The official allegation is that using AI is generating "inauthentic" content, what is true, given you have a blueprint to how to get those videos e views. But I suppose having so many shady people literally farming real money out of Youtube must be hurting Googles pocket more than saying something is not authentic.
But its authentically AI. And it generates authentic money. Which AI slop is good at on other media so why doesn't Utube lie back, relax, and enjoy it?
I suppose the answer is the same answer to the same question asked to SE:
If RMTs/bots still pay subscription, if RMTs/bots keeps buying news accounts, why SE still takes measures against them?
The answer? That is up to debate.
I would say both SE and Youtube must consider, despite not hurting the revenue at first sight, in a long run if left unchecked the predatory behavior of those people exploiting the system will inevitably lead to the system downfall.
Youtube in particular has an extra argument. The platform itself is a hugh datalake that can be used to train new models.
But if most content on Youtube doest have human value embued, then the models arent learning new things, thus, it does make sense Youtube to want content that is at least some bit enriched by human insights, than straight AI generated content.
The official allegation is that using AI is generating "inauthentic" content, what is true, given you have a blueprint to how to get those videos e views. But I suppose having so many shady people literally farming real money out of Youtube must be hurting Googles pocket more than saying something is not authentic.
But its authentically AI. And it generates authentic money. Which AI slop is good at on other media so why doesn't Utube lie back, relax, and enjoy it?
They still get a cut after all ...
The question is: If you create 100x the amount of content (absolute AI trash) and only a few of those become obscenely popular - are the costs offsetting the sudden need to host and distribute millions of new videos that generate you 0 revenue.
That was already an economics problem - hence why Youtube needed to be sold to google in the first place, but now the problem is much worse.
Generative AI works really well with curated datasets that you own and therefor no liability issues. Since you already draw / write / generate with that style, it will just generate what it thought you would do anyway. Instead of someone having to generate something they've generated a hundred times in the past, they can have the computer do that job for them while they supervise and work on unique / new stuff.
Lately, Ive been considering the chatGPT service to be really behind compared to Google free AI chatbot.
Some specific questions I sent to chatGPT were returning as a denial, or simply hallucinations, while Google AI has been giving back answer saying something that chatGPT denied as possible,
At this point, the best use Ive been doing with OpenAI chatbot is to edit images, what Ive been relatively successful, and I do think it would be worth the price alone IF... not for this damn limit of a handful amount of images I can sent there per day.
In other words, Ive been using chatGPT solely for image editing just because Ive been really lazy to look for a more qualified image generation tool.
But unfortunelly I already lost my trust on that chatbot, and if my plan were expiring tomorrow, I would certainly not renew.
Serious question for the aspiring vibe coders out there.
Why would a company pay you money just so you can write prompts into an AI to then use for work.
Why would that company not just hire someone that knows nothing about development, with no degree what so ever, half way around the world where ever the labour is cheapest, to do the same thing.
How is "vibe coding" a useful "skill" to develop for someone that is aspiring to be a developer.
What makes the vibe coder any different than a 13 year old kid playing around with prompts.
What makes the vibe coder any different than a 13 year old kid playing around with prompts.
That's what most vibe coders are for the most part.
To understand what's going on, we have these executives with either very little or very old IT knowledge hearing things like "80% improvement in coding efficiency" and "AI will replace all tech jobs" and think to themselves "80% reduction in IT staffing costs". Then they issue a directive to the middle management to "use AI to enhance productivity" and make it a part of their performance goals, which are usually tied to bonus compensation.
Those middle management then do everything humanly possible to force "AI" into the workflow, so they can make those performance goals and get that yearly bonus. Vibe coding resulted from this, hiring low cost non-developers to mess with prompts trying to generate code, then pushing it into production and causing outages and bugs.
Just look at the sheer number of crash causing bug Windows patches have caused in the last year, since MS did it's big AI mandates.
What makes the vibe coder any different than a 13 year old kid playing around with prompts.
