My mother tongue is Spanish, yet I can’t understand at all what the laughing guy is saying. He has an extreme Castillian accent.
Something about he being in swimming trunks while attempting to deliver a Paella.

ChatGPT translation.

 

That is hilarious!
EDIT; beyond hilarious.

 

 

https://archive.is/NgILh

Exclusive: Meta planning sweeping layoffs as AI costs mount

Top executives have recently signaled the plans to other senior leaders at Meta and told them to begin planning how to pare back, two of the people said. The sources spoke anonymously because they were not authorized to disclose the cuts.

"This is speculative reporting about theoretical approaches," Meta spokesperson Andy Stone ⁠said in response to questions about the plan.

 

https://archive.is/NgILh

Exclusive: Meta planning sweeping layoffs as AI costs mount

Top executives have recently signaled the plans to other senior leaders at Meta and told them to begin planning how to pare back, two of the people said. The sources spoke anonymously because they were not authorized to disclose the cuts.

"This is speculative reporting about theoretical approaches," Meta spokesperson Andy Stone ⁠said in response to questions about the plan.

So... "pare back" is somehow indistinguishable from "sweeping layoffs"?

The link wouldn't load, so I don't know if the article resolved that or not.

Personally, I think those two people should be fired. They were unauthorized to disclose the cuts (which may or may not be something that has been decided upon), yet they did. I sure wouldn't want them working for me (and this is hardly a case of whistle blowing). On top of that, the media is clearly willing to believe anything they say, despite the fact that they've demonstrated that they can't be trusted.

 

We’re experiencing an uncanny valley of autonomy. Computer systems aren’t just almost human; they are almost capable of working on their own. When they fail, someone has to absorb the cost. Right now, that someone is us. But when we pay for a self-driving car or an AI tool, we think we’re buying a finished product, not signing up to test a work in progress.
This “almost” phase isn’t a brief transition. It’s the product—one that will be with us for years, maybe decades. So it’s important to notice the patterns. When an AI system never admits uncertainty, or when a car’s marketing says “self-driving” but the fine print says “driver responsible,” that’s a warning sign. When you realize that you haven’t really been paying attention for the past 10 miles, or the past 10 auto-composed emails, that’s the trap.

https://archive.ph/VQQs1

 

The current tech tendency is to convince the clients themselves to pay for being the beta testers ...

EDIT: And to make them consider it a privilege ...

 

https://arstechnica.com/gaming/2026...ng-disgust-to-dlss-5s-generative-ai-glow-ups/
Gamers react with overwhelming disgust to DLSS 5’s generative AI glow-ups
Nvidia’s next frame-gen tech goes way beyond upscaling, and not in a good way.

Spoiler

I'm not a 'gamer' so I've no idea what's good or bad in that universe.

 

 

https://www.cnn.com/2026/03/19/politics/super-micro-computer-founder-charged-ai-chips-china
Co-founder of tech company charged with diverting $2.5 billion in Nvidia AI chips to China in violation of export laws

In another instance, prosecutors said surveillance cameras documented individuals using hair dryers to remove labels and add labels and serial number stickers to the boxes and dummy servers.

 

https://www.hollywoodreporter.com/b...i-shutting-down-sora-ai-video-app-1236546187/
Disney Exits OpenAI Deal After AI Giant Shutters Sora
The studio giant will no longer move forward with its OpenAI investment, as the AI company exits the video generation business

 

https://techcrunch.com/2026/03/26/openai-abandons-yet-another-side-quest-chatgpts-erotic-mode/
OpenAI abandons yet another side quest: ChatGPT’s erotic mode

 

 

https://cacm.acm.org/blogcacm/model-collapse-is-already-happening-we-just-pretend-it-isnt/

Model Collapse Is Already Happening, We Just Pretend It Isn’t
The weird, rare, surprising patterns that make data rich slowly get smoothed out when an AI model trains on outputs from a previous model.

 

https://finance.yahoo.com/video/wha...-contain-leaked-claude-ai-code-141759728.html

https://www.straitstimes.com/busine...ally-releases-source-code-for-claude-ai-agent

What day is today?

 

For me, AI in general, and Chat GPT in particular, have been a "mixed bag".

Have there been times where it has attempted to "lead me down the garden path" with directions, answers, and advice that it just plain wrong? ... absolutely. Frequently in-fact. But on the flip-side, has it also given me verifiably accurate credible information and/or done a good job articulating something that I've been struggling with? ... again, absolutely. And frequently too.

