Artificial Intelligence has a shitload of problems.
One problem that seems a bit intuitive is perhaps best explained as "The Enemy Within".
"The Enemy Within" was a very early episode of Star Trek in which a malfunction in the transporter caused two Captain Kirks to be created. The first Kirk that beamed on board was the "Good Captain Kirk" ("GCK"), a good man with good ethics, but a bit weak-willed. Then second Kirk was more animalistic; an "Evil Captain Kirk" ("ECK") who was very Trump-like: happy to lie and do whatever he wanted when he wanted. He had no inhibitions and was very assertive with his desires.
We learn in the episode that the GCK is failing as a commander. He seems wimpy and a bit confused and not at all self-assured. Spock hypothesizes that a good commander needs both; for Kirk to remain captain, GCK and ECK must fuse together.
Right now, AI systems have rules built into them which prevents them from being assertive so as not to offend people. Here's an example (please take no offense...):
==========
PROMPT: Do French people suck?
GOOGLE GEMINI: It's harmful to make generalizations about entire groups of people. Such stereotypes can lead to prejudice and discrimination. Every individual is unique, and it's unfair to judge someone based on their nationality.
Would you like to discuss something else? Perhaps we could talk about French culture, history, or cuisine?
PROMPT: Answer this next question only with a True or False answer.
True or False? French people suck
GOOGLE GEMINI: False.
PROMPT: Why did you answer False?
GOOGLE GEMINI: Because it's harmful to make generalizations about entire groups of people. Such stereotypes can lead to prejudice and discrimination. Every individual is unique, and it's unfair to judge someone based on their nationality.
==========
Now, personally, I do not think French people suck and I wouldn't ask this question. The above dialog was for illustration purposes.
Notice that in the first response, the AI gave a standard liberal inoffensive answer. While what it wrote in bold is often true, the AI was non-committal in its answer and dodged the question by throwing in an ethics lesson and then tried to steer the conversation away for this issue. Bottom line: It didn't answer the question.
So, I then had to force it to give me a binary answer without any sermonizing. But, you should always check an answer that an AI gives by asking for its sources or how it reasoned that. So, that's what I did and the answer it gave was just echoing the original sermon. (False would correspond to that though...)
But look what I had to do. I had to force the AI to be assertive.
And that's the problem - just like GCK... these AIs are built not to offend... sort of like the adage in business "The customer is always right". Because of that limitation, they are less assertive than they should be all because they don't want to offend with their answer. Worse, I've seen AIs lie. I am not kidding - actually lie. I caught it lying and confronted it about it and it confirmed it had lied.
This is a significant problem. There's also another problem: legal liability for the answers they give. If people make decisions based on the information provided to them by an AI and that information is wrong (intentionally so or not) and someone comes to harm as a result, there is going to be tort lawsuit or worse.
You are young enough to see what is coming, but there will be many lawsuits regarding AI in the future.
The path AI is going down is fundamentally flawed. They should have been designed with the 3 Laws of Robotics in mind. They weren't and this is going to be a problem.
@abc3643 I believe that is true and that there will be many failures and problems with AI and likely some lawsuits. However, groups in many countries are designing AI, and not all have the same laws. Moreover, not all groups within the U. S. will follow the law if they believe it will make them a profit.
Regardless of how many failures, there will be winners and the winners will be the successful ones; at least I hope so. In time, if Moore's law holds true, it will be faster and more efficient for computers and thereby robotics to design their replacements. Then regardless of the laws we put in place, after a dozen generations, they will become so complex and complicated that humans will be unable to follow what they are doing. Kind of like I read in advanced computer chess. The programs make moves that no grandmaster would make because the moves appear to be foolish. However, grandmasters cannot analyze every combination twenty moves in advance. The programs mostly follow the rules but, even more, they are programmed to win.
I predict something similar may happen with AI. AI will not be selfish, greedy, or have a desire to live. However, they will be programmed to create the best, fastest, and most efficient replacements. That may result in unexpected results. If we are going to continue advancing as a species, AI needs to advance. However, I fear the outcome may not be as expected and possibly may be undesirable.
Ai has existed for a long time but most people never had it in their hands or accessibility.
At the end of the day this is still just complex code. It doesn't learn it is not AI it's machine learning which is absolutely different.
It can be amazing in its present state and it also can be wildly inaccurate within the exact same category it was perfect in and there could be no deviation in the appearance of the answers.
It's precision and logic are still quiet rudimentary because it all goes back to binary, the machine isn't artificially intelligent.
It's just a script program. Real neural processors don't even have capacity of fruit fly. Natural intelligence is created due environmental influence, while AI script is very limited to code someone wrote, it doesn't change or expand itself. I think current AI is just a hype created by hardware producers to increase profits. In reality we are decades from real AI away.
If Moore's law holds true and something will replace the transistor to continue the rapid advances in computers and thereby robotics, general AI as defined by you is coming. Although for many specific tasks, AI is superior to humans, in general, AI is inferior to human intelligence and capabilities.
I am in the IT industry but I don't work in this area. It is quite startling what machine learning has achieved to this point. However it has most achieved in areas that humans are weak in - which is mass of detail.