Right now, it seems to be ability to orchestrate multiple agents in a way that's notably better than browser-based AI. There's a huge gap between someone asking ChatGPT how to do something and someone using a multiple agent setup with local file access. People who can't code think that because they've learned a few basic techniques to orchestrate agents, they're AI professionals.
Give it a few years (maybe not even) and the agent setups will be abstracted away (and paywalled past what a hobbyist is likely to spend). Then, the 'skillset' they've been developing ceases to have value.
at least half of the positions at a lot of these companies are entirely redundant or non-essential. much rather have a surplus of vibe coders over middle management, but they are quite good at keeping their own jobs and getting rid of the ones that actually help the business.
at least half of the positions at a lot of these companies are entirely redundant or non-essential. much rather have a surplus of vibe coders over middle management, but they are quite good at keeping their own jobs and getting rid of the ones that actually help the business.
Funny thing couple of weeks ago, we had a large meeting where all our managers were presenting the years delivery plans.
But in reality, it was just a show of buzzwords, always using the trendy words to show they are “modern”.
My department is under the company AI major department, so AI is the hot topic here.
At their level, they just want us to use AI and see if we can get something out of it.
Knowing how to use AI is a plus.
But if I already have a license paid by the company, not hurt in trying to do my best anyway.
I just saw more a couple of videos teaching new tricks to improve vibe coding.
That sounds like something for an IDE to automatically do.
I hope my IDE doesn't invoke agents automatically. But, yes, it's obviously going to be streamlined into plugins (it already is for many configurations). Look at the nonsense Panta is always spewing about agent orchestration and which agents are best to combine. This is the kind of junk youtubers are talking about to convince people they're learning a skill.
It's a simple enough process that the most effective agent workflows will be automated and streamlined into one tool. Select which agent setup you want, offer your prompt, and let it go. Learning common agent configurations isn't a skill, as much as vibe coders want to think it is.
What makes the vibe coder any different than a 13 year old kid playing around with prompts.
Right now, it seems to be ability to orchestrate multiple agents in a way that's notably better than browser-based AI. There's a huge gap between someone asking ChatGPT how to do something and someone using a multiple agent setup with local file access. People who can't code think that because they've learned a few basic techniques to orchestrate agents, they're AI professionals.
Give it a few years (maybe not even) and the agent setups will be abstracted away (and paywalled past what a hobbyist is likely to spend). Then, the 'skillset' they've been developing ceases to have value.
In reality there is a huge gap between basic function blocks and making those blocks actually do something useful while being safe and stable. Experience and context aware judgement are the missing pieces.
Case study would be all the rollbacks MS has had to do lately as update after update breaks windows as they insist their developers use AI to write system code.
I hope my IDE doesn't invoke agents automatically. But, yes, it's obviously going to be streamlined into plugins (it already is for many configurations). Look at the nonsense Panta is always spewing about agent orchestration and which agents are best to combine. This is the kind of junk youtubers are talking about to convince people they're learning a skill.
LLMs are programms that are based on language analyzation and guessing with an added database. not intelligent at all. more like kelly bundy with access to wikipedia.
Unbound ai that is actually intelligent and learns/evolves/optimizes itself exists since 2016 as far as i know, but wont be anything a consumer will ever get their hands on. also it requires insane amounts of hardware.
And if not what would or could be? (This assumes that we are intelligent.)
Sub questions:
1, Is self awareness needed for intelligence?
2, Is conciseness needed for intelligence?
3, Would creativity be possible without intelligence?
Feel free to ask more.
I say they aren't. To me they are search engines that have leveled up once or twice but haven't evolved.
They use so much electricity because they have to sift through darn near everything for each request. Intelligence at a minimum would prune search paths way better than LLMs do. Enough to reduce power consumption by several orders of magnitude.
After all if LLMs aren't truly AI then whatever is will suck way more power unless they evolve.
I don't think that LLM's hallucinations are disqualifying. After all I and many of my friends spent real money for hallucinations.