Given that the world is currently awash with disinformation and misinformation, I think it's critically important that we all take the time to evaluate information we receive (from any source) to hedge against bias; my personal protocol before giving credibility to anything significant is to ask myself "what training, qualifications, and experience the asserter has?" (ie "is a stated assertion on vaccines coming from a vaccinologist or does it come from a pastry chef who got it from a YouTube video made by a taxi driver?") and "Do assertions made by suitably trained, qualified, and experienced experts enjoy peer support from other appropriately trained, qualified, and experienced experts?" (to mitigate outliers).

Because I'm not an expert in most things, I'm aware that any positions I form on most topics would consist of more "bias" than "truth & wisdom" if I didn't defer to people who actually do know what they're talking about.

So, for me anyway, that's the "rub" with AI; for me, it's just a tool. Sometimes it's the right tool for a job and sometimes it's the wrong tool for a job; either way, it's my responsibility to do a good job evaluating the results it's given me.

I've had occasions where an Arduino sketch won't compile and I'll be damned if I can see the issue (most often when I'm tired). Many is the time when Chat GPT has given me a "doh" moment in a couple of seconds when it highlights my error. Also, many is the time when I ramble on for too long (like I am now) and Chat GPT can re-work my ramblings in to something tight and refined - and for that I'm grateful. I frequently use it to summarise long news articles to "remove the spin and get to the facts" (or get it to explain something to me). As a teacher it's been great in explaining things to me - things that I'm careful to make it justify (or point me to credible references).

I've had several instances this year alone where it's been able to assist in the repair of electronic items where what I've been able to charge for has more than covered the subscription cost for the entire year. And yes, frequently, it also gives me information that absolutely, totally, and completely nonsense. The only danger for me is if it were to give me wrong info that I didn't pick up on; does that ever happen? ... I can't say without being biased, but I do "call it out" on a regular basis. In reality, that's no different to how the world works anyway; we constantly get information from many sources and if we dont take the time to separate the "truth & wisdom" from the "ignorant bias" then that's on us.

For what it's worth, personally, I can have far more "intelligent" conversations on topics with Chat GPT than I can with people who's positions are so obviously biased that discussing things with them is pointless.

So, again for me, the benefits it brings easily out-weigh the disadvantages - even if it's only to search other forums - identify relevant information that fits my scope - and then points me to it.

TLRD: (summarised by Chat GPT):

"AI, including ChatGPT, is a mixed tool.

It often produces confident but incorrect answers. It also often provides accurate information and helps articulate ideas clearly. Both happen regularly.

The only reliable approach is to evaluate everything. Two checks matter:

• What relevant training, qualifications, and experience does the source have?
• Do other qualified experts broadly agree?

Without that, most positions end up being more bias than knowledge.

AI is no different from any other source. It is a tool. Sometimes it is the right tool, sometimes it is not. The responsibility to verify remains with the user.

In practice, it is useful for:

• spotting simple errors quickly, such as debugging an Arduino sketch
• tightening and rewriting text
• summarising long material
• explaining concepts, provided the explanation is checked

It has also been useful in technical work, including electronics repair, where correct suggestions can save time and money.

The risk is simple: incorrect output that goes unnoticed. That risk already exists with any information source. The difference is that AI can generate it quickly and convincingly.

Used critically, the benefits outweigh the drawbacks. Used uncritically, it is unreliable."

-- I'd call that a 9/10 summary.

 

Not worth the risk of brain rot from a system without any intelligence, understanding or reasoning. It only has correct answers because previous humans have found the correct answer in the training data. Not knowingly playing the AI game.

 

Not worth the risk of brain rot from a system without any intelligence, understanding or reasoning.

By my observations the world is already awash with brain-rot-inducing mechanisms; from social media that seems to exist solely as a platform for 99% of people to document their bias and ignorance - to "reality" TV consisting of content edited to be distorted and misleading (but entertaining) - to false and misleading advertising that we allow ourselves to be bombarded with - to the conduct of world "leaders" who communicate "spun" truth (and outright falsehoods) with the sole objective of gaining/retaining power. For me, AI has been supreme in reducing my exposure to many of these brain-rotting things - which as been a definite plus for my mental health.

It only has correct answers because previous humans have found the correct answer in the training data. Not knowingly playing the AI game.

To say "only correct because it saw it in training data” is not how systems like ChatGPT work in practice. Yes, they are trained on large datasets, but they don't retrieve and repeat stored answers - instead, they learn statistical relationships and generate responses based on patterns. That allows them to generalise to new problems not explicitly seen before, combine concepts across domains, and produce correct outputs in cases where no exact prior example exists.