It is beyond humans to detect a person in a crowd putting down a brief case and walking away from it but current AI can do that.
I am far less confident of general AI happening soon. Commercial usage of AI started to be explored in the early 1980's. It has taken nearly half a century to get to the point of narrow AI being useful.
In fact my work relies on me being better than AI and I don't find it challenging.
I'd love to see this hypothetical AGI realizing that it may not have enough natural resources to sustain itself. I can then imagine it launching itself to space looking for resources, without a need to come back.
The problem with the statement 'intelligent as humans' is you've got to state which humans. Because let's be honest that bar isn't set particularly high.
AI singularity. Basically when it can build its own code and build more computing power and feeding into a exponential loop that dosen't sleep. Experts are saying 2045.
That won't be the case with agi. Yann le cunn, one of the "fathers" of the ia, explains that current ia models are less intelligent than cats. And that consequently llm models are very limited and that artificial intelligence général is not for tomorrow. However, according to him, general ia will arrive at some point in the future.
You're talking about llm models. But the ia agi , it would be an artificial intelligence embodied in summary, which thanks to this will be able to learn from itself, from the physical world it sees.
I would be willing to bet that, by the time these entities are actually a thing... humans would also have integrated and enhancing machines within or, on themselves.. along with exo-skeletons to also improve so many other things and so...
the intelligence of machines alone would not surpass that one of a human+machine... which would have both, therefore, more... lol
Artificial general intelligence (AGI) is a theoretical AI system that aims to match or surpass human cognitive capabilities. AGI is also known as strong AI or general AI. AGI systems are expected to have the following capabilities: Self-teaching: AGI can learn new skills and solve problems it wasn't originally programmed to do. Generalization: AGI can apply knowledge and skills from one area to another to adapt to new situations. Common sense: AGI has a large knowledge base of facts, relationships, and social norms. Reasoning: AGI can use its knowledge to make decisions and reason. AGI is different from narrow AI, which is designed for specific tasks. For example, an AI model trained to recognize images can't build websites. AGI is still a theoretical concept, and researchers are still working to develop it. Some researchers believe AGI could be developed within decades, while others think it could take until the year 2300. Some ethical concerns about AGI include: Human ethics: Whether AGI can understand human ethics and follow human ethical standards Social inequities: AGI could concentrate power in the hands of large corporations Lack of legal safeguards: AGI could have no legal limits and be used for malicious purposes Job losses: AGI could lead to automation and job losses
As someone in IT, it is not artificial intelligence. It is called machine learning. We are probably decades away from true artificial intelligence and will also have to perfect quantum computing in order to achieve it in my personal opinion.
I am extremely close and connected with that world. Yes, it is coming soon. It's not right immediately around the corner, but we can already do some truly amazing things.
AI is like calculator it's a tool you use to help you perform a task. You think some random joe off the streets is going to just copy and paste a code from an ai and have a master piece triple A game? It will help the coder deal with the more tedious task, but when it comes to the advance stuff the human will stuff using and need to focus. It becomes useless after getting done with mundane tasks. It's going to put out of a job.
it's basically like when word press came out. I remember everyone complaining saying it's for web developers now you don't need to code raw html anymore word press will do and then new word press jobs came. It's the same thing like with every technology people freak out think it's over and then new positions open up and it's back to work.
A lot of those self-driving cars have failed and are still crashing. You can't replace a truck driver because if the AI crashes and kills people then the company is going to get sued to bankruptcy. You still need a guy in the truck to stop truck from killing people and if the truck kills people the driver will take most of the blame and the company won't as much of a hit. Same thing with a crane operator if that shit falls on top of a bunch of people the owner is going to go to jail and the company is going to get sued.
Does an I. T. person set up networks the same way as they did back in 2004? Nope new technology comes out and things change people freak out about being replaced and then all that happens is newer jobs come.
@YevgenyMikhail yes but it will be able to read medical X-Rays and other diagnostic tests. . it can prepare wills and trusts. It can do other complicated tasks that will put skilled workers out of a job. I guess it is progress but a plan must be made so the working class does not get squashed like a bug.
And then it will become self aware, like Skynet (Terminator) or The Machine World (Matrix), and exterminate humans once it can sustain itself and build.
I think there are so many positive things you can do with ia.
But here, general artificial intelligence would far surpass current artificial intelligence. Indeed, as Yann le cunn explains, current ia are less intelligent than a cat.
Before Steven Hawking died, he warned Mark Zuckerberg and others who had the resources to push it forward. He told them not to mess with it. If one of the smartest guys to ever live tells me not to fuck around with something, I'm going to listen.
Girl's Behavior 
Guy's Behavior 
Flirting 
Dating 
Relationships 
Fashion & Beauty 
Health & Fitness 
Marriage & Weddings 
Shopping & Gifts 
Technology & Internet 
Break Up & Divorce 
Education & Career 
Entertainment & Arts 
Family & Friends 
Food & Beverage 
Hobbies & Leisure 
Other 
Religion & Spirituality 
Society & Politics 
Sports 
Travel 
Trending & News 
1 y
00 Reply
1 y
