is more like a slacker intern than the game-changer that I hoped for. Yeah, yeah, we had our fun, asking ChatGPT dumb questions and making AI art that was meme-gold. But the shine has worn off, big time.
And I am not the only one, experiencing this…
Dr. Linda Zhao from Stanford did a study showing that American workers’ excitement for AI tools tanked from 45% to 36% in just three months.
I’ll let this sink in for a while…
🕐
🕑
🕒
🕓
The quote that I presented above is total bullshit.
It is straight-up fabricated. Not by me, but by ChatGPT itself when I dared to ask it for some “research”. Oh, and Dr. Linda Zhao is real, alright. She is a statistician at Wharton School who works on inference and machine learning. But this so-called study never happened. Not even close.
Yet here we are, staring down the barrel of AI’s ability to conjure up plausible-sounding nonsense. If you needed a reminder of the wobbly, error-prone mess our current AI models are in, well, this right here, peeps, is your Exhibit A. (for more gnarly reads on this topic, go look at: It has been 2 years since ChatGPT launched. What has it done to us? | LinkedIn, AI is a compulsive liar | LinkedIn, or I’ve seen the dark side of AI, and you need to know about it | LinkedIn)
The research done by the AI tool Perplexity is much, much better. The research figures that ChatGPT came up with, are factual, but comes from Slack’s Fall 2024 workforce index.
Other research shows similar movements:
Vast Majority of Americans Pessimistic on How AI Will Affect Employment
Employees are hiding their AI use from their managers.
A Comprehensive Survey on AI in the Workplace
AI’s Impact on Jobs, Time Saved Has Employees Divided
And now, it turns out, that once the hype dies down, no one wants to deal with glitchy tech that spits out nonsense. I mean, Google’s Gemini AI literally told someone to “please die”, and another bot, but this time from character AI told a 14 yo kid to commit suicide. Which he ultimately decided to do. (Read: A tragic bond with a chatbot: The story of Sewell Setzer III | LinkedIn).
The machines meant to help us are serving up existential crises instead.
OpenAI has been the untouchable AI queen ever since it launched, and they are now desperately trying to keep the hype train running. Uncle Sam A. is still talking up the “Intelligence Age”, and he whispers sweet nothings about his path to AGI to people, but even he admits now that GPT-5 isn’t happening anytime soon, and if it will, it will be more or less agentic and meant for businesses.
So this means that we are going to be stuck with GPT-4o for an unforeseeable amount of time, which feels about as groundbreaking as a new iPhone wallpaper.
And Meanwhile, Cambridge University found that these “advanced” models have a 25% bias rate in healthcare diagnostics. Helloooo?!?! 25%. That is one in four that gets a wrong diagnosis. So, not only is AI useless, but it’s also actively harmful. Definitely not the “intelligence” that we signed up for, but more like a data-driven dumpster fire with a fancy interface.
And now we have Microsoft which is proudly calling itself the “Copilot company”.
Sounds cool, I must admit.
And their CEO Satya Nadella’s is all about “agentic AI” that is going to transform our lives, but there’s a catch: Bing’s market share hasn’t moved an inch despite the AI integration. And even Slack users are now are ditching AI tools, and they are preferring the old-school grind. Researchers found that AI-powered moderation tools flag accurate info while letting hate speech run wild. If this is Microsoft’s idea of a “full-scale transformation,” hard pass.
Now take Apple. They are always ready to join the tech circus. They were kinda late at the game, and their excuse was published in the form of a research study into the “reasoning” capabilities of various LLM’s. They concluded that it was nothing but smoke and mirrors (read: We should all start spelling AI as Ai because LLM’s are full of shit (according to Tim) | LinkedIn). But despite their critique, and lured into the game with the promise of untold riches, they tried their hand at AI with “Intelligence Updates” for Siri. Are you serious, Apple? You get confusing notifications and a glowing redesign that’s all style but no substance. This is just the kind of lame innovation that makes you miss the days when AI was just a scary sci-fi plot, and not a frustrating reality. AI tools like Copilot often increase error rates – by up to 41%, and this is proving to me, that the tech is raw dough.