They also demonstrably perform multi-step reasoning tasks, code generation, and error diagnosis. That is not simple lookup behaviour.

On one hand AI would be neutered without access to training data - but on the other hand it's far better at spotting relationships between data than humans are (and is getting better). Used critically, it’s a tool that can improve efficiency and understanding. Used uncritically, it’s no better than any other low-quality information source. Again, the responsibility to verify what it gives us remains with us; again, for me, when it generates something that I'm just not capable of doing myself then, for me, that's a huge win. When it generates nonsense then that's a minor annoyance. "huge wins" trump "minor annoyances" for me - by a big margin.

 

By my observations the world is already awash with brain-rot-inducing mechanisms; from social media that seems to exist solely as a platform for 99% of people to document their bias and ignorance - to "reality" TV consisting of content edited to be distorted and misleading (but entertaining) - to false and misleading advertising that we allow ourselves to be bombarded with - to the conduct of world "leaders" who communicate "spun" truth (and outright falsehoods) with the sole objective of gaining/retaining power. For me, AI has been supreme in reducing my exposure to many of these brain-rotting things - which as been a definite plus for my mental health.



To say "only correct because it saw it in training data” is not how systems like ChatGPT work in practice. Yes, they are trained on large datasets, but they don't retrieve and repeat stored answers - instead, they learn statistical relationships and generate responses based on patterns. That allows them to generalise to new problems not explicitly seen before, combine concepts across domains, and produce correct outputs in cases where no exact prior example exists.

They also demonstrably perform multi-step reasoning tasks, code generation, and error diagnosis. That is not simple lookup behaviour.

On one hand AI would be neutered without access to training data - but on the other hand it's far better at spotting relationships between data than humans are (and is getting better). Used critically, it’s a tool that can improve efficiency and understanding. Used uncritically, it’s no better than any other low-quality information source. Again, the responsibility to verify what it gives us remains with us; again, for me, when it generates something that I'm just not capable of doing myself then, for me, that's a huge win. When it generates nonsense then that's a minor annoyance. "huge wins" trump "minor annoyances" for me - by a big margin.

I understand how they are supposed to work, yet we can still get AI systems to output almost verbatim text from the datasets they used.
https://arstechnica.com/ai/2026/02/...verbatim-copies-of-novels-from-training-data/

“There’s growing evidence that memorization is a bigger thing than previously believed,” said Yves-Alexandre de Montjoye, a professor of applied mathematics and computer science at Imperial College London.
AI groups have long argued that memorization does not happen. In a 2023 letter to the US Copyright Office, Google said “there is no copy of the training data—whether text, images, or other formats—present in the model itself.”
The AI industry also claims that training models on copyrighted books is “fair use,” arguing that the technology transforms the original work into something meaningfully new.
But a study published last month showed that researchers at Stanford and Yale Universities were able to strategically prompt LLMs from OpenAI, Google, Anthropic, and xAI to generate thousands of words from 13 books, including A Game of Thrones, The Hunger Games, and The Hobbit.
By asking models to complete sentences from a book, Gemini 2.5 regurgitated 76.8 percent of Harry Potter and the Philosopher’s Stone with high levels of accuracy, while Grok 3 generated 70.3 percent.

They were also able to extract almost the entirety of the novel “near-verbatim” from Anthropic’s Claude 3.7 Sonnet by jailbreaking the model, where users can prompt LLMs to disregard their safeguards.

They don't actually reason or understand the questions asked because, as you say, it's statistical relationships, that leads to a smoothing of results and a loss of creativity in solving problems IMO. There is nothing an AI can help me with at this point that I can do as good or better myself (sure it might take longer but I will have actually reasoned out the answers by methods that could far from the statistical norm)

I see it as a great destroyer of precious resources like the human mind and the environment. Most of these companies, like Google, a few years ago pretended to be for 'Green Energy', not at least now they don't pretend anymore with the current data-center energy madness of burning anything made from oil for electrical power. These systems are costing regular people money from increased prices for energy, water and computer related hardware.

 

I understand how they are supposed to work, yet we can still get AI systems to output almost verbatim text from the datasets they used.

Maybe so, but storing text from sources (notwithstanding legal debate) doesn't mean that AI output is merely subsequent verbatim regurgitation of portions of that data (unless that's what the user is trying to achieve - in which case it's happy to comply with that too).

They don't actually reason or understand the questions asked because, as you say, it's statistical relationships, that leads to a smoothing of results and a loss of creativity in solving problems IMO.

