๐ฅ Introducing @PepeExplorers DEMOLITION EXPERT ๐ฅ
Some solve problems. Some avoid them. He blows them up. ๐๐ฃ
๐ด Armored combat helmet
๐งจ Explosive specialist
๐ค Tactical support drone
๐ฏ Precision destruction expert
๐ Chaos is part of the plan
Built for high-risk missions and impossible odds, @PepeExplorers 's Demolition Expert is always one detonation ahead.
Would you recruit him for your squad? ๐
Thank you @pduivenb for the prompt
@TrystanNFT
@Dorialexander Gemini for vibecoding I think is vastly underrated...mainly because of issues many now take for granted in other platforms. When google fixes the issues...it will be a totally different ball game...
https://t.co/nTqoSnDUME
If it were up to Yann LeCun, We would all be Burger King Bound
Yann LeCun is yapping again!
The Chief AI Scientist at Meta, has been consistent in his skepticism for years, but his claims became a major public focal point around late 2022 and early 2023.
While the rest of the world was celebrating the release of ChatGPT (November 2022), LeCun famously took a different stance. He argued that Large Language Models (LLMs) like GPT-4 are merely "off-ramps" on the highway to true intelligence, not the destination.
The "Dead End" Timeline
Pre-2022 - LeCun had long argued in research circles that text alone is not enough to train a machine to be smart. He believed AI needed to experience the physical world, much like a baby does.
Early 2023 - As the AI hype exploded, LeCun amplified his warning. He stated that LLMs are limited because they only predict the next word in a sentence. He called this a "dead end" for achieving Artificial General Intelligence (AGI) because these models lack common sense, logic, and an understanding of physical reality.
2024โ2025 - he began aggressively promoting his alternative vision called "World Models" (specifically an architecture called JEPA). His argument is that AI needs to understand how the world works (cause and effect), not just mimic human language.
Industry Achievements Since LeCun's Claim (2023โ2025)
Despite LeCunโs warnings that LLMs might be a scientific dead end, they have been a financial and industrial rocket ship. Since late 2022, the industry has seen explosive growth in funding, revenue, and capability.
Massive Financial Growth
The amount of money pouring into AI has shattered records.
Funds Invested - by 2024, corporate investment in AI reached approximately $252 billion. Private investment specifically for Generative AI (the tech behind ChatGPT) jumped to over $33 billion in a single year.
Company Revenues:
OpenAI - skyrocketed from roughly $200 million in 2023 to an annualized revenue rate of $13 billion by August 2025.
Anthropic - saw incredible growth, moving from roughly $87 million in early 2024 to an annualized rate of $7 billion by late 2025.
Nvidia - the company making the chips that power AI (GPUs) saw its market share of data center chips hit 86%, becoming one of the most valuable companies on earth.
Adoption & Usage
AI moved from a "fun experiment" to a core business tool.
Global Usage - in 2023, only about 33% of organizations used Generative AI. By 2024, that number more than doubled to 71%.
Consumer Reach - apps like ChatGPT exceeded 5 billion cumulative users globally, becoming some of the fastest-growing consumer products in history.
Technological Milestones
The capabilities of these models improved drastically, challenging LeCun's claim that they would hit a wall quickly.
Reasoning Models - new models (like OpenAIโs o1 and others) were released with "reasoning" capabilities, designed to "think" before they speak, directly addressing some of LeCunโs criticisms about logic.
Multimodal AI - AI moved beyond just text. Models can now instantly generate realistic video, analyze complex images, and speak with emotional voice tones in real-time.
Context Windows - the "memory" of these models expanded from a few pages of text to entire books' worth of information (millions of tokens), allowing them to process vast amounts of data at once.
The Verdict So Far
LeCun argues that while these products are useful and profitable, they are scientifically stuck and won't evolve into "superintelligence." and continues to yap.
However, the market disagrees, having poured hundreds of billions of dollars into proving that LLMs can continue to scale.
Google and Google Ai Studio have come a real far way....i even find the vibe coding experience much better than some paid services.
However, so may issues still exist and must be turning away many, especially the wannabe developers.
Here are some blatant issues:
1. the constant mixing or react 18 and 19 that invariably breaks apps and prevent proper deployment
2. the Ai often being prompted, applies the fix but forgetting to update the code.
3. having to keep a watchful eye of the chat dialogue...for whatever reason, the Ai will recode the same prompt over and over again...this at times, at the expense of new prompts...
4. the Ai seems lost when it comes to handling uploaded logos and favicons....even using the logo its created as a favicon seems to be an issue.
5. the Ai seemingly lost in preparing meta tags and sitemaps...you are hardpressed to get Google to index a site its own Ai has built.
6. support is sparse or non-existent...right now the Github integration has been broken for days and nothing/nobody sees it/responds to to the hail marys.
7. Not often, but there are times the Ai simply can't seem to update the code at all...and this with repeated attempts...
8. It's not straight foward getting apps to use like a Netlify to publish site on a custom domain...it was impossible it seemed at one stage...some fix apparently has been applied but it's not perfect.
9. The Ai seemingly does not see and interpret hi res uploaded images very well...it will say it understands but reproduces something completely different.
10. The studio does not seemlessly connect to a backend...not even to an expensive option for multiple projects like supabase.
These are some of my not so peaceful experiences using the studio...I often wonder if Google is fully onboard with the solution.
While other vibe coding platforms have their own issues, I cant say that any of these problems ail them anymore.
I really do hope Google pays more attention as the coding itself is pretty good and it does more complex apps than most as far as i have experienced, but to be real a factor in such a competitive market...it's still a long way off.
Gemini just seems to lack many of the creature comforts you take for granted in other vibecoding platforms.
And what's the point if the app is a beauty but you can't deploy or if you deploy but the site can't be indexed.
Something tells me they will have to do mighty better even on top of the leaps and bounds they have made over the past few months.
Anyone else using Google Ai studio to vibecode? What are your thoughts?
So, if it's no longer hype, and companies are now building/growing their Ai revenue base...wouldn't MS now be even in a better position?
BTW, from Google Ai search...
"Microsoft's AI revenue growth is accelerating rapidly, driven by strong adoption of its Azure cloud services and AI-powered productivity tools (Copilot), with Azure and other cloud services revenue growing 40% in the first quarter of fiscal 2026 (ending Sept. 30, 2025). The company's AI initiatives have pushed its annualized Azure revenue to over $75 billion for fiscal year 2025."
A fall in revenue does not have to be Ai. Please repeat after me and 1000 times by yourself.
You could do the research but no, you have to frame your folly to make OAi the fall guy.