For all the big talk about universal transformation, it has become clear that generative AI has hit what Gartner’s hype cycle calls the “trough of disillusionment”. It is that phase where tech has to prove itself after the initial excitement wears off.
And so far, AI is failing hard.
We are left with a parade of broken promises..
We have tools that can’t cut it, corporations cashing in on our frustrations, and the there’s this one lingering question: “was this ever really about innovation, or just another money grab disguised as progress”?
In 2024, our dearly beloved AI has failed to live up to its hype, and it is actively sabotaging its own credibility.
Between buggy updates, biased algorithms, and straight-up offensive chatbots, the tech is as much a liability as it is a tool. Unless Silicon Valley gets its act together, this once-revolutionary field risks becoming just another chapter in the history of overhyped tech fads. And let’s be real, that’s an ending nobody wants.
And now, riding in on a wave of corporate desperation, comes the next big thing: agentic AI.
Oh, sure, it is futuristic and exciting.
Again.
AI that just doesn’t wait for your command but takes initiative!
But at this stage, in reality it is still a teenager with the car keys and you are hoping for the best. We have all experienced error rates and hallucinations that are already spiraling out of control, and the idea of AI “making decisions for us” feels less and less like a technological breakthrough and more like a high-stakes game of roulette (read: Brace, brace brace! AI takes the stick at Heathrow’s air traffic control center | LinkedIn, Flamethrower dogs, kamikaze cars, and bomb-planting humanoids. | LinkedIn)
So 2025 will be the year that AI shows profitability. And I am hoping it is a little more than just wishful thinking wrapped in a buzzword.
Microsoft is betting its future on this so-called “full-scale agentic transformation”, but when your star product is buggy enough to flag accurate information while it gleefully letting hate speech roam free at the same time, how exactly does that inspire trust. And even Apple, who is the king of making style-over-substance profitable, can’t get this one right. They are offering glowing redesigns and AI tools that are more error-prone than the oompa loompas you hire to fetch your coffee.
In 2024, I have gone from dreaming of intelligent machines to watching these systems flounder like an overzealous Roomba yelling slurs (read: Robovacs chase pets and yell slurs | LinkedIn). Unless Silicon Valley can figure out how to make AI tools actually work, and maybe not lie, insult or endanger us in the process, we’re looking at yet another overhyped, underdelivering disaster.
So I have decided to pitch in my 2 cents.. this week, I’ll share the results of a project I’ve been working on for a few days: a conceptual framework for Enterprise Agentic AI. Because if we really want agentic AI to be successful, we need to start fixing things.
Signing-off from the deep trenches of AI,
Marco
Well, that’s a wrap for today. Tomorrow, I’ll have a fresh episode of TechTonic Shifts for you. If you enjoy my writing and want to support my work, feel free to buy me a coffee ♨️
Think a friend would enjoy this too? Share the newsletter and let them join the conversation. Google appreciates your likes by making my articles available to more readers.
Top-rated articles
- Anticipating AI’s next move • article ① • | LinkedIn
- Anticipating AI’s next move • article ② • | LinkedIn
- Anticipating AI’s next move • article ③ • | LinkedIn
- Anticipating AI’s next move • article ④ • | LinkedIn
- Richard Feynman’s notebook method in a modern age with Obsidian and Zeta Alpha
- AI based Intelligent Automation: [RPA + BPM] x AI = ROI
- A model for strategy development in the AI age | LinkedIn
- Fake is the new real
- The Borg are here to assimilate you through digital telepathy
- 150 Digital Transformation templates new ones added
- Chip happens ! This article explains the alphabet soup of chip names (CPU, GPU, LPU, TPU)
- Now why do AI/ML project FAIL!
- The age of hyper-personalization
Leave a Reply