To me, terms like "reason" and "understand" are ones that I would use more in a relating-to-humans context; when relating to AI I prefer to ask "does it accurately interpret my input" - and in my experience the answer is a resounding yes - it's been incredibly accurate. It may not always get the right answer but by golly it virtually never misinterprets what I'm asking - including subtlety and nuance. I find that quite incredible to be honest. It doesn't always do a great job with sarcasm, but that's probably more an indicator of my frustration with current world affairs ...

So although it does function on a statistical relationship basis, that doesn't mean that it suffers from any significant limitations to me when it comes to accurately interpreting what is being asked or the steps required to solve the problem; it was designed that way because that's the way it works best.

Does it lead to a "smoothing of results and a loss of creativity in solving problems"? Maybe - maybe not - it really depends on the problem; in my experience, for what I require of it, it's exceptional at identifying the crux of what I'm asking & then ruthlessly executing the required steps in the required order to produce the answer required. Personally. I do value and admire creativity - especially in art - but I also value and admire ruthless efficiency.

What creativity is being lost when I'm asking it to calculate the force on a 1m umbrella in a force 1 hurricane, or to summarise a long waffling news article? On the occasions where I'll ask it to generate Arduino code, it'll often fail at the overall task - but it'll show me a plethora of coding approaches that are far tighter and more efficient than I could ever have thought of along the way - and I subsequently learn from those. Could it be said that those code concepts (not just code examples) originally came from humans? ... absolutely ... but so what? AI is just another tool that also came from humans - no different to spanners, saws, computers, MRIs; it's completely unsurprising that it contains our "DNA".

There is nothing an AI can help me with at this point that I can do as good or better myself (sure it might take longer but I will have actually reasoned out the answers by methods that could far from the statistical norm)

For me, AI can do many things that I could do myself - but because it does them faster it gives me more time to focus on higher-level tasks - and that's a win for me. Because it can tap into an extremely wide range of resources - and it can make decisions based on an incredibly complex array of variables - it can also produce things that I'm just not capable of doing. Case in point - I'm not a psychiatrist, but I do find the workings of the human mind quite interesting. One of the things I find quite facinating is habits - and I personally have a few bad ones. In James Clear's excellent Atomic Habits, he writes extensively about their 4 component parts cue, craving, response, and reward - the mechanism is well understood and yet try as I did, I wasn't able to relate those 4 parts to a particular habit of mine, I just couldn't do it. Could I have paid a therapist and got the right answers? Maybe - but probably not. Could I have "posted online" and got the answers? ... yes to the first part, probably no to the 2nd part. I put it to Chat GPT and it absolutely nailed it - not only with the answers but with a substantial explanation of them. What's not to like in situations like that?

I see it as a great destroyer of precious resources like the human mind and the environment. Most of these companies, like Google, a few years ago pretended to be for 'Green Energy', not at least now they don't pretend anymore with the current data-center energy madness of burning anything made from oil for electrical power.

The same could be said for almost every tool ever invented. Should I walk to work instead of using my motorcycle "tool"? ... 10 hours of walking a day will do wonders for my health and the environment ... but at the expense of all office productivity - the productivity that generates my income and solves problems for other people. Nothing is ever a single-sided transaction.

If the majority people truly valued their minds then they would find something far more healthy than engaging with social media, watching "reality" TV, allowing themselves to be bombarded with false and misleading advertising, and supporting people whose net contribution to the world is a net vast destruction of values; unfortunately, I think that horse has well and truly bolted. As I see it, AI is possibly the first tool capable of reducing a lot of that harm - if people choose to use it that way; I use it to produce daily news briefs so I don't get exposed to as much "spin" and "lean" and advertising. I use it to "cut through the cr^p" to produce something of merit from otherwise low-value presentations (eg summarise a YouTube video in seconds to answer questions without me having to wade through 20 minutes of filler). I use it to learn; I've spent around 2 hours last week getting tailored instruction on IGBTs that I've now consolidated in practical experiments; it's been like having my own private tutor. I can honestly say that it was the happiest 2 hours of my week.

Personally, I'm not a fan of excessive burning anything made from oil for electrical power either, but that's probably another conversation for another day. If the datacenters that run social media disappeared overnight then I'd be quite happy about it - if the same happened to the datacentres hosting Google search or eBay / Amazon / AliExpress commerce then I think that would have a far more negative impact on the world. That being said - given the amount of oil used daily - how much less (or more?) would be used if there were no data centers? ... my guess is that despite the enormous amounts of power they use, it wouldn't actually make any meaningful difference. Case in point - here in NZ - a small country in the middle of nowhere we apparently use around 25 million litres of oil-related fuels per day ... and I suspect we're pretty typical of many countries.