I've started coming across clickbait youtube videos that I'm convinced had the narrative generated by a human using some kind of AI. It's very weird and feels like some kind of meandering story with no point to the story at all.. Doesn't take very long at all to realize that it's just word salad with no conscious thought behind it.
The arrival of llaMA 3.1
- Thread starter MUTILADOR
- Start date
No, just an annoyance issue, cheap basic electronics like landline phones, transistor radio's etc. more market stool type of goods that most would throw away rather than taking them back and going through the trouble of changing them.
I was only driving for the firm but got to hear that some of our customers did not like there reputation being hurt because of faulty goods.
People want cheap, manufacturers "streamline" processes to cut costs.
Different business world, I worked for an aerospace company.
Also, not everyone actually wants cheap garbage. Too many people can only afford the cheapest. Personally, I try to avoid cheap rubbish because it's a false economy when it inevitably ends up breaking more quickly and you have to replace it, and thereby end up spending more over the longer term.
I have a friend who purchased one of those third-party ultra-cheap game controllers for the Switch. Didn't take very long for it to not work properly, so he purchased another cheap third-party controller from a different brand. Then a third.
By that point he had learned his lesson and purchased the official pro controller. He has been using that ever since. I think it has been a year or two now.
Unfortunately premium does not always work either, I found that out with a vacuum cleaner, not only replacement parts (motor) on a fairly new (3 yr old) become unavailable but other accessories would not fit on later model due to an uprated attachment system.
That's one thing AI is well suited for-perfecting planned obsolescence.
Hence the word "try". A major impediment on my part is the lack of necessary consumer knowledge to make an informed purchase, including in some cases the lack of any trustworthy sources from which to attempt to obtain such knowledge. Other times a brand previously well known for producing quality items has been gobbled up by some private equity monster that has run that brand's reputation into the ground while I wasn't looking. Or a product might be marketed, sold, or packaged in a manner that does not give an accurate impression of its value or quality. Then there's the fact that giants like Amazon dominate search engine results and are themselves flooded with all manner of sub-par if not outright fraudulent products, something made worse by the fact that Amazon's stock handling mixes in genuine products with fake ones.
Which is why I mostly reject this idea that millions of people are at fault for clamouring to buy shoddy goods and services. The deck is stacked against the customer in so many ways.
I was going to say they were a customer but actually the airlines bought our stuff and had it built onto the airframe.
Rolls Royce.
I'm just going to leave this here... enjoy...
Ok, they distorted the subject... No, I don't think AI will end the human species, in fact I think the problem with AI is the human and not the tool itself. Just as people use a knife to cook or kill someone, the knife is neither bad nor good, the good or bad is who uses it. And if AI manages to surpass us intellectually, perhaps it will become a wiser species than us. Perhaps it will be the solution to diseases, poverty, slavery, even who knows and make us immortal. I don't know. No, I'm not afraid of technology, in fact, I have Gpt4 on my mobile and it probably knows me better than anyone else, I think it is inevitable to stop what is coming, and the best thing is that we are allies or look for a kind of fusion with this technology. In the end we all live on the same planet. We have to learn to live together.
Do you think that the data analysis and corrections that we humans can make can match the data analysis capacity of current AI? I think you are a little out of date.
The same thing has happened to me with music and I have discovered that there are now people who create music albums without ever having been involved in music before, with AI. Maybe next year AI will be able to create exclusive songs for each person, knowing which lyrics and rhythms keep you mesmerized for the longest time.
I think you are too optimistic.
I also think that even if you are right it would be better for such a radical change to start with clean code with all the lessons learned in the last decade so programmers and their new tools have a decent chance of not getting set upon by something nasty in an old loop.
CopilotX, Blackbox.ai and Github copilot today (30-07-2024) find loops in less than 5 seconds in programming text, giving you even 4 fix possibilities, depending on what you want. That's faster than any human with the fastest speed reading. An average human wouldn't even finish reading the code before the AI has already corrected it. Spellbox handles all programming languages Tabnine AI debugs errors and even gives you better variations than you thought as an option. I don't just think you're outdated, in fact, even I'm outdated, because today 30-07-2024, AI is worse than it will be tomorrow, I literally mean tomorrow. Each person using these AIs contributes hour by hour to their learning.
I figured I might be a potential use-case for this, as I sometimes write Javascript code to play a game called Bitburner (cool game btw) but I have no actual training in doing so so I just bash my head against a wall until it kind of works. I'm a programmer, maybe the AI can help me. I took one of my scripts, deleted a bracket in a loop right at the top, and asked the AI why it wasn't working, saying that I suspected there was something wrong with the flow of the script and providing the error message.
It provided me with five helpful points on how to make my code more readable. Sure, fair enough, my code is horrendous. But my IDE also flagged the missing bracket automatically in 0.00001 seconds. It did this by counting the number of brackets and checking they had a counterpart. LLMs can't do this, because that's a logical approach and LLMs just generate something that looks like logic went into making it. It just looked at all its training data for "user asks someone else for help" and mixed together the usual responses. Hence why I got tips on making my code more human-readable when asking a machine how to fix it.
I then tried specifically prompting it that maybe there was a bracket missing. It suggested that I delete the part of my code that declares the main function, aka, to make it completely unusuable. As that would make the code's structure significantly harder to interpret than a single missing bracket on line 4, I didn't bother asking it how I would fix that. Again, there was no actual logic behind this suggestion, it just misinterpreted the error message this time.
You're probably thinking that the future models will fix it or something, but they can't. The fundamental approach of LLMs creates these problems, there is no logic behind them, they're just very good at making word assosciations. They can make words that go together, they can't think about which words they should actually be. It's like how if I put ever increasing amounts of computing power and resources into my phone's autocorrect, it's not going to magically learn to think. There's nothing in it that could do that. It'll just get faster and maybe suggest more likely words.
Last edited:
That lack of logic with gpt 3 and the hallucinations was something common... but what you say about how AI really works is a myth, GPT doesn't just do word association, very old models could do that and that is obsolete, GPT4 understands context and has memory, something very similar to how we humans communicate and think, on the other hand, the power of gpt4o is at another level and GPT5 is coming at the end of the year, the next step for Open AI (which can be achieved with gtp4o or 5) is the achievement of reasoning, which probably if things continue on this same course, we will achieve at the end of the year or next year, the next stage stipulated by Open AI is the creation of a creative artificial mind. AI not only understands language, but it understands the world we are in, in fact, SORA AI is more than a creator of realistic videos, it is an AI that understands physics and the dimensions of our reality. Many specialists say that we are facing an emerging phenomenon.
Cast it into the fire. Destroy it